Comparative evolutionary pharmacogenomics of human prostaglandin-endoperoxide synthase paralogs identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions

Reuben S Maghembe

doi:10.1371/journal.pone.0358222

. 2026 Sep 25;21(9):e0358222. doi: 10.1371/journal.pone.0358222

Comparative evolutionary pharmacogenomics of human prostaglandin-endoperoxide synthase paralogs identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions

Reuben S Maghembe ^1,^2,^*

Editor: Chandrabose Selvaraj³

PMCID: PMC13614540 PMID: 42789539

Abstract

Human PTGS1 and PTGS2 encode cyclooxygenase paralogs that regulate prostaglandin biosynthesis and are major targets of nonsteroidal anti-inflammatory drugs (NSAIDs). Interpreting population-differentiated PTGS variation requires integration of allele-frequency structure with transcript consequence, protein topology, splice-prediction evidence and structural context. This study integrated allele-resolved population differentiation, transcript-aware consequence annotation, splice-prediction boundary checks, protein-domain mapping, direct leader-sequence property calculations and controlled structural analyses. Candidate classes distinguished high-FST synonymous contextual markers, N-terminal PTGS1 leader-sequence missense variants, splice-region candidates and PTGS2 p.Val511Ala. Web-based SpliceAI/Pangolin evaluation of seven splice-relevant or comparator variants provided limited support for splice alteration. A retrospective comparison of 31 missense candidates showed that Val511Ala was not the most differentiated missense variant overall, but ranked first by allele frequency, allele-frequency range and maximum pairwise FST within four channel-context candidates; FST was used only to describe population differentiation. Canonical Val511 mapped to 5KIR Val525, a second-shell position adjacent to the rofecoxib-contact network. Controlled docking-score, contact and direct-frame pose analyses detected no systematic variant-associated shift under the tested conditions. PTGS1 p.Trp8Arg and p.Pro17Leu produced distinct directly calculated changes in net-charge proxy, hydrophobic-residue count, mean Kyte-Doolittle hydropathy, aromatic-residue count and proline count across the 23-residue leader sequence and descriptive H-region. No signal-peptide predictor output was used, and effects on SRP recognition, ER targeting, translocation, cleavage, membrane insertion, maturation, abundance, localization, enzyme activity or drug response were not demonstrated. These results support a calibrated prioritisation framework in which population differentiation identifies structured variation, while transcript consequence, protein context and reproducible quantitative analyses define experimentally testable candidates.

Introduction

Prostaglandin-endoperoxide synthases 1 and 2 (PTGS1 and PTGS2; COX-1 and COX-2) catalyse prostaglandin biosynthesis and are major anti-inflammatory drug targets. PTGS1 and PTGS2 share cyclooxygenase chemistry but differ in tissue distribution, inducibility, regulatory control, cellular context and inhibitor sensitivity [1,2]. These paralog-specific differences make PTGS variation a useful model for testing how population-differentiated coding variants map onto protein topology and pharmacogenomic interpretation [2–4].

Human PTGS variation has been linked to inflammatory phenotypes, NSAID response and adverse-event susceptibility [1,5–8]. Variant interpretation remains challenging because reported associations may depend on phenotype, drug class, exposure definition and population background [1,5,9,10]. While population differentiation can identify structured PTGS variation [7,11,12], frequency differences require transcript, coding, splice and protein-context information before they become mechanistic hypotheses [10,13–15]. A consequence-first framework provides a compact route from population signal to biological interpretation [16]. In this study, variants were organised by transcript consequence, coding effect, principal-transcript relevance, protein topology, splice-prediction support and structural interpretability. This hierarchy separates contextual synonymous markers from PTGS1 N-terminal leader-sequence candidates, PTGS2 channel-adjacent candidates and splice-region variants requiring RNA-level follow-up.

Splice prediction, structural modelling and docking add functional context to these candidate classes [17,18]. SpliceAI and Pangolin support splice-region interpretation [18,19], while protein modelling and docking place PTGS2 Val511Ala into catalytic-channel ligand context. Common NSAIDs were selected because they represent clinically established COX inhibitor chemotypes, including nonselective and COX-2-selective scaffolds [3,20]. Selected natural anti-inflammatory products were included as a complementary non-NSAID panel to sample diverse plant-derived and nutraceutical scaffolds, including curcuminoids, flavonoids, stilbenes, ginger phenolics, terpenoids, boswellic acids and salicin [8,21,22]. The combined panel contextualises catalytic-channel compatibility across established NSAID and natural-product anti-inflammatory chemical space.

The aim of this study was to define how global human PTGS1 and PTGS2 coding variation is organised across population-genetic, transcript and protein contexts. The analysis integrated allele-resolved population differentiation, transcript-aware consequence annotation, splice-prediction boundary checks, direct PTGS1 leader-sequence property calculations and controlled PTGS2 structural analyses to distinguish contextual population markers from protein-contextual candidates. Population differentiation was used for description and prioritisation, not as evidence of molecular function.

Results

Global PTGS variation reveals paralog-specific population-genetic architecture

To define the population-genetic architecture of human PTGS variation, this study integrated global genotype variation, transcript-aware annotation, population differentiation metrics, consequence-first prioritisation, protein-domain mapping, splice-prediction boundary checks and restrained structural interpretation into a unified analytical framework (Fig 1A). This design allowed PTGS1 and PTGS2 to be evaluated not only as homologous inflammatory enzymes, but also as a test case for distinguishing population-differentiated markers from evidence-bounded functional pharmacogenomic candidates.

Mapping prioritised coding variants onto PTGS paralog architecture identified distinct candidate classes across the two enzymes (Fig 1B and Table 1). PTGS1 Trp8Arg and Pro17Leu mapped to the N-terminal leader sequence, motivating direct comparison of sequence physicochemical properties rather than a demonstrated targeting or maturation mechanism. PTGS2 Val511Ala mapped to the catalytic cyclooxygenase/peroxidase domain; canonical Val511 corresponds to 5KIR Val525, a second-shell, channel-adjacent position that motivated controlled structural follow-up. The differentiated synonymous variants PTGS1 Gly213Gly and PTGS2 Val102Val were interpreted as high-FST contextual coding markers that describe population structure without changing amino-acid identity.

Table 1. Functional PTGS candidate classes defined by consequence-first prioritisation.

Candidate class	No.	Genes	Representative variants	Max FST	Evidence-bounded interpretation
High-FST contextual coding marker	5	PTGS1, PTGS2	PTGS1 p.Gly213Gly (c.639C > A); PTGS1 p.Gln41Gln (c.123G > A); PTGS2 p.Val102Val (c.306G > C); PTGS2 p.His403His (c.1209T > C); plus 1 additional candidate	0.601	Population-differentiated synonymous coding markers retained as contextual locus-level signals rather than direct protein-altering candidates.
Signal-peptide missense candidate	4	PTGS1, PTGS2	PTGS1 p.Trp8Arg (c.22T > C); PTGS1 p.Pro17Leu (c.50C > T); PTGS1 p.Trp8Ser (c.23G > C); PTGS2 p.Ala3Pro (c.7G > C)	0.184	N-terminal signal-peptide missense candidates interpreted as protein-biogenesis hypotheses requiring experimental validation.
Catalytic-channel missense candidate	1	PTGS2	PTGS2 p.Val511Ala (c.1532T > C)	0.040	PTGS2 active-site-channel candidate supporting catalytic-channel structural follow-up.
Additional protein-topology missense candidate	24	PTGS1, PTGS2	PTGS1 p.Leu237Met (c.709C > A); PTGS1 p.Lys185Thr (c.554A > C); PTGS2 p.Gly587Arg (c.1759G > A); plus 21 additional candidates	0.036	Additional missense candidates recovered by consequence-first filtering and prioritised by protein topology and population context.
Splice-region candidate downgraded by prediction boundary	3	PTGS1, PTGS2	PTGS2 c.724–10_724–7delATTT; PTGS1 p.Pro71Ser (c.211C > T); PTGS1 p.Arg119Ser (c.355C > A)	0.052	Splice-region annotations with weak or absent web SpliceAI/Pangolin support; interpreted as boundary candidates, not evidence of altered splicing.
Splice-region candidate requiring prediction/validation	3	PTGS1, PTGS2	PTGS2 c.458-8T > C; PTGS1 c.1009 + 8C > T; PTGS2 c.970 + 3A > G	0.003	Splice-region candidates requiring additional computational or experimental validation before functional interpretation.

Open in a new tab

The table summarises the final functional candidate classes generated after transcript-aware consequence filtering, allele-resolved population differentiation analysis, topology-aware interpretation and splice-prediction boundary checking. Counts indicate the number of variants retained in each evidence class. Representative variants are shown to illustrate each class, while the complete corrected candidate list is provided in S1 Table in S1 File. FST, fixation index; PTGS, prostaglandin-endoperoxide synthase.

Across global superpopulations, PTGS variant burden showed broad population structure while preserving paralog-specific differences (Fig 1C). African populations carried the highest population-present variant burden, whereas East Asian populations showed comparatively lower burden, with admixed American, European, and South Asian groups occupying intermediate or distinct profiles. This pattern provided the first indication that PTGS diversity is not uniformly distributed across human populations, and that PTGS1 and PTGS2 may differ subtly in the scale and distribution of population-present variation. A multi-metric comparison of conservation and diversification further supported paralog-specific evolutionary structure (Fig 1D). PTGS1 showed slightly higher mean population allele-frequency standard deviation, allele-frequency range, allele entropy, and mean pairwise FST than PTGS2. Although these differences were modest, their consistency across complementary metrics suggested that PTGS1 carries a somewhat broader population-diversification signature, whereas PTGS2 appears more tightly constrained at the global locus scale. Together, these analyses established a population-genetic foundation for subsequent mechanistic interpretation: PTGS diversification is not simply a question of variant frequency, but of where differentiated variants occur within the functional topology of each paralog.

Consequence-first candidate classes link population differentiation to protein topology and splice-prediction support

Prioritised PTGS variants were consolidated into a candidate-class table linking variant identity, population contrast, coding consequence, transcript context, protein-domain or splice-region context, maximum pairwise FST and final evidence category (Table 1). This organisation connected population differentiation with molecular consequence while keeping FST descriptive rather than functional. Candidate categories included contextual coding variation, N-terminal leader-sequence variation, channel-adjacent structural context, other protein-topology context and splice-region boundary evidence.

The candidate comparison showed that strong population differentiation and protein-contextual interpretability do not necessarily coincide. Among 31 missense candidates, PTGS1 p.Trp8Arg and p.Pro17Leu were classified as N-terminal leader-sequence candidates. PTGS2 p.Val511Ala was selected for structural follow-up because it was the most population-visible of four channel-context candidates and mapped unambiguously to a human ligand-bound structure; it was not the highest-FST missense variant overall. PTGS1 p.Gly213Gly and PTGS2 p.Val102Val captured marked coding differentiation without changing amino-acid identity and were therefore retained as contextual population markers. The differentiation-first versus consequence-first comparison is provided in S2 Table in S1 File, and aggregate prioritisation metrics are provided in S4 Table in S1 File.

The same framework identified 35 additional consequence-first candidates with interpretable coding, topology or splice-region context. These included additional missense candidates prioritised by protein topology and population context, splice-region candidates requiring further prediction or validation, and splice-region annotations with weak SpliceAI/Pangolin support. The result is a richer PTGS candidate map in which population differentiation highlights structured variation, while transcript consequence and protein topology identify the most biologically informative follow-up axes. The complete candidate set and integrated variant-level evidence are provided in S1 and S5 Tables in S1 File.

SpliceAI/Pangolin analysis defines splice-region boundary candidates

Seven PTGS variants with splice-region or comparator relevance were evaluated using web-based SpliceAI/Pangolin prediction. These included principal-transcript splice-region candidates, non-principal isoform audit findings and non-splice comparator variants. No evaluated variant reached a SpliceAI maximum delta score of 0.20, the permissive threshold used here for notable computational splice support. Detailed web-based prediction outputs are provided in S3 Table in S1 File.

The key PTGS2 splice-region deletion, c.724–10_724–7delATTT, remained a population-variable principal-transcript splice-region variant, but its web-prediction scores did not support a strong splice-altering effect. Its maximum SpliceAI delta score was 0.01 and its maximum Pangolin score was 0.05. Therefore, this deletion should be interpreted as a splice-region boundary candidate requiring further validation, not as evidence of altered splicing. The curated splice-prediction boundary evidence supporting this interpretation is provided in S3 Table in S1 File.

PTGS2 Val511Ala maps to a catalytic-channel environment sampled by NSAIDs and natural anti-inflammatory products

Wild-type (WT) and Val511Ala PTGS2 models were compared within the catalytic-channel context to determine how the variant sits relative to the inhibitor-binding environment (Fig 2A and 2D). Val511Ala localized within the cyclooxygenase/peroxidase domain and near the active-site channel, placing it in a region directly relevant to substrate access and NSAID accommodation. This spatial position supports prioritisation of Val511Ala as a catalytic-channel candidate for structural and biochemical follow-up.

Structural close-up analysis indicated that Val511Ala could plausibly alter local packing geometry without producing a gross structural rearrangement (Fig 2D). Because valine has a larger branched hydrophobic side chain than alanine, the substitution may subtly change side-chain volume and local hydrophobic packing within the channel-adjacent environment. This provides a concrete structural rationale for evaluating the variant across ligand chemotypes.

Docking analyses therefore used two complementary ligand groups (Fig 2B and 2C). Common NSAIDs were included to sample clinically relevant cyclooxygenase inhibitor chemotypes, whereas selected natural anti-inflammatory products were included to sample non-NSAID, plant-derived and nutraceutical anti-inflammatory chemical space. The natural-product panel included curcuminoids, flavonoids, stilbenes, ginger phenolics, terpenoids, boswellic acids, salicin and related compounds. Across this ligand panel, the docking-score landscape provided an exploratory view of how Val511Ala may relate to catalytic-channel accommodation, with rofecoxib retained as an illustrative modelling observation rather than as a single defining result. The paired WT-versus-Val511Ala docking-statistics summary is provided in S6 Table in S1 File.

Together, these analyses place PTGS2 Val511Ala within a functionally meaningful catalytic-channel context. The result does not depend on one compound alone; instead, it identifies a channel-proximal coding variant whose structural position can be interpreted across pharmacological NSAID scaffolds and natural-product anti-inflammatory chemical space. This makes Val511Ala a focused candidate for future enzyme, binding or cellular pharmacology assays.

Evolutionary and protein-context integration distinguishes PTGS1 leader-sequence and PTGS2 channel-adjacent candidate contexts

Population differentiation was integrated with protein topology to organise candidate contexts (Fig 3A). High-FST synonymous variants described contextual coding differentiation, PTGS1 leader-sequence missense variants defined a directly measurable sequence-property context and PTGS2 Val511Ala defined a second-shell, channel-adjacent structural context. These categories guide prioritisation; they do not establish functional effects.

Evolutionary constraint modelling added background to this topology-aware interpretation (Fig 3B). PTGS2 Val511Ala occurred within a constrained catalytic-region landscape, whereas PTGS1 Trp8Arg and Pro17Leu occurred within the N-terminal leader sequence. This contrast identifies different protein locations and experimental questions, but it does not by itself demonstrate altered protein handling, catalytic activity or drug response.

Population-functional burden mapping provided an additional synthesis layer (Fig 3C). Prioritized PTGS variants showed different population-differentiation architectures across superpopulations, but the strongest mechanistic candidates were not simply those with the highest differentiation values. Instead, mechanistic priority emerged from the intersection of population differentiation, coding consequence, and functional topology.

An integrative model summarised this evidence hierarchy (Fig 3D). PTGS1 candidates were represented as N-terminal leader-sequence variants with directly calculable physicochemical differences, whereas PTGS2 Val511Ala was represented as a channel-adjacent structural candidate. Population differentiation was treated as a discovery signal whose interpretation depends on transcript consequence, protein context and reproducible quantitative analysis; downstream function remains to be tested.

A focused structural pharmacogenomics analysis mapped canonical PTGS2 Val511 to chain A Val525 in the rofecoxib-bound 5KIR structure and evaluated its second-shell geometry using paired docking-score, direct-frame pose-recovery and contact-set comparisons (Fig 4A–4D). Val525 was adjacent to, but not a direct member of, the crystallographic rofecoxib-contact network, and no systematic variant-associated shift was detected under the tested docking conditions.

Fig 4 — (A) Val511Ala was selected from four channel-context missense candidates based on population visibility, channel context and availability of a ligand-bound structural model; FST is used only as a population-differentiation measure. (B) Canonical Val511 maps to chain A Val525 in PDB 5KIR, a second-shell position adjacent to, but not directly contacting, the crystallographic rofecoxib-contact network; minimum any-heavy-atom and side-chain distances were 6.004 Å and 7.559 Å, respectively. (C) Across five paired docking seeds, WT and Val525Ala showed similar docking scores and direct-frame pose recovery, with mean paired differences of −0.047600 kcal/mol and −0.003538 Å, respectively. (D) Contact-set Jaccard similarity was 0.952941, 0.972727 and 0.915269 at 4-, 5- and 6-Å thresholds; direct 4-Å residue-525 occupancy was 0/5 in both groups, and side-chain distance ranges overlapped. These analyses support structural mapping and reproducibility assessment but do not establish altered affinity, catalysis, pharmacology or clinical effects.

PTGS1 Trp8Arg and Pro17Leu produce distinct leader-sequence physicochemical changes

Because the prioritised PTGS1 missense variants mapped to the N-terminal 23-residue leader sequence rather than the catalytic domain, their directly calculable sequence properties were compared (Fig 5A–5C). The affected wild-type residues are Trp8 and Pro17. Tryptophan is aromatic and hydrophobic, whereas proline is a cyclic residue that constrains local backbone geometry. Wild-type, Trp8Arg and Pro17Leu sequences were aligned, and net-charge proxy, positive-residue count, hydrophobic-residue count, mean Kyte-Doolittle hydropathy, aromatic-residue count and proline count were calculated across the full leader sequence and the descriptive H-region. The N-, H- and C-region partitions are computational descriptions rather than experimentally established boundaries. No signal-peptide predictor output was used. The underlying property-shift metrics are provided in S7 Table in S1 File.

The two substitutions produced different directly calculated physicochemical changes. Relative to WT, Trp8Arg increased the full-leader net-charge proxy from 1 to 2, reduced the hydrophobic-residue count from 17 to 16, lowered mean Kyte-Doolittle hydropathy from 1.98 to 1.82 and reduced the aromatic-residue count from 3 to 2, while the proline count remained 3. Pro17Leu left the net-charge proxy and aromatic-residue count unchanged, increased the hydrophobic-residue count from 17 to 18 and mean hydropathy from 1.98 to 2.21, and reduced the proline count from 3 to 2. Within the descriptive H-region, Trp8Arg increased the positive-residue count from 0 to 1, reduced the hydrophobic-residue count from 11 to 10, lowered mean hydropathy from 2.45 to 2.18 and reduced the aromatic-residue count from 3 to 2, while the proline count remained 2. Pro17Leu left the positive- and aromatic-residue counts unchanged, increased the hydrophobic-residue count from 11 to 12 and mean hydropathy from 2.45 to 2.87, and reduced the proline count from 2 to 1 (Fig 5B and 5C and S7 Table in S1 File). These values are direct sequence calculations and do not demonstrate altered SRP recognition, ER targeting, translocation, membrane insertion or maturation.

Both substitutions fall within the descriptive H-region spanning residues 6–18 (Fig 5A and 5D). Their positions therefore support comparison of local leader-sequence properties, but do not establish altered localisation, processing or protein handling. This PTGS1 sequence context is analytically distinct from the second-shell, channel-adjacent location of PTGS2 Val511Ala. The positional contrast does not by itself establish a difference in biological effect.

Together, the PTGS1 analysis establishes that Trp8Arg and Pro17Leu alter directly calculable leader-sequence properties in different directions. The analysis does not demonstrate effects on SRP recognition, ER targeting, translocation, cleavage, membrane insertion, maturation, abundance, localisation, enzyme activity or drug response. These downstream outcomes remain hypotheses for experimental testing.

Discussion

This study shows that integrating population differentiation with transcript consequence, protein topology, splice-prediction evidence and structural ligand context yields distinct evidence categories rather than demonstrated mechanistic axes. Synonymous coding variants serve as contextual population markers; PTGS1 Trp8Arg and Pro17Leu show direct leader-sequence property differences; and PTGS2 Val511Ala is a tractable channel-adjacent structural candidate for which controlled docking, contact and pose analyses detected no systematic variant-associated shift. Additional consequence-first candidates broaden the set available for future functional prioritisation [13,14,23].

These findings can be interpreted against the established biology of PTGS paralogs, which share cyclooxygenase chemistry but differ in regulatory behaviour, inducibility, cellular context and pharmacological sensitivity [7,8,12]. The present analysis identifies different protein contexts for human variation, but it does not show that the PTGS1 variants alter targeting or processing, or that PTGS2 Val511Ala alters ligand response. This calibrated separation converts population-genetic observations into specific, testable hypotheses.

High FST variants in this analysis provide an informative map of PTGS population structure. Some of the strongest differentiated coding signals were synonymous variants, including PTGS1 Gly213Gly and PTGS2 Val102Val, indicating that coding-region differentiation can capture locus-level population architecture even when amino-acid sequence is unchanged. These markers are therefore useful for understanding PTGS haplotype and differentiation patterns, while missense and splice-region candidates provide complementary routes toward functional interpretation [10,13–15].

The PTGS1 candidates highlight N-terminal leader-sequence variation. Trp8Arg increased the net-charge proxy and positive-residue count while reducing hydrophobicity and aromatic-residue count. Pro17Leu increased hydrophobicity and removed one proline without changing the net-charge proxy. Because N-terminal signal sequences generally participate in protein targeting [24,25], these direct sequence-property changes motivate future cellular and biochemical experiments. However, no signal-peptide predictor output was used, and effects on SRP recognition, ER targeting, translocation, cleavage, membrane insertion, maturation, abundance or localisation were not measured in this study.

PTGS2 Val511Ala defines a complementary catalytic-channel axis. Val511 lies within the cyclooxygenase/peroxidase domain near the active-site-channel environment, and the valine-to-alanine substitution reduces side-chain volume while preserving hydrophobic character. This makes the variant structurally meaningful for local packing and ligand-channel interpretation. The docking panel was designed to contextualise this environment across two anti-inflammatory chemical spaces: common NSAIDs, which represent clinically established COX inhibitor chemotypes, and selected natural anti-inflammatory products, which represent structurally diverse plant-derived or nutraceutical scaffolds with reported inflammation-related activity. The resulting docking landscape supports Val511Ala as a focused structural candidate for experimental enzyme, binding or cellular pharmacology assays.

The splice-prediction results add a third interpretation layer. PTGS2 c.724–10_724–7delATTT remained a population-variable principal-transcript splice-region deletion, while SpliceAI and Pangolin scores indicated limited computational support for altered splicing. This places the deletion in a splice-region boundary category: it is biologically notable because of its transcript position and population variability, and it is experimentally approachable through RNA-level assays, but the available prediction profile prioritises it for validation rather than immediate functional assignment [18,19].

The 35 additional consequence-first candidates show that functional prioritisation becomes richer when coding consequence and topology are considered alongside population differentiation. A purely frequency-centred view highlights differentiated markers, whereas a consequence-first view recovers variants with lower differentiation but stronger protein-contextual relevance. For pharmacogenomic discovery, these approaches are complementary: population differentiation identifies structured variation, and consequence/topology mapping indicates which variants are most suitable for mechanistic follow-up.

Several features should guide interpretation of these findings. The analysis is computational and is intended to prioritise testable hypotheses rather than substitute for biochemical, cellular, transcriptomic or clinical validation. Docking and structural modelling provide spatial and chemical context for Val511Ala, while splice-prediction analyses provide a computational boundary for RNA-level follow-up. Comparator-gene generalisation remains an extensibility module for future work, and population datasets remain shaped by available reference genomes, transcript models, allele-frequency resources and variant-calling assumptions.

Overall, the study provides a practical framework for moving from population-genetic discovery to calibrated pharmacogenomic prioritisation. The framework preserves population-differentiation information while adding transcript consequence, protein topology, splice-prediction evidence, direct leader-sequence property calculations and controlled channel mapping, docking, contact and pose comparisons. In PTGS1, this approach highlights N-terminal sequence-property candidates; in PTGS2, it highlights a channel-adjacent structural candidate for which no systematic docking shift was detected under the tested conditions. The design can be extended to broader pharmacogene panels requiring joint evaluation of population differentiation and protein context.

Conclusion

In conclusion, global PTGS variation separates into contextual population markers and protein-contextual candidates. PTGS1 Trp8Arg and Pro17Leu show distinct directly calculated leader-sequence physicochemical changes, whereas PTGS2 Val511Ala is a second-shell, channel-adjacent structural candidate for which no systematic docking-score, contact or pose shift was detected under the tested conditions. By combining population differentiation with transcript consequence, protein context, splice-prediction evidence and reproducible quantitative analyses, this study provides a calibrated prioritisation framework for future PTGS functional and pharmacogenomic investigation.

Methods

Study design and analytical overview

This study used an integrative computational pharmacogenomics framework to compare the population-genetic, transcript and protein-context architecture of PTGS1 and PTGS2. The workflow combined global human variant data, population-level genotype summaries, pairwise population differentiation, transcript-aware functional annotation, a retrospective comparison of 31 missense candidates, protein-domain mapping, splice-prediction boundary analysis, crystal-structure mapping, controlled docking-score, contact and direct-frame pose comparisons, and direct PTGS1 leader-sequence property calculations. FST described population structure and was not treated as functional evidence.

Variant datasets and population-level genotype matrices

PTGS1 and PTGS2 variants were extracted from global human population variant resources and summarized into gene-specific and combined population genotype matrices [15]. Population-level mean genotype profiles were generated for PTGS1, PTGS2, and combined PTGS loci. Superpopulation groupings were used to compare major global population structures, including African, admixed American, East Asian, European, and South Asian groups. The complete integrative variant-level dataset is provided in S5 Table in S1 File.

Population differentiation and conservation metrics

Population differentiation was evaluated using pairwise FST-based summaries across PTGS loci [13,14]. Top differentiated variants and hotspot windows were identified and annotated by genomic and transcript region. Conservation and diversification metrics were calculated for each paralog using population allele-frequency dispersion, allele-frequency range, allele entropy, mean pairwise FST, median pairwise FST, and maximum pairwise FST. PTGS1-versus-PTGS2 differentiation distributions were compared using a Mann–Whitney U test. For per-variant differentiation ranking, FST estimates were interpreted as effect-size summaries of allele-frequency divergence rather than as direct evidence of selection. Gene-level conservation and diversification summaries were derived by aggregating retained pairwise estimates and allele-frequency dispersion measures across PTGS1 and PTGS2. Negative or unavailable differentiation estimates were not treated as functional evidence; mechanistic interpretation required transcript consequence, protein topology or splice-prediction context.

Transcript-aware and coding consequence annotation

Differentiated variants were mapped to transcript and coding sequence context using gene annotation resources. Coding variants were classified by predicted consequence, including synonymous and missense effects. High-FST coding variants were further annotated using snpEff-derived effect classes and integrated with exon/CDS coordinates [17]. Coding-versus-noncoding enrichment patterns were compared between PTGS1 and PTGS2 using a chi-square test.

Consequence-first prioritisation and FST-first comparison

A consequence-first prioritisation framework was implemented to connect population differentiation with molecular consequence. Variants were first classified by transcript-aware coding consequence, principal-transcript status and protein-topology context, and population differentiation was then integrated as an evidence layer. This allowed high-FST synonymous coding markers to be separated from missense, splice-region and structural candidates with more direct biological interpretability.

Candidate classes were defined using biological variant identity rather than exact identifier matching when necessary, because some source identifiers differed in allele representation. Biological identity was defined by gene, HGVS.c and HGVS.p. Candidates were grouped as contextual coding markers, N-terminal leader-sequence missense candidates, additional protein-topology missense candidates, splice-region boundary candidates or PTGS2 channel-adjacent structural candidates. Population-differentiation measures were retained as descriptive prioritisation variables rather than functional scores.

Splice-prediction boundary analysis

Splice-region and splice-relevant PTGS candidates were evaluated using web-based SpliceAI and Pangolin prediction tools [22,23]. Seven validated hg38/GRCh38 PTGS alleles were submitted, including principal-transcript splice-region candidates, non-principal isoform audit variants and non-splice comparator variants. Local SpliceAI execution was not used for final scoring after machine-safety interruptions during test runs; therefore, the reported splice scores derive from the web-based prediction outputs recorded in the project audit tables [18,19]. The curated web SpliceAI/Pangolin scoring table is provided in S3 Table in S1 File.

SpliceAI maximum delta scores were interpreted using predefined evidence categories: values below 0.20 were treated as no notable computational splice-prediction support, 0.20 to below 0.50 as permissive computational support, 0.50 to below 0.80 as substantial computational support and 0.80 or higher as strong computational support. Pangolin scores were used as supporting computational evidence only. No splice-prediction result was interpreted as experimental proof of altered splicing.

Comparator-gene generalisation resource audit

A compact comparator-gene generalisation module was planned to evaluate whether the FST-first versus consequence-first interpretive lesson extends to pharmacogenes with different architectures, including VKORC1, CYP2C9, SLCO1B1, DPYD and TPMT. A local resource audit showed that genome-wide comparator-gene VCF resources were not available within the project at the time of revision, although supporting sample metadata and annotation resources were present. Therefore, comparator-gene generalisation was retained as an extensibility framework rather than executed as a full empirical comparator analysis in this manuscript.

Functional-domain and topology mapping

Prioritised coding variants were mapped onto PTGS protein architecture using a manually curated domain map. Protein contexts included the N-terminal leader sequence, membrane-binding domain, catalytic cyclooxygenase/peroxidase domain, catalytic residues and channel-adjacent regions. Variants were classified by protein context and evidence type rather than by an assumed functional axis. For PTGS2 Val511Ala, canonical-to-structure mapping was resolved to 5KIR Val525, and proximity to the crystallographic rofecoxib ligand and surrounding contact network was quantified.

PTGS2 structural modeling and docking analyses

PTGS2 wild-type (WT) and Val511Ala structural models were used to compare ligand accommodation in the catalytic-channel context. Human PTGS2 was referenced to UniProt P35354 and AlphaFold model AF-P35354-F1 [26], with active-site interpretation guided by available cyclooxygenase-2 structural information, including the ibuprofen-bound COX-2 structure PDB 4PH9 [27]. Ligand structures were retrieved as PubChem SDF records where available and converted/prepared for docking using the project cheminformatics workflow. The NSAID panel was selected to represent common clinically used cyclooxygenase inhibitors and COX-2-relevant scaffolds, including aspirin (CID 2244), ibuprofen (CID 3672), naproxen (CID 156391), diclofenac (CID 3033), indomethacin (CID 3715), celecoxib (CID 2662) and rofecoxib (CID 5090). The natural-product panel was selected to represent structurally diverse anti-inflammatory phytochemical and nutraceutical-derived compounds reported in COX-, NF-kB-, cytokine-or inflammation-related literature, including curcuminoids, flavonoids, stilbenes, ginger phenolics, phenylpropanoids, terpenoids, boswellic acids, salicin and related plant-derived scaffolds. Representative examples included curcumin, demethoxycurcumin, bisdemethoxycurcumin, resveratrol, pterostilbene, quercetin, kaempferol, luteolin, apigenin, baicalein, baicalin, wogonin, epigallocatechin gallate, naringenin, hesperetin, silybin, 6-gingerol, 6-shogaol, zingerone, cinnamaldehyde, eugenol, thymol, carvacrol, alpha-boswellic acid, beta-boswellic acid, acetyl-11-keto-beta-boswellic acid, ursolic acid, oleanolic acid, parthenolide and salicin. Receptors and ligands were prepared by adding hydrogens, assigning protonation/charge states as supported by Open Babel (v3.1.1) and RDKit (2023.09 series), generating three-dimensional conformers for ligands where needed and defining a grid around the catalytic-channel region. Docking was performed with AutoDock Vina (v1.2.5; Trott and Olson framework) through the project workflow, and docking-score differences were calculated as Val511Ala minus WT. Ranked affinity plots, WT-versus-mutant scatter plots and delta-score response landscapes were used to compare ligand accommodation across NSAID and natural anti-inflammatory product panels [28]. The full ligand-level docking-statistics summary is provided in S6 Table in S1 File.

PTGS2 docking statistics

Docking-score distributions between WT and Val511Ala PTGS2 were compared using paired Wilcoxon signed-rank tests for NSAIDs, selected natural anti-inflammatory products and the combined ligand set. Statistical outputs included the number of ligands, median WT docking score, median Val511Ala docking score, median delta, mean delta, test statistic and p-value. These analyses were used to evaluate whether the variant produced a consistent ligand-panel shift or a compound-specific structural signal within the catalytic-channel model. The detailed statistical output is provided in S6 Table in S1 File.

PTGS2 rofecoxib structural contact analysis

Rofecoxib pose-contact summaries were retained as an illustrative structural example because rofecoxib is a COX-2-selective inhibitor scaffold and showed an interpretable local contact pattern in the Val511Ala modelling audit. WT and Val511Ala rofecoxib pose-contact summaries were used to examine local channel geometry and side-chain packing around the catalytic-channel environment. These comparisons provide a focused structural example within the broader NSAID and natural anti-inflammatory product docking panel.

PTGS1 leader-sequence physicochemical analysis

PTGS1 Trp8Arg and Pro17Leu were evaluated using direct sequence-property calculations. The 23-residue sequences were WT, MSRSLLLWFLLFLLLLPPLPVLL; Trp8Arg, MSRSLLLRFLLFLLLLPPLPVLL; and Pro17Leu, MSRSLLLWFLLFLLLLLPLPVLL. Residues 1–5, 6–18 and 19–23 were treated as descriptive N-, H- and C-region partitions, respectively, rather than experimentally validated boundaries. For the full 23-residue sequence and the descriptive H-region, the analysis calculated net-charge proxy, positive-residue count, hydrophobic-residue count, mean Kyte-Doolittle hydropathy, aromatic-residue count and proline count. Five-residue windows centred on each substitution were also compared, and variant-minus-WT differences were reported. No signal-peptide predictor or cleavage-site output was used. The analysis therefore did not estimate effects on SRP recognition, ER targeting, translocation, membrane insertion, maturation, abundance, localisation, enzyme activity or drug response. The complete property table is provided in S7 Table in S1 File.

Statistical summary integration

Major statistical comparisons were consolidated into an integrative statistical summary table (S4 Table in S1 File). PTGS1-versus-PTGS2 pairwise FST distributions were compared using the Mann-Whitney U test because the differentiation summaries were not assumed to be normally distributed. Coding-versus-noncoding distributional differences between PTGS genes were evaluated using a chi-square test. WT-versus-Val511Ala paired docking-score distributions were compared using paired Wilcoxon signed-rank tests for NSAIDs, selected natural anti-inflammatory products and all ligands combined; ligand-level docking statistics are provided in S6 Table in S1 File. Statistical significance was evaluated at alpha = 0.05. Because the tests represented a small set of predefined, distinct analytical comparisons used for evidence summarisation rather than genome-wide association testing, p-values were interpreted descriptively and no formal multiple-testing correction was applied.

Ethics statement

This study used publicly available, de-identified population-genetic, annotation and structural resources and did not involve recruitment of human participants, collection of new human samples or generation of new individual-level clinical data. Formal institutional ethics approval was therefore not required.

Figure generation and reproducibility

Figures were generated using reproducible Python scripts and command-line workflows. High-resolution PNG, PDF, and SVG outputs were generated where possible. Final figure composites were assembled from panel-level outputs and organized under results/figures/final_panels. All scripts, tables, and figure outputs were tracked using Git to preserve reproducibility and version history.

Supporting information

S1 File. Combined Supporting Tables S1–S7.

Excel workbook containing the complete supporting tables used in this study: S1, full consequence-first PTGS functional candidate table; S2, population-differentiation and consequence-first PTGS candidate comparison; S3, web SpliceAI/Pangolin prediction-boundary summary for PTGS splice-relevant and comparator variants; S4, summary metrics for population-differentiation and consequence-first prioritisation; S5, integrative PTGS variant master table; S6, PTGS2 WT-versus-Val511Ala docking statistics; and S7, PTGS1 leader-sequence physicochemical property changes.

(XLSX)

pone.0358222.s001.xlsx^{(100.1KB, xlsx)}

Acknowledgments

The author has no acknowledgments to report.

Data Availability

All analyses were performed using publicly available human population-variation, annotation, protein-structure and web-prediction resources as described in the Methods. Derived tables, audit files and manuscript-revision outputs generated during this study are organized within the project repository. The repository is available at: https://github.com/rmaghembe1/ptgs-evolutionary-pharmacogenomics. The main derived datasets underlying the paper are also provided as Supporting information files S1-S7 Tables.

Funding Statement

The author(s) received no specific funding for this work.

References

1.Nagao M, Sato Y, Yamauchi A. A meta-analysis of PTGS1 and PTGS2 polymorphisms and NSAID intake on the risk of developing cancer. PLoS One. 2013;8(8):e71126. doi: 10.1371/journal.pone.0071126 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Smith WL, Garavito RM, DeWitt DL. Prostaglandin endoperoxide H synthases (cyclooxygenases)-1 and -2. J Biol Chem. 1995;270:14843–7. doi: 10.1074/jbc.270.24.14843 [DOI] [PubMed] [Google Scholar]
3.Blobaum AL, Marnett LJ. Structural and functional basis of cyclooxygenase inhibition. J Med Chem. 2007;50(7):1425–41. doi: 10.1021/jm0613166 [DOI] [PubMed] [Google Scholar]
4.Cebrián-Prats A, González-Lafont À, Lluch JM. Unraveling the molecular details of the complete mechanism that governs the synthesis of prostaglandin G2 catalyzed by cyclooxygenase-2. ACS Omega. 2019;4:2063–74. doi: 10.1021/acsomega.8b03575 [DOI] [Google Scholar]
5.Trinh HKT, Pham LD, Le KM, Park HS. Pharmacogenomics of hypersensitivity to non-steroidal anti-inflammatory drugs. Front Genet. 2021;12. doi: 10.3389/fgene.2021.647257 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Sheehan BJ, Edwards B, Medrano IS, El-Saidi MA, Zaidan WR, El-Ezzi AA, et al. Association between two single nucleotide polymorphisms of the Prostaglandin-Endoperoxide Synthase 1 and 2 genes and cell proliferative prostatic diseases in Lebanon. Oncotarget. 2025;16:262–72. doi: 10.18632/oncotarget.28710 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Lucena MI, García-Martín E, Daly AK, Blanca M, Andrade RJ, Agúndez JAG. Next-generation sequencing of PTGS genes reveals an increased frequency of non-synonymous variants among patients with NSAID-induced liver injury. Front Genet. 2019;10. doi: 10.3389/fgene.2019.00134 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Wosniaki DK, Kusma BN, Marin AM, Munhoz EC, Farias JS de H, Mendes EB, et al. Divergent gene expression of PTGS1 and PTGS2 along the disease course of chronic myeloid leukaemia. Front Oncol. 2025;15:1556956. doi: 10.3389/fonc.2025.1556956 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Li X, Mazaleuskaya LL, Yuan C, Ballantyne LL, Meng H, Smith WL, et al. Flipping the cyclooxygenase (Ptgs) genes reveals isoform-specific compensatory functions. J Lipid Res. 2018;59(1):89–101. doi: 10.1194/jlr.M079996 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Thorn CF, Grosser T, Klein TE, Altman RB. PharmGKB summary: Very important pharmacogene information for PTGS2. Pharmacogenet Genom. 2011;21(9):607–13. doi: 10.1097/FPC.0b013e3283415515 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Zhang C, Jiang X, Chen W, Li Q, Yun F, Yang X, et al. Population genetic difference of pharmacogenomic VIP gene variants in the Lisu population from Yunnan Province. Medicine (Baltimore). 2018;97(52):e13674. doi: 10.1097/MD.0000000000013674 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Yucesoy B, Kashon ML, Johnson VJ, Lummus ZL, Fluharty K, Gautrin D, et al. Genetic variants in TNFα, TGFB1, PTGS1 and PTGS2 genes are associated with diisocyanate-induced asthma. J Immunotoxicol. 2016;13(1):119–26. doi: 10.3109/1547691X.2015.1017061 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Bhatia G, Patterson N, Sankararaman S, Price AL. Estimating and interpreting FST: The impact of rare variants. Genome Res. 2013;23(9):1514–21. doi: 10.1101/gr.154831.113 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Holsinger KE, Weir BS. Genetics in geographically structured populations: Defining, estimating and interpreting F(ST). Nat Rev Genet. 2009;10(9):639–50. doi: 10.1038/nrg2611 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.1000 Genomes Project Consortium, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. doi: 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Ota M, Spence JP, Zeng T, Dann E, Milind N, Marson A, et al. Causal modelling of gene effects from regulators to programs to traits. Nature. 2026;650(8101):399–408. doi: 10.1038/s41586-025-09866-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Cingolani P, Platts A, Wang LL, Coon M, Nguyen T, Wang L, et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff. Fly. 2012;6(2):80–92. doi: 10.4161/fly.19695 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.de Sainte Agathe J-M, Filser M, Isidor B, Besnard T, Gueguen P, Perrin A, et al. SpliceAI-visual: A free online tool to improve SpliceAI splicing variant interpretation. Hum Genom. 2023;17(1):7. doi: 10.1186/s40246-023-00451-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Zeng T, Li YI. Predicting RNA splicing from DNA sequence using Pangolin. Genome Biol. 2022;23(1):103. doi: 10.1186/s13059-022-02664-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Orlando BJ, Lucido MJ, Malkowski MG. The structure of ibuprofen bound to cyclooxygenase-2. J Struct Biol. 2015;189(1):62–6. doi: 10.1016/j.jsb.2014.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Zhang L, Siyiti M, Zhang J, Yao M, Zhao F. Anti-inflammatory and anti-rheumatic activities in vitro of alkaloids separated from Aconitum soongoricum Stapf. Exp Ther Med 2021. doi: 10.3892/etm.2021.9924 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Goel A, Boland CR, Chauhan DP. Specific inhibition of cyclooxygenase-2 (COX-2) expression by dietary curcumin in HT-29 human colon cancer cells. Cancer Lett. 2001;172(2):111–8. doi: 10.1016/s0304-3835(01)00655-3 [DOI] [PubMed] [Google Scholar]
23.Cingolani P. Variant annotation and functional prediction: SnpEff. In: Ng C, Piscuoglio S, editors. Variant calling: Methods and protocols. New York (NY): Springer US; 2022. p. 289–314. doi: 10.1007/978-1-0716-2293-3_19 [DOI] [PubMed] [Google Scholar]
24.Hegde RS, Kang SW. The molecular code of the signal peptide and its recognition by the signal recognition particle. Genes Dev. 2006;20:2377–89. doi: 10.1101/gad.1493006 [DOI] [Google Scholar]
25.Kellogg MK, Tikhonova EB, Karamyshev AL. Signal recognition particle in human diseases. Front Genet. 2022;13. doi: 10.3389/fgene.2022.898083 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Jain P, Satija J, Sudandiradoss C. Discovery of andrographolide hit analog as a potent cyclooxygenase-2 inhibitor through consensus MD-simulation, electrostatic potential energy simulation and ligand efficiency metrics. Sci Rep. 2023;13(1):8147. doi: 10.1038/s41598-023-35192-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Orlando BJ, Lucido MJ, Malkowski MG. The structure of ibuprofen bound to cyclooxygenase-2. J Struct Biol. 2015;189(1):62–6. doi: 10.1016/j.jsb.2014.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Trott O, Olson AJ. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–61. doi: 10.1002/jcc.21334 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0358222.r001

Decision Letter 0

Chandrabose Selvaraj

23 Jul 2026

Dear Dr. Maghembe,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 21 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors.

We look forward to receiving your revised manuscript.

Kind regards,

Chandrabose Selvaraj, Ph.D.

Academic Editor

PLOS One

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1.Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in your Competing Interests section:

“The author declares no competing interests.”

Please complete your Competing Interests on the online submission form to state any Competing Interests. If you have no competing interests, please state "The authors have declared that no competing interests exist.", as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now

This information should be included in your cover letter; we will change the online submission form on your behalf.

3. Thank you for uploading your study's underlying data set. Unfortunately, the repository you have noted in your Data Availability statement does not qualify as an acceptable data repository according to PLOS's standards.

At this time, please upload the minimal data set necessary to replicate your study's findings to a stable, public repository (such as figshare or Dryad) and provide us with the relevant URLs, DOIs, or accession numbers that may be used to access these data. For a list of recommended repositories and additional information on PLOS standards for data deposition, please see https://journals.plos.org/plosone/s/recommended-repositories.

4. Please upload a new copy of Figure 1, 2, and 3 as the detail is not clear. Please follow the link for more information: https://journals.plos.org/plosone/s/figures.

5. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

[Note: HTML markup is below. Please do not edit.]

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #2: Yes

**********

Reviewer #1: Revision for PTGS Manuscript

This PTGS1/PTGS2 manuscript presents a computational prioritization study that classifies human variants by population differentiation, coding consequence, splice prediction, and structural context, but the current text overstates functional and pharmacogenomic inference relative to the evidence shown. The core premise is reasonable because PTGS pharmacogenomics is biologically important and population structure alone is often insufficient for interpretation, yet the manuscript needs sharper methods reporting, tighter claims, and stronger validation standards before it is ready for PLOS ONE

Overall Assessment

The paper’s main contribution is a consequence-first ranking framework for PTGS1/PTGS2 variation, with three emphasized outputs: contextual high-FST markers, PTGS1 signal-peptide candidates, and a PTGS2 catalytic-channel candidate. That framing is plausible and fits the broader literature showing that PTGS genes are variable, pharmacologically important, and difficult to interpret mechanistically from allele frequency alone.

The manuscript is strongest when it treats variants as hypothesis-generating candidates rather than as established functional determinants. That stance is supported by the broader PTGS literature, where many reported associations depend on phenotype, drug, and population context, and only a small subset of variants have direct functional evidence.

The manuscript is weakest where it implies mechanistic confidence from in silico evidence alone. Across pharmacogenomics, population differentiation metrics can be dominated by common-variant frequency structure, while rare and functionally important variants are often missed or misranked by simple FST summaries. Across COX structural studies, docking is useful for contextualizing ligand binding, but it does not establish altered drug response for a human variant without experimental follow-up

Major Comments

The central claims need downtoning

“Support a model” and “biologically grounded prioritisation framework” are acceptable, but phrases implying functional separation of PTGS diversification into distinct mechanistic axes are too strong without cell-based or biochemical validation.

Population differentiation is not a surrogate for function.

The manuscript recognizes this, but the Results still privilege differentiated variants in a way that risks overinterpretation, especially because FST-based summaries are influenced by allele frequency structure and common variants

Signal-peptide interpretation is biologically plausible but still speculative.

Independent work shows that signal-peptide missense variants can disrupt ER targeting, translocation, cleavage, secretion, mRNA stability, or splicing, and sometimes act through multiple mechanisms, so the PTGS1 Trp8Arg/Pro17Leu hypotheses are credible, but those same studies show that some apparently concerning signal-peptide substitutions prove nonpathogenic experimentally, so this manuscript should avoid implying likely functional damage from physicochemical reasoning alone

The PTGS2 p.Val511Ala docking section is oversold.

COX-2 ligand recognition is structurally meaningful, and active-site differences are central to NSAID selectivity. But docking mainly supports possible compatibility within the catalytic channel, not altered inhibition, selectivity, or clinical pharmacogenomic relevance of Val511Ala.

The splice section needs stricter logic.

Low SpliceAI/Pangolin support is useful negative evidence, but absence of prediction signal is not proof of no splice effect, especially because signal-peptide and near-boundary variants can alter splicing modestly or alongside protein-level effects

Specific Recommendations

PTGS evidence already includes examples where broader sequencing plus replication identified clinically relevant PTGS1 variants, while other PTGS2 candidates showed weak or negligible relevance after deeper study

That is a useful benchmark, but this manuscript should more explicitly separate prioritization from evidence of effect.

The Methods need enough detail for reproduction.

There is need for exact data sources, population definitions, transcript selection rules, filtering thresholds, FST estimator, handling of multiallelic sites, principal isoform criteria, structural template choice, docking protocol, scoring functions, and any manual curation steps. Recent PGx population studies now integrate SNVs with structural variants and phenotype frameworks, underscoring how incomplete a variant landscape can be when methods are narrowly scoped or underreported

Reviewer #2: Major comments

1. Authors recommended providing information about the specific knowledge gap that this study addresses beyond previous investigations of PTGS1/PTGS2 genetic variation and pharmacogenomics?

2. Authors have mentioned that PTGS2 Val511Ala variant was specifically selected for structural modeling and docking analysis. Why are other coding variants not considered?

3. Authors were claiming that PTGS2 Val511Ala as residing within the catalytic channel. If authors provide additional information in the context of quantitative structural analyses like residue distances, binding-pocket interactions, or energetic changes, it would be more informative and supportive.

4. Authors have mentioned that PTGS1 Trp8Arg and Pro17Leu play a crucial role and affecting protein biogenesis. Is there any experimental or published evidence supporting this?

5. How can authors be claiming that PTGS2 p.Val511Ala as the primary structural candidate? Provide clear justification.

6. Authors have mentioned that protein topology is the primary criterion, providing information about what structural features were considered like catalytic residues, membrane-binding regions, dimer interfaces, solvent accessibility.

7. Authors recommended performing molecular dynamics simulation or binding free-energy calculation which may give better elusive structural effects.

8. Authors recommended increasing the overall figure quality.

**********

what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

PLoS One. 2026 Sep 25;21(9):e0358222. doi: 10.1371/journal.pone.0358222.r002

Author response to Decision Letter 1

5 Aug 2026

Dear Dr. Selvaraj and Reviewers,

Thank you for the constructive assessment of this manuscript. I have revised the study to make the evidence hierarchy explicit, reduce functional and pharmacogenomic overstatement, strengthen the quantitative structural record, clarify the rationale for selecting PTGS2 Val511Ala, define the limits of the PTGS1 leader-sequence analysis, expand methodological reporting, and improve the figure package. The final title—“Comparative evolutionary pharmacogenomics of human PTGS1 and PTGS2 identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions”—now names the two principal protein contexts highlighted by the revised analysis while retaining the evidence-calibrated phrase “protein-contextual candidates.”

The marked and clean manuscripts retain all figure legends but contain no embedded figure panels; Figures 1–5 are supplied separately. Supporting Tables S1–S7 have been consolidated into a single XLSX workbook with a Contents worksheet and dedicated worksheets preserving the complete S1–S7 data, consistent with the earlier technical-check instruction. A single-author language audit was completed. The only journal-mandated plural wording retained is the exact PLOS ONE competing-interests statement: “The authors have declared that no competing interests exist.”

Responses are provided point by point below. Section names, figure numbers and supporting-table identifiers are used as stable locations because pagination differs between the marked and clean files.

Summary of principal revisions

• Refined the title to name the PTGS1 N-terminal leader-sequence and PTGS2 catalytic-channel regions while preserving the evidence-calibrated phrase “protein-contextual candidates.”

• Reframed FST strictly as a measure of population differentiation and a descriptive prioritisation variable.

• Compared 31 missense candidates and clarified that Val511Ala was selected from four channel-context candidates because of population visibility, channel context and ligand-bound structural tractability—not because of a favorable docking outcome.

• Mapped canonical PTGS2 Val511 to 5KIR Val525 and added crystallographic distances, five paired docking seeds, direct-frame pose recovery, contact-set similarity, residue-525 occupancy and side-chain distance ranges.

• Replaced strong PTGS1 biogenesis claims with direct leader-sequence property calculations and explicit statements that targeting, cleavage, abundance, localization, activity and drug response were not demonstrated.

• Clarified that low SpliceAI/Pangolin scores reduce computational support but do not exclude context-dependent splice effects.

• Rebuilt Figures 1–3 for clarity and revised Figures 4–5 to display the evidence boundaries directly.

• Consolidated Supporting Tables S1–S7 into one XLSX workbook with a Contents worksheet; all source data and the original S6 and S7 worksheet structures were preserved.

• Updated Data and Code Availability statements, exact PLOS competing-interests wording, and single-author narrative language.

Academic Editor and journal requirements

Editor requirement 1 — PLOS ONE style and file organization

Comment: Please ensure that the manuscript meets PLOS ONE style requirements, including file naming.

Response: The revision package has been separated into a marked manuscript, a clean manuscript and this Response to Reviewers document. Figure panels are no longer embedded in the manuscript; their legends remain in the text and Figures 1–5 are supplied separately with standardized names. Supporting Tables S1–S7 have been consolidated into a single XLSX workbook, as requested during the earlier technical assessment. The title, declarations and submission filenames have also been checked for consistency.

Location in revised manuscript: Front matter; figure legends in Results; separate submission files.

Editor requirement 2 — Competing interests

Comment: Use the exact wording “The authors have declared that no competing interests exist.” when there are no competing interests.

Response: The Competing Interests statement has been replaced with the exact journal wording. Although this is a single-author manuscript, the plural form is retained because it is the wording specifically required by PLOS ONE.

Location in revised manuscript: Declarations and availability statements → Data Availability and Code Availability; combined Supporting Tables S1–S7 workbook.

Editor requirement 3 — Underlying data and repository status

Comment: The previously cited repository does not qualify as the sole acceptable data repository. Supply the minimal dataset in a stable public form.

Response: The Data Availability statement now identifies the article-associated Supporting Information as a single combined S1–S7 XLSX workbook containing the minimal derived dataset necessary to reproduce the reported findings. GitHub is no longer presented as the sole data repository; it is described as the public location for scripts, audit files and version history. No unverified DOI or accession has been inserted.

Location in revised manuscript: Declarations and availability statements → Data Availability and Code Availability; combined Supporting Tables S1–S7 workbook.

Editor requirement 4 — Figure clarity

Comment: Upload clearer copies of Figures 1, 2 and 3.

Response: Figures 1–3 were rebuilt and exported as separate publication files with enlarged typography and simplified evidence displays. Figures 4 and 5 were also revised to align with the calibrated interpretation. The manuscript now contains legends only, preventing reduced-quality embedded panels from being mistaken for submission figures.

Location in revised manuscript: Separate figure files; legends retained in the Results section.

Editor requirement 5 — Recommended citations

Comment: Evaluate any reviewer-recommended works for relevance; citation is not mandatory unless directed by the editor.

Response: The cited literature was reviewed for direct relevance. References were retained or added where they support PTGS biology, population differentiation, splice-prediction interpretation, signal-sequence biology and structural/docking methodology. No citation was added solely to satisfy a recommendation when it did not directly support the revised text.

Location in revised manuscript: Introduction, Discussion and Methods; References.

Reviewer 1

Reviewer 1, major comment 1 — Central claims require downtoning

Comment: The manuscript overstates functional and pharmacogenomic inference from in silico evidence, particularly when describing distinct mechanistic axes.

Response: I agree. The final title is “Comparative evolutionary pharmacogenomics of human PTGS1 and PTGS2 identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions.” It explicitly identifies the PTGS1 N-terminal leader-sequence and PTGS2 catalytic-channel regions while retaining “protein-contextual candidates” to avoid implying established molecular effects. Across the Abstract, Results, Discussion and Conclusion, mechanistic language was replaced with evidence-bounded terms such as candidate, prioritisation, sequence-property difference and structural context. Explicit statements now clarify that downstream function remains experimentally untested.

Location in revised manuscript: Title; Abstract; Results sections on consequence-first candidates, PTGS2 Val511Ala and PTGS1 leader-sequence variation; Discussion; Conclusion.

Reviewer 1, major comment 2 — Population differentiation is not function

Comment: FST-based summaries may privilege common-variant population structure and should not be treated as a surrogate for molecular function.

Response: I agree. FST is now described only as population differentiation and a descriptive prioritisation variable. The revised analysis separates differentiation from transcript consequence and protein context. A retrospective comparison of 31 missense candidates shows that Val511Ala was not the highest-FST missense variant overall, and the manuscript explicitly states that FST does not establish molecular function or selection.

Location in revised manuscript: Abstract; final Introduction paragraph; Results → consequence-first candidate classes; Methods → Population differentiation and conservation metrics and Consequence-first prioritisation; Discussion.

Reviewer 1, major comment 3 — PTGS1 signal-peptide interpretation remains speculative

Comment: Physicochemical reasoning alone does not establish altered targeting, cleavage, secretion, stability or pathogenicity.

Response: I agree. The revision no longer reports predicted biogenesis effects. Instead, it presents direct calculations for the 23-residue leader sequence and descriptive H-region: net-charge proxy, positive-residue count, hydrophobic-residue count, mean Kyte–Doolittle hydropathy, aromatic-residue count and proline count. The manuscript states that no signal-peptide predictor output was used and that effects on SRP recognition, ER targeting, translocation, cleavage, membrane insertion, maturation, abundance, localization, enzyme activity and drug response were not demonstrated. General signal-sequence biology is supported by references, but the exact variants remain experimental hypotheses.

Location in revised manuscript: Abstract; Results → PTGS1 Trp8Arg and Pro17Leu; Fig 5 legend; Methods → PTGS1 leader-sequence physicochemical analysis; Discussion; combined S1–S7 workbook (S7 worksheets).

Reviewer 1, major comment 4 — Val511Ala docking is oversold

Comment: Docking can provide structural context but cannot establish altered inhibition, selectivity, clinical response or pharmacogenomic effect.

Response: I agree. Canonical Val511 is mapped to 5KIR Val525, which is explicitly described as second-shell and channel-adjacent rather than a direct rofecoxib contact. Across five paired docking seeds, no systematic variant-associated shift was detected under the tested conditions. Docking scores are not presented as binding affinities, and the manuscript explicitly states that the analyses do not establish altered affinity, catalysis, pharmacology or clinical effects.

Location in revised manuscript: Abstract; Results → focused PTGS2 structural analysis; Fig 4 legend; Discussion; Methods → PTGS2 structural modeling, docking statistics and rofecoxib contact analysis; combined S1–S7 workbook (S6 worksheets).

Reviewer 1, major comment 5 — Splice logic requires caution

Comment: Low SpliceAI/Pangolin support is useful negative evidence but is not proof that a splice effect is absent.

Response: I agree. The seven evaluated alleles are now described using explicit score categories. PTGS2 c.724-10_724-7delATTT had a maximum SpliceAI delta score of 0.01 and maximum Pangolin score of 0.05. These values are interpreted as limited computational support, not proof of no effect; the variant remains a splice-region boundary candidate requiring RNA-level validation.

Location in revised manuscript: Results → SpliceAI/Pangolin analysis; Methods → Splice-prediction boundary analysis; Discussion; combined S1–S7 workbook (S3 worksheet).

Reviewer 1, specific recommendation — Separate prioritisation from evidence of effect

Comment: The manuscript should distinguish candidate prioritisation from established biological or clinical effects.

Response: This distinction is now a central organizing principle. Table 1 classifies contextual markers, leader-sequence candidates, channel-adjacent structural candidates, additional topology candidates and splice-region boundary candidates. The revised text repeatedly distinguishes measured quantities from untested downstream consequences.

Location in revised manuscript: Table 1 and accompanying text; Abstract; Discussion; Conclusion; Figs 3–5.

Reviewer 1, specific recommendation — Reproducible Methods

Comment: Provide substantially more detail on data sources, transcript rules, differentiation analysis, structural models, docking, thresholds and manual curation.

Response: The Methods were expanded to describe the population and superpopulation framework, allele-frequency and FST summaries, consequence-first classification, biological-identity reconciliation, web-based SpliceAI/Pangolin thresholds, manually curated topology categories, UniProt and structural references, ligand CIDs, Open Babel/RDKit preparation, AutoDock Vina version, paired Wilcoxon analyses, direct PTGS1 sequences and calculated properties. Detailed candidate, splice, structural and ligand-level records are supplied in the single combined Supporting Tables S1–S7 XLSX workbook and the public project repository.

Location in revised manuscript: Methods in full; combined Supporting Tables S1–S7 workbook; Code Availability.

Reviewer 2

Reviewer 2, comment 1 — Specific knowledge gap

Comment: Clarify the knowledge gap addressed beyond previous PTGS1/PTGS2 genetic-variation and pharmacogenomics studies.

Response: The Introduction now distinguishes the present study from association-focused or frequency-only analyses. The stated gap is the absence of an integrated framework that jointly evaluates population differentiation, transcript consequence, splice evidence and protein context while keeping differentiation separate from functional evidence.

Location in revised manuscript: Introduction, especially the final two paragraphs and study aim.

Reviewer 2, comment 2 — Why were other coding variants not modeled?

Comment: Explain why PTGS2 Val511Ala was selected and why other coding variants were not taken forward for structural analysis.

Response: A retrospective comparison of 31 missense candidates was added. Val511Ala was not the most differentiated missense variant overall. It was selected from four channel-context candidates because it ranked first within that subgroup by allele frequency, allele-frequency range and maximum pairwise FST, and because it mapped unambiguously to a human ligand-bound structural model. Selection was therefore based on population visibility, channel context and structural tractability rather than the docking result.

Location in revised manuscript: Abstract; Results → consequence-first candidate classes; Fig 4A; combined S1–S7 workbook (S1, S2 and S5 worksheets).

Reviewer 2, comment 3 — Quantitative structural context

Comment: Add residue distances, binding-pocket interactions or energetic changes to support the structural interpretation.

Response: The revised analysis reports that canonical Val511 maps to chain A Val525 in 5KIR. Minimum crystallographic distances from Val525 to rofecoxib were 6.004 Å for any heavy atom and 7.559 Å for the side chain. Across five paired seeds, the mean docking-score difference was −0.047600 kcal/mol and the direct-frame RMSD difference was −0.003538 Å. Contact-set Jaccard similarity was 0.952941, 0.972727 and 0.915269 at 4, 5 and 6 Å, respectively; direct 4-Å residue-525 occupancy was 0/5 in both groups, and side-chain distance ranges overlapped. These measurements support reproducibility and structural mapping, not altered affinity.

Location in revised manuscript: Results immediately before Fig 4 and Fig 4 legend; Methods → PTGS2 structural analyses; combined S1–S7 workbook (S6 worksheets).

Reviewer 2, comment 4 — Evidence for Trp8Arg and Pro17Leu biogenesis effects

Comment: State whether experimental or published evidence supports a crucial role for these exact variants in protein biogenesis.

Response: No direct experimental evidence was identified for these exact variants, and the revised manuscript states this explicitly. General literature supports the biological relevance of signal sequences, but the present study reports only direct sequence-property changes. The prior language implying a crucial biogenesis role was removed.

Location in revised manuscript: Results → PTGS1 leader-sequence physicochemical changes; Fig 5 legend; Discussion; Methods; combined S1–S7 workbook (S7 worksheets).

Reviewer 2, comment 5 — Why is Val511Ala the primary structural candidate?

Comment: Provide a clear justification for treating Val511Ala as the primary structural candidate.

Response: The wording “primary structural candidate” has been replaced by “channel-adjacent structural candidate.” The explicit selection rationale is the combination of population visibility, membership in a four-variant channel-context subgroup and availability of a ligand-bound structural mapping. No functional superiority is claimed.

Location in revised manuscript: Abstract; Results → consequence-first candidate classes; Fig 4A; Discussion.

Reviewer 2, comment 6 — Structural features used for topology classification

Comment: Describe the structural features considered, including catalytic residues, membrane-binding regions, interfaces and solvent accessibility.

Response: The topology framework now describes the N-terminal leader sequence, membrane-binding region, catalytic cyclooxygenase/peroxidase domain, catalytic residues and channel-adjacent context. For Val511Ala, canonical-to-structure mapping and distances to the crystallographic ligand/contact network were quantified. The classification is presented as protein context rather than a functional score. Dimer-interface or solvent-accessibility claims are not made where they were not directly quantified.

Location in revised manuscript: Methods → Functional-domain and topology mapping and PTGS2 structural modeling; Table 1; Figs 1, 3 and 4.

Reviewer 2, comment 7 — Molecular dynamics or binding free energy

Comment: Consider molecular dynamics simulation or binding free-energy calculations to reveal subtle structural effects.

Response: I appreciate this recommendation. MD and binding free-energy calculations were not added. Instead, the revision strengthens the directly supported static evidence through five paired docking seeds, crystallographic distance measurements, direct-frame pose recovery, contact-set similarity, residue-specific occupancy and side-chain distance ranges. The revised manuscript reports the absence of a systematic shift under these conditions and explicitly acknowledges that these analyses do not establish altered affinity or pharmacology. MD/free-energy analysis remains an appropriate future extension rather than a basis for claims in the present study.

Location in revised manuscript: Abstract; Results and Fig 4; Discussion; Methods → PTGS2 structural analyses; combined S1–S7 workbook (S6 worksheets).

Reviewer 2, comment 8 — Figure quality

Comment: Increase the overall quality and clarity of the figures.

Response: Figures 1–3 were rebuilt with clearer panel organization, enlarged labels and calibrated terminology. Figures 4 and 5 were redesigned to show the quantitative evidence and validation boundaries directly. All figures are supplied as separate files, while the manuscript retains the complete legends only.

Location in revised manuscript: Separate Fig1–Fig5 files; legends in Results.

Closing statement

I thank the Academic Editor and both reviewers for comments that materially improved the rigor, reproducibility and interpretive calibration of the manuscript. The revised study now presents population differentiation as context, reports the directly measured structural and sequence-property quantities, and reserves biological and pharmacogenomic effects for future experimental testing.

Sincerely,

Reuben S. Maghembe

Corresponding author

Attachment

Submitted filename: PONE-D-26-28410_Response_to_Reviewers_TITLE_UPDATED.docx

pone.0358222.s003.docx^{(34.9KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0358222.r003

Decision Letter 1

Chandrabose Selvaraj

30 Aug 2026

Comparative evolutionary pharmacogenomics of human prostaglandin-endoperoxide synthase paralogs identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions

PONE-D-26-28410R1

Dear Dr. Maghembe,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Chandrabose Selvaraj, Ph.D.

Academic Editor

PLOS One

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

**********

Reviewer #1: Accept the manuscript. Right now, I don't have concerns about dual publication, research ethics, or publication ethics.

**********

what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #1: No

**********

PLoS One. doi: 10.1371/journal.pone.0358222.r004

Acceptance letter

Chandrabose Selvaraj

PONE-D-26-28410R1

PLOS One

Dear Dr. Maghembe,

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS One. Congratulations! Your manuscript is now being handed over to our production team.

At this stage, our production department will prepare your paper for publication. This includes ensuring the following:

* All references, tables, and figures are properly cited

* All relevant supporting information is included in the manuscript submission,

* There are no issues that prevent the paper from being properly typeset

You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps.

Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing.

If we can help with anything else, please email us at customercare@plos.org.

Thank you for submitting your work to PLOS One and supporting open access.

Kind regards,

PLOS One Editorial Office Staff

on behalf of

Dr. Chandrabose Selvaraj

Academic Editor

PLOS One

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 File. Combined Supporting Tables S1–S7.

(XLSX)

pone.0358222.s001.xlsx^{(100.1KB, xlsx)}

Attachment

Submitted filename: PONE-D-26-28410_Response_to_Reviewers_TITLE_UPDATED.docx

pone.0358222.s003.docx^{(34.9KB, docx)}

Data Availability Statement

[pone.0358222.ref001] 1.Nagao M, Sato Y, Yamauchi A. A meta-analysis of PTGS1 and PTGS2 polymorphisms and NSAID intake on the risk of developing cancer. PLoS One. 2013;8(8):e71126. doi: 10.1371/journal.pone.0071126 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref002] 2.Smith WL, Garavito RM, DeWitt DL. Prostaglandin endoperoxide H synthases (cyclooxygenases)-1 and -2. J Biol Chem. 1995;270:14843–7. doi: 10.1074/jbc.270.24.14843 [DOI] [PubMed] [Google Scholar]

[pone.0358222.ref003] 3.Blobaum AL, Marnett LJ. Structural and functional basis of cyclooxygenase inhibition. J Med Chem. 2007;50(7):1425–41. doi: 10.1021/jm0613166 [DOI] [PubMed] [Google Scholar]

[pone.0358222.ref004] 4.Cebrián-Prats A, González-Lafont À, Lluch JM. Unraveling the molecular details of the complete mechanism that governs the synthesis of prostaglandin G2 catalyzed by cyclooxygenase-2. ACS Omega. 2019;4:2063–74. doi: 10.1021/acsomega.8b03575 [DOI] [Google Scholar]

[pone.0358222.ref005] 5.Trinh HKT, Pham LD, Le KM, Park HS. Pharmacogenomics of hypersensitivity to non-steroidal anti-inflammatory drugs. Front Genet. 2021;12. doi: 10.3389/fgene.2021.647257 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref006] 6.Sheehan BJ, Edwards B, Medrano IS, El-Saidi MA, Zaidan WR, El-Ezzi AA, et al. Association between two single nucleotide polymorphisms of the Prostaglandin-Endoperoxide Synthase 1 and 2 genes and cell proliferative prostatic diseases in Lebanon. Oncotarget. 2025;16:262–72. doi: 10.18632/oncotarget.28710 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref007] 7.Lucena MI, García-Martín E, Daly AK, Blanca M, Andrade RJ, Agúndez JAG. Next-generation sequencing of PTGS genes reveals an increased frequency of non-synonymous variants among patients with NSAID-induced liver injury. Front Genet. 2019;10. doi: 10.3389/fgene.2019.00134 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref008] 8.Wosniaki DK, Kusma BN, Marin AM, Munhoz EC, Farias JS de H, Mendes EB, et al. Divergent gene expression of PTGS1 and PTGS2 along the disease course of chronic myeloid leukaemia. Front Oncol. 2025;15:1556956. doi: 10.3389/fonc.2025.1556956 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref009] 9.Li X, Mazaleuskaya LL, Yuan C, Ballantyne LL, Meng H, Smith WL, et al. Flipping the cyclooxygenase (Ptgs) genes reveals isoform-specific compensatory functions. J Lipid Res. 2018;59(1):89–101. doi: 10.1194/jlr.M079996 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref010] 10.Thorn CF, Grosser T, Klein TE, Altman RB. PharmGKB summary: Very important pharmacogene information for PTGS2. Pharmacogenet Genom. 2011;21(9):607–13. doi: 10.1097/FPC.0b013e3283415515 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref011] 11.Zhang C, Jiang X, Chen W, Li Q, Yun F, Yang X, et al. Population genetic difference of pharmacogenomic VIP gene variants in the Lisu population from Yunnan Province. Medicine (Baltimore). 2018;97(52):e13674. doi: 10.1097/MD.0000000000013674 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref012] 12.Yucesoy B, Kashon ML, Johnson VJ, Lummus ZL, Fluharty K, Gautrin D, et al. Genetic variants in TNFα, TGFB1, PTGS1 and PTGS2 genes are associated with diisocyanate-induced asthma. J Immunotoxicol. 2016;13(1):119–26. doi: 10.3109/1547691X.2015.1017061 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref013] 13.Bhatia G, Patterson N, Sankararaman S, Price AL. Estimating and interpreting FST: The impact of rare variants. Genome Res. 2013;23(9):1514–21. doi: 10.1101/gr.154831.113 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref014] 14.Holsinger KE, Weir BS. Genetics in geographically structured populations: Defining, estimating and interpreting F(ST). Nat Rev Genet. 2009;10(9):639–50. doi: 10.1038/nrg2611 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref015] 15.1000 Genomes Project Consortium, Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. doi: 10.1038/nature15393 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref016] 16.Ota M, Spence JP, Zeng T, Dann E, Milind N, Marson A, et al. Causal modelling of gene effects from regulators to programs to traits. Nature. 2026;650(8101):399–408. doi: 10.1038/s41586-025-09866-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref017] 17.Cingolani P, Platts A, Wang LL, Coon M, Nguyen T, Wang L, et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff. Fly. 2012;6(2):80–92. doi: 10.4161/fly.19695 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref018] 18.de Sainte Agathe J-M, Filser M, Isidor B, Besnard T, Gueguen P, Perrin A, et al. SpliceAI-visual: A free online tool to improve SpliceAI splicing variant interpretation. Hum Genom. 2023;17(1):7. doi: 10.1186/s40246-023-00451-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref019] 19.Zeng T, Li YI. Predicting RNA splicing from DNA sequence using Pangolin. Genome Biol. 2022;23(1):103. doi: 10.1186/s13059-022-02664-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref020] 20.Orlando BJ, Lucido MJ, Malkowski MG. The structure of ibuprofen bound to cyclooxygenase-2. J Struct Biol. 2015;189(1):62–6. doi: 10.1016/j.jsb.2014.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref021] 21.Zhang L, Siyiti M, Zhang J, Yao M, Zhao F. Anti-inflammatory and anti-rheumatic activities in vitro of alkaloids separated from Aconitum soongoricum Stapf. Exp Ther Med 2021. doi: 10.3892/etm.2021.9924 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref022] 22.Goel A, Boland CR, Chauhan DP. Specific inhibition of cyclooxygenase-2 (COX-2) expression by dietary curcumin in HT-29 human colon cancer cells. Cancer Lett. 2001;172(2):111–8. doi: 10.1016/s0304-3835(01)00655-3 [DOI] [PubMed] [Google Scholar]

[pone.0358222.ref023] 23.Cingolani P. Variant annotation and functional prediction: SnpEff. In: Ng C, Piscuoglio S, editors. Variant calling: Methods and protocols. New York (NY): Springer US; 2022. p. 289–314. doi: 10.1007/978-1-0716-2293-3_19 [DOI] [PubMed] [Google Scholar]

[pone.0358222.ref024] 24.Hegde RS, Kang SW. The molecular code of the signal peptide and its recognition by the signal recognition particle. Genes Dev. 2006;20:2377–89. doi: 10.1101/gad.1493006 [DOI] [Google Scholar]

[pone.0358222.ref025] 25.Kellogg MK, Tikhonova EB, Karamyshev AL. Signal recognition particle in human diseases. Front Genet. 2022;13. doi: 10.3389/fgene.2022.898083 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref026] 26.Jain P, Satija J, Sudandiradoss C. Discovery of andrographolide hit analog as a potent cyclooxygenase-2 inhibitor through consensus MD-simulation, electrostatic potential energy simulation and ligand efficiency metrics. Sci Rep. 2023;13(1):8147. doi: 10.1038/s41598-023-35192-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref027] 27.Orlando BJ, Lucido MJ, Malkowski MG. The structure of ibuprofen bound to cyclooxygenase-2. J Struct Biol. 2015;189(1):62–6. doi: 10.1016/j.jsb.2014.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0358222.ref028] 28.Trott O, Olson AJ. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–61. doi: 10.1002/jcc.21334 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Comparative evolutionary pharmacogenomics of human prostaglandin-endoperoxide synthase paralogs identifies population-structured coding variation and protein-contextual candidates in N-terminal leader-sequence and catalytic-channel regions

Reuben S Maghembe

Roles

Abstract

Introduction

Results

Global PTGS variation reveals paralog-specific population-genetic architecture

Fig 1. Global population-genetic architecture of PTGS variation.

Table 1. Functional PTGS candidate classes defined by consequence-first prioritisation.

Consequence-first candidate classes link population differentiation to protein topology and splice-prediction support

SpliceAI/Pangolin analysis defines splice-region boundary candidates

PTGS2 Val511Ala maps to a catalytic-channel environment sampled by NSAIDs and natural anti-inflammatory products

Fig 2. PTGS2 ligand perturbation and catalytic-channel pharmacogenomic remodeling.

Evolutionary and protein-context integration distinguishes PTGS1 leader-sequence and PTGS2 channel-adjacent candidate contexts

Fig 3. Evolutionary and protein-context integration of PTGS paralog variation.

Fig 4. Population prioritisation, structural mapping and controlled docking assessment of PTGS2 Val511Ala.

PTGS1 Trp8Arg and Pro17Leu produce distinct leader-sequence physicochemical changes

Fig 5. PTGS1 Trp8Arg and Pro17Leu produce distinct leader-sequence physicochemical changes.

Discussion

Conclusion

Methods

Study design and analytical overview

Variant datasets and population-level genotype matrices

Population differentiation and conservation metrics

Transcript-aware and coding consequence annotation

Consequence-first prioritisation and FST-first comparison

Splice-prediction boundary analysis

Comparator-gene generalisation resource audit

Functional-domain and topology mapping

PTGS2 structural modeling and docking analyses

PTGS2 docking statistics

PTGS2 rofecoxib structural contact analysis

PTGS1 leader-sequence physicochemical analysis

Statistical summary integration

Ethics statement

Figure generation and reproducibility

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Chandrabose Selvaraj

Roles

Author response to Decision Letter 1

Decision Letter 1

Chandrabose Selvaraj

Roles

Acceptance letter

Chandrabose Selvaraj

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases