Serum Proteomics of Multiple Menstrual Symptoms in Female Athletes: A Pilot Study.

Conceptualization: N.I., M.Y., M.S.‐S., N.H. Methodology and investigation: S.Y.W., N.H. Data curation: K.F., T.K. Formal analysis: K.F., Writing – original draft: K.F. Writing – review and editing: S.Y.W., N.I., M.Y., M.S.‐S., N.H. Supervision and funding acquisition: S.Y.W., M.S.‐S., N.H.

This work was supported by the Japan Sports Agency (Grant Number 999999).

This observation study recruited a total of 54 female students from a softball club at Nippon Sport Science University. Sample collection was conducted in 2019 (November 5, 12, 19, 27, and December 3) and 2020 (November 19, 26, and December 3, 10, 17). The volunteers had the following clinical characteristics: 19.6 ± 1.0 year, 160.9 ± 4.6 cm, 60.4 ± 5.0 kg, BMI 23.5 ± 1.9 m 2 /kg. In addition, all participants reported regular training habits consisting of six training days per week. Training duration was approximately 4 h/day on weekdays and 7 h/day on weekends. Five blood samples were collected from each participant once in a week over 5 weeks, with an interval of a 6–8 days. Serum was obtained from whole blood by allowing to stand at room temperature for 20–30 min, followed by centrifugation at 3000 rpm for 10 min to isolate the serum components. The obtained serum from each participant was aliquoted into five tubes, each containing 100 μL each. After sample collection, serum samples were transported to the laboratory within 30 min on ice and stored at −80°C until further experiments. In this study, both proteomic and hormone analyses were performed. Two ovarian hormones, estradiol (E2) and progesterone (P4) in the blood samples were analyzed at LSI Medience Corporation, a contact laboratory for clinical tests (Itabashi, Tokyo, Japan). This was an observation study in which no intervention or treatment was administered to participants. This study was approved by the ethical committees Tokyo Institute of Technology (2019094) and Nippon Sports Science University (019‐H090). All volunteers were provided with a verbal explanation of the study and subsequently gave their written informed consent prior to participation. The study was carried out in full compliance with the relevant guidelines and regulations. Menstrual distress questionnaire (MDQ) established by Moos is often used for evaluation of menstrual symptoms [ 28 ]. In this study, the MDQ was translated into Japanese and participants were asked to respond to 46 questions on a five‐point scale ranging from 0 to 4, with 0 indicating no symptoms and 4 indicating very severe symptoms (Appendix  S2 ). The scores for question No. 36 to No. 40 were reversed due to their positive phrasing. To account for individual differences in symptom perception, scores of 1 to 4 were classified into the symptom group, while a score of 0 was classified into the non‐symptom group. In this study, several exclusion criteria were set. The first criterion was attendance during sampling; among 54 participants, only 46 participants completed the blood collection for all 5 weeks, and the remaining 8 were excluded from the analysis. The second criterion was irregular menstruation; three participants were excluded due to reported abnormal menstruation. The third criterion was medical history; one participant reported a diagnosis of Grave's disease. Highly abundant proteins in serum were removed using affinity columns since they can interfere with the analysis by masking low‐abundance protein in serum. Albumin and IgG were selectively depleted from the serum using the Aurum serum protein mini kit (Bio‐Rad, Hercules, CA, USA). Furthermore, Multiple Affinity Removal Spin Cartridge Human 14 (Agilent, Santa Clara, CA, USA) was used to remove the following 14 highly abundant proteins from the sample: albumin, immunoglobulin G (IgG), immunoglobulin A (IgA), transferrin, haptoglobin, antitrypsin, fibrinogen, alpha‐2‐macroglobulin, alpha‐1‐acid glycoprotein, apolipoprotein A‐I, apolipoprotein A‐II, complement C3, transthyretin, and immunoglobulin M (IgM). The depleted fraction was then concentrated to a final volume of 100 μL using a Vivaspin 2 MWCO 3000 (Cytiva, Tokyo, Japan). Concentrated samples were further purified to remove contaminants using a 2D clean up kit (Cytiva, Tokyo, Japan), and protein concentration was quantified with the 2D quant kit (Cytiva, Tokyo, Japan). All procedures were performed according to the manufacturers' protocols. 2‐DE experiments were conducted according to the method described in a previous study [ 29 ]. For 2‐DE, 15 μg of protein sample dissolved in Destreak Rehydration Solution (Cytiva, Tokyo, Japan) was used. The protein sample was loaded onto an Immobiline Drystrip (7 cm, pH 3–10, Cytiva, Tokyo, Japan) at room temperature. Isoelectric focusing was performed under the following conditions: (i) 0–300 V for 30 min; (ii) 300–1000 V for 60 min; (iii) 1000–5000 V for 90 min; (iv) 5000 V for 36 min. The gel strips containing proteins separated by isoelectric points were equilibrated in equilibration buffer [6 M urea; 1.5 M Tris‐Hydrochloric Acid (Tris‐HCl), pH 8.8; 30% v/v glycerol; 2% sodium dodecyl sulfate (SDS)]. For the first 15 min, the strips were reduced in equilibration buffer containing 1% (w/v) dithiothreitol (DTT). For the next 15 min, they were alkylated in equilibration buffer containing 4.5% iodoacetamide (IAA). The secondary‐dimension electrophoresis was carried out using NuPAGE 4%–12% Bis‐Tris ZOOM GEL (Thermo Fisher Scientific, Waltham, MA, USA). Gels were set in XCell‐SureLock (Thermo Fisher Scientific, Waltham, MA, USA) and run at 200 V, 2 mA for 40 min. SYPRO RUBY stain (Invitrogen, Waltham, MA, USA) was used for staining, and the stained gels were scanned using Typhoon FLA 9500 (Cytiva, Tokyo, Japan). The resulting 2‐DE images were analyzed using a gel analysis software, Melanie (Cytiva, Tokyo, Japan), and spot detection was performed on all gel images. The %vol, representing the percentage contribution of each protein spot to total protein volume, was obtained and used for subsequent statistical analysis. Hormone variation among all participants was assessed before analyzing protein changes associated with menstrual symptoms, as individual differences in hormone fluctuations during the menstrual cycle could affect the results. The difference in two ovarian hormones is shown, and this study focused on a limited sample for statistical analysis. The criteria are shown in Figure  1A . Participants were selected based on having continuous menstrual phases of luteal, menstruation, and follicular. A total of 14 participants and 42 2DE images were used for statistical analysis. Data mining process. (A) The analysis focused on participants (highlighted in yellow) whose menstrual phases changed consecutively over 3 weeks—from the luteal phase to menstrual phase to follicular phase (indicated by black square). As a result, fourteen volunteers and 42 samples were included in the statistical analysis. (B) After sample selection, interacted proteins between menstruation and menstrual symptoms were explored using multiple regression. Subsequently, the Mann–Whitney U test was performed in order to confirm significant protein changes associated with symptoms in each specific menstrual phase. All statistical analyses were conducted using R, version 4.4.1. Figure  1B illustrates the data mining approach in this study. The %vol values were log‐transformed to normalize the data, ensuring equal treatment of both large and small spots. Multiple regression analysis was applied to selected samples as the statistical model to identify proteins associated with both menstrual symptoms and menstruation. The formula is as follows; (1) log 2 % vol = β 0 + β 1 menstruation 1 + β 2 symptom 2 + β 3 menstruation 1 × symptom 2 β 0 ~ β 3 represents regression coefficient. In this study, proteins influenced by the interaction between menstruation and symptoms ( β 3 menstruation 1 × symptom 2 ) were of primary interest. Here, symptom refers to the individual items from the MDQ questionnaire, and all 46 questions were analyzed independently. The significant interactions were considered as a local false discovery rate (lFDR) < 0.1, to account for multiple testing. Next, Mann–Whitney U tests were performed during both the menstrual and non‐menstrual phases for proteins that showed significant interactions. This test aimed to identify proteins that differed significantly between symptom and non‐symptom in each phase. When multiple testing correction was required, p ‐values from Mann–Whitney U test were adjusted by the Benjamini‐Hochberg procedure, and q ‐values < 0.1 was considered statistically significant (i.e., when each MDQ question was associated with multiple significant interaction proteins). In cases where only one significant interaction protein was identified for each MDQ question, a p ‐value < 0.05 was considered statistically significant. However, the significance threshold may be too stringent to capture these subtle changes because this study recruited healthy participants with regular exercise habits and no underlying diseases, which this detectable protein changes in this study are likely subtle compared to disease‐related studies. Significant spots identified in the statistical analysis were automatically excised from the gels by the Ettan Spot Picker (Cytiva, Tokyo, Japan) for subsequent protein identification by mass spectrometry. The excised gel plugs were reduced and alkylated with 100 mM DTT and 50 mM IAA, respectively. Protein‐containing gel plugs were incubated with trypsin (+) solution [20 μg/μL trypsin (Trypsin Gold, Mass Spectrometry Grade, Promega, Madison, WI, USA), 40 mM ammonium bicarbonate, 0.2 mM HCl, 5 mM calcium chloride (CaCl2), and 10% acetonitrile (ACN)] for 5 min at room temperature, and then incubated at 37°C overnight with trypsin (−) solution, which is consisted of the same components as the trypsin (+) solution except for the absence of trypsin. For peptide extraction, supernatants were collected after incubating the digested gels with the following solvents, in order, for 10 min at room temperature: ultra‐pure water, 60% ACN, 80% ACN, 100% ACN. The collected peptide solutions were concentrated to approximately 2 μL using a Smart Evaporator C10 (BioChromato, Kanagawa, Japan). Prior to mass spectrometry, the peptide solution was desalted using C‐tip (AMR, Tokyo, Japan) according to the manufacturer's protocol. Protein contained in significant protein spots was identified by liquid chromatography–tandem mass spectrometry (LC–MS/MS). Desalted peptide samples were dissolved in TFA‐A (0.1 trifluoroacetic acid (TFA), 2% ACN in ultra‐pure water). The peptide solution was injected into LC (Vanquish Neo, Thermo Fisher Scientific, Waltham, MA, USA) equipped with an AURORA Elite column (75 μm × 15 mm, packed with 1.7 μm particles). Peptides were separated and eluted using two mobile phases: solvent A (ultrapure water including 1% formic acid) and solvent B (80% acetonitrile including 1% formic acid). The gradient condition was set to change from 5% to 57% B buffer over 20 min with a 0.3 μL flow rate. Peptides were ionized using the electrospray ionization (ESI) technique and analyzed by a mass spectrometer (Orbitrap Exploris 480, Thermo Fisher Scientific, Waltham, MA, USA). Data were analyzed using NIMS Proteome Discoverer 3.1 software (Thermo Fisher Scinetific, Waltham, MA, USA). Protein identification was performed using MASCOT (Version 2024_3, Matrix Science, United Kingdom) and the Swiss‐Prot database. The following parameters were applied for the mass spectrometry analysis: (1) precursor ion mass range: 350–5000 Da, (2) mass tolerance: 10.0 PPM, (3) fixed modification: (4) carbamidomethylation of Cys residues, (5) variable modification: (6) oxidation of Met residues. Proteins were filtered using the following exclusion criteria: (1) FDR > 0.01, (2) common contaminants, such as keratin, potentially introduced during sample preparation. The mass spectrometry data have been deposited in the ProteomeXchange via the PRIDE database (Accession: PXD065313, DOI: 10.6019/PXD065313 ). Western blot was performed to validate the potential proteins associated with menstrual symptoms. In this study, transferrin was selected for validation because it showed significant association with two MDQ questions. Band intensities were compared between symptom and non‐symptom groups for MDQ20 and MDQ32. Four samples were randomly selected from the non‐symptom group, while four samples with the highest scores were selected for the symptom group. Serum protein concentrations were measured, and a total of 0.04 μg of protein per sample was used in the experiments. Precision Plus Protein WesternC Standard (Bio‐Rad, Hercules, CA, USA) was used as the molecular weight marker. Samples and protein standards were loaded onto 12.5% SDS‐PAGE gels and separated by electrophoresis. Proteins were then transferred to a PVDF membrane at 15 V and 0.16A for an hour at room temperature. To detect transferrin, a primary antibody against transferrin (Proteintech, Cat#: 17435‐AP, RRID:AB_2035023) was used. The antibody was diluted 1:5000 in blocking buffer consisting of 5% skim milk/TBS‐T (0.1% Tween 20), following the manufacturer's recommendation. The membrane was incubated overnight at 4°C. Subsequently, HRP Goat anti‐rabbit IgG (abcam, Cat#: ab6721, RRID:AB_955447) was used as the secondary antibody. The dilution rate is 1:10000 in the same blocking buffer. After an hour's incubation at room temperature, protein bands were visualized using the Merck Immobilon Western Chemiluminescent HRP substrate. Band intensities were quantified from the scanned images using Image J.

To assess hormone imbalance among the participants, the variance of two ovarian hormones was analyzed. Figure  2A,B show the variance of estradiol and progesterone over 5 weeks, respectively. Color plots show each hormone's level at five sampling weeks on both figures. The estradiol levels among the 14 participants varied considerably over the five‐week period, with individual fluctuations spanning a wide range. For example, the average range among 14 participants was 166–22 (ng/mL). ID3 exhibited the widest range from 31 to 415 (ng/mL), while ID49 exhibited the narrowest range from 24 to 38 (ng/mL) (Figure  2A ). Similarly, for progesterone, the average range among 14 participants was 0.11 to 6.14. ID3 exhibited the widest range from 0.10 to 26.3 (pg/mL), while the progesterone levels of ID49 and ID51 did not fluctuate (Figure  2B ). These variations highlight the significant individual differences in hormone levels, which could influence the analysis of menstrual symptoms‐related protein changes due to their various functions. Ovarian hormone fluctuations among the participants. (A) Variability in estradiol level. The vertical axis represents individual participants, and the horizontal axis indicates estradiol concentrations. Colored points indicate sampling time points over the five‐week period. (B) Variability in progesterone level. The vertical axis represents individual participants, and the horizontal axis indicates progesterone concentrations. Colored points indicate sampling time points over the five‐week period. (C) A table showing menstrual cycle phases across the sampling weeks. Rows represent participants, and columns correspond to the sampling weeks. Each cell is colored according to the assigned phase based on estradiol and progesterone levels, following the legend. The follicular phase week that includes menstruation is defined as the menstrual phase. From 2DE images, a total of 648 spots were detected using the software, Melanie. The obtained 2DE images are shown in Figure  S2 %vol were obtained from each image and multiple regression analysis was conducted. This study focused on protein changes resulting from the interaction between menstrual symptoms and menstruation. The result is shown in Table  1 . A total of 19 MDQ items showed significant interaction proteins, with 29 proteins identified as having significant interaction. Significant interaction protein spots and corresponding MDQ questions identified by multiple regression. Note: MDQ No. refers to the question number, and its content corresponds to the items in the MDQ questionnaire. Significant spots are protein spots that showed a statistically significant interaction between menstrual phase and symptom scores related to the respective MDQ question. The significance threshold was set at lFDR < 0.1. Significant interaction proteins were analyzed by the Mann–Whitney U test. This analysis was conducted to detect significant protein changes associated with symptoms in either the non‐menstruation or menstruation. The results can be categorized into four patterns: (1) significance only in the non‐menstruation phase, (2) significance only in the menstruation phase, (3) significance in both phases, and (4) no significance in either phase. In this study, patterns 1 and 2 were considered important, as they may indicate that menstruation either triggers the symptoms or that the symptoms disappear with the onset of menstruation. As a result, a total of eight proteins showed significant changes across six MDQ items (Table  2 and Figure  3 ). One of the six items showed significant proteins during the non‐menstrual phase group, while the remaining five were associated with significant proteins during the menstrual phase. Among the identified proteins, spot 281 was common to more than two questions, while the other significant protein spots were unique to individual questions. In this analysis, two significant thresholds were used. When multiple comparisons were performed, a q ‐value < 0.1 was applied. In the case of single comparisons, a p ‐value < 0.05 was considered significant. Significant protein spots and corresponding MDQ questions identified by Mann–Whitney U tests. Note: MDQ No. refers to the question number, and its content corresponds to the items in the MDQ questionnaire. Significant spots are protein spots that showed a statistically significant change depending on symptom presence in either non‐menstrual or menstrual phase. The significance threshold was set at q  < 0.1 when adjusting for multiple tests; otherwise, a threshold of p  < 0.05 was applied. Phase indicates the menstrual phase in which the significant change in protein expression was observed. Changes in protein spots intensity by menstruation and symptom presence. Each boxplot shows the %vol (vertical axis) for protein spots across menstrual and non‐menstrual phases, with and without symptom (horizontal axis). (A) MDQ1 (muscle stiffness): Spot592 (B) MDQ6 (general aches and pains): Spot599 (C) MDQ20 (tension): Spot281 (D) MDQ20 (tension): Spot580 (E) MDQ20 (tension): Spot607 (F) MDQ21 (depression): Spot288 (G) MDQ28 (distractible): Spot636 (H) MDQ32 (take naps; stay in bed): Spot281 (I) MDQ32 (take naps; stay in bed): Spot285. These eight significant protein spots were identified by LC–MS/MS (Figure  4 ). Identification results were shown in Table  S4 . Identification results indicate each significant protein spot has several candidates. In this study, for each protein spot, the protein candidates with the highest Mascot score were prioritized. However, in the case where the highest Mascot score protein is actin, the protein with the second highest score is selected. This is because the present study focuses on serum proteins rather than cell structure proteins. Location of the eight significant spots on the 2DE image. The vertical axis of the 2DE image represents molecular weight (MW), while the horizontal axis represents isoelectric point (pI). Each spot ID was assigned by Melanie software. Western blotting was conducted for transferrin (spot 281) as a validation experiment for MDQ20 (tension) and MDQ32 (take naps; stay in bed) (Figure  5 ). For each MDQ item, a total of eight subjects were selected for this experiment (four per group). A single band appeared at around 75 kDa corresponding to the theoretical molecular weight of transferrin, that is approximately 77 kDa. As a result of the quantification of these bands, all quantified values were visualized in boxplot (Figure  5C,D ). 2DE results showed that the %vol of spots 281 was higher in participants from the non‐symptom group compared to the symptom group. Consistent with the result of 2DE, Western blot analysis also demonstrated higher transferrin intensity in the non‐symptom group than in the symptom group for both MDQ20 and MDQ32. These results indicate that the Western blot findings corroborate the 2DE results. Western blot results for protein validation. (A) Cropped western blot (WB) image showing detection of transferrin protein for MDQ20 at approximately 75 kDa, comparing symptom and non‐symptom groups. (B) Cropped western blot (WB) image showing detection of transferrin protein for MDQ32 at approximately 75 kDa, comparing symptom and non‐symptom groups. Each lane was loaded with a serum sample obtained from a different individual. “ID” represents the individual number, and “W” indicates the week of collection (e.g., ID9W3 refers to a sample collected from individual 9 at week 3). In the non‐symptom group, four samples were randomly selected, while in the symptom group, the four samples with the highest scores were chosen. The values below each band indicate the band intensity quantified using Image J. A total of 0.04 μg of serum protein was loaded for detection of Tf bands. Proteins were separated by SDS‐PAGE using a 12.5% gel. Precision Plus Protein WesternC Standard (Bio‐Rad) was used as the molecular weight marker. For protein detection, a 1:5000 diluted anti‐transferrin antibody was used as the primary antibody, and HRP Goat Anti‐Rabbit IgG (1:10000 dilution) was used as the secondary antibody. Band intensities were quantified using Image J, and the results were visualized as boxplots for comparison between symptom and non‐symptom groups. (C) Boxplot showing transferrin band intensities for the symptom and non‐symptom groups in MDQ20. (D) Boxplot showing transferrin band intensities for the symptom and non‐symptom groups in MDQ32. Each boxplot represents band intensity values from four independent samples within each group (symptom and non‐symptom).

Hormone fluctuation is considered to play an important role in women's well‐being, with much previous studies focusing on the role of ovarian hormones such as estradiol and progesterone. In contrast, the present study aimed to explore the relationship between menstrual symptoms, assessed by a self‐reported questionnaire, and the potential role of proteins in women's physiological and psychological changes during the menstrual cycle. To achieve this, two‐dimensional gel electrophoresis (2DE) was used for proteome analysis. The participants in this study were all members of a softball club and lived in the same dormitory. Although residual external variability may persist, their shared lifestyle and training environment likely reduced it to a minimum. Based on the statistical analysis of 14 participants, six questionnaire items were associated with significant changes in proteins. Transferrin (Tf) showed significant associations with MDQ20 (tension) and MDQ32 (take naps; stay in bed). Tf is a glycoprotein of approximately 77 kDa that functions as an iron transporter [ 30 ]. It is known that Tf expression level is related to serum iron level; in the case of iron deficiency, Tf synthesis increases [ 31 , 32 ]. Previous studies have shown Tf decreases during menstruation [ 33 , 34 ]. In the present study, 2DE analysis revealed lower Tf level in the symptom group compared to the non‐symptom group, and Western blot (WB) results supported the 2DE findings (Figure  5 ). Although a direct relationship between Tf and menstrual symptoms has not been reported, iron—its primary cargo of Tf—has been implicated in emotional changes and sleep disturbance [ 35 , 36 , 37 ]. Several possible explanations may account for our findings. First, Tf is classified as an acute protein, and its levels are known to decrease in response to inflammation [ 30 ]. Menstruation involves the shedding of the endometrial lining, which induces inflammation [ 38 ]. This inflammatory response may lead to decreased Tf level, thereby reducing iron transport. Second, iron loss due to menstruation may also contribute to the changes in Tf level. Women lose serum iron during menstruation, and in response to decreased iron levels, Tf synthesis typically increases [ 31 , 32 ]. Although the present study did not examine this aspect, the amount of menstrual blood loss may differ between the symptom group and non‐symptom group. Third, ovarian hormone may influence Tf levels. In particular, estrogen is known to promote Tf expression [ 39 , 40 ]. In the present study, participants in the symptom group had higher estradiol levels during the luteal phase (prior to menstruation) on MDQ20 and MDQ32 compared to the non‐symptom group (S1E and S1K Figs). This may suggest that a subsequent sharp decrease in estrogen led to a delayed increase in Tf synthesis, resulting in reduced iron transport to cells. Hemopexin (Hx) is a protein that showed significant changes associated with MDQ32 (tendency to take a nap or stay in bed). Hx is a 60 kDa plasma protein classified as an acute phase protein [ 41 , 42 ]. It has high affinity for heme and functions as a scavenger of free heme, which can exert cytotoxic effects [ 43 , 44 ]. Although changes in Hx during the menstrual cycle have not been reported, some studies have described its involvement in iron metabolism. Interestingly, Fiorito et al. reported increased iron uptake in the duodenum of hemopexin knockout mice [ 45 ]. Since iron is lost through menstrual blood, Hx may play a feedback role in maintaining iron homeostasis. It is well known that there are sex differences in sleep quality [ 46 ]. Sex steroids are considered key factors underlying this difference, but low iron levels have also been associated with sleep quality [ 47 ]. In the present study, Tf levels were also altered in association with MDQ32. Therefore, a decrease in iron supply to the brain may occur, potentially triggering a compensatory increase in iron absorption in the duodenum. For MDQ21 (depression), kininogen‐1 (KNG‐1) showed a significant association. KNG‐1 is a component of the kallikrein‐kinin system (KKS), which has diverse functions, including roles in inflammation, blood coagulation, and blood pressure regulation [ 48 , 49 ]. During the luteal phase, increased levels of estradiol and progesterone lead to elevated blood pressure, resulting in the secretion of aldosterone [ 50 , 51 , 52 ]. In the present study, the observed downregulation of KNG‐1 during the menstrual phase suggests that blood pressure may not return to normal levels immediately after menstruation begins. There is a known association between hypertension and depression [ 53 ]. Therefore, altered blood pressure regulation may influence mental health during menstruation. Ceruloplasmin (CP), which showed significant association with MDQ20 (tension), is a glycoprotein that functions as a copper transporter [ 54 ]. In addition to copper transport, CP has various physiological roles, including the regulation of iron metabolism, enzymatic activities such as ferroxidase activity, antioxidant effects, and pro‐oxidant properties [ 55 ]. Although no direct connection between CP and the menstrual cycle has been reported, approximately 95% of copper binds to CP. Therefore, copper may be involved in the manifestation of menstrual symptoms. One study reported an inverse correlation between copper intake and depression, suggesting that low copper levels may negatively affect mental health [ 56 ]. Moreover, CP possesses ferroxidase activity and is involved in iron metabolism. Since both CP and Tf showed changes in the present study, alterations in iron dynamics mediated by CP and Tf may contribute to the observed symptoms. Interestingly, the four proteins mentioned above are synthesized in the liver. The liver plays a role in sex hormone metabolism, and it is established that impaired liver function can lead to irregular menstrual cycle [ 57 ]. Furthermore, a study reported that curcumin supplementation in women with PMS and dysmenorrhea improved liver function test values [ 58 ]. These findings suggest a connection between liver function and the menstrual cycle. Therefore, the decreased hepatic proteins observed in this study may reflect reduced liver function. Additionally, individual differences in the alterations of specific liver‐synthesized proteins may account for the variety of menstrual symptoms observed among individuals. Next, four proteins—serine protease‐1, junction plakoglobin (JUP), desmoglein (Dsg), and dermcidin (DCD)—were identified in association with various menstrual symptoms. Serine protease‐1, known for its role in pancreatitis, showed a significant association with MDQ1 (Muscle stiffness) [ 59 , 60 ]. Its up‐regulation has been reported in cervical cancer [ 61 ]. JUP and Dsg were associated with MDQ6 (General aches and pains) and MDQ20 (Tension), respectively. As structural proteins, they are typically not present in serum under normal conditions, although serum JUP levels are correlated to endometriosis [ 62 , 63 , 64 ]. Finally, DCD was found to be associated with MDQ28 (distractable). DCD, an antimicrobial protein secreted by sweat glands, has also been secreted from skeletal muscle and promotes apoptosis [ 65 , 66 ]. Notably, serum DCD levels were lower in female basketball players compared to control women, whereas this decrease was not observed in male basketball players [ 67 ]. These findings suggest that DCD may have a sex‐specific role related to muscle inflammation in women. In contrast, the present study found that DCD levels significantly increased during menstruation in women who reported being distractible. However, research on these four proteins and their roles as reflected in serum during the menstrual cycle is limited, and further investigation is required to clarify their biological significance. This study has several limitations. The first is potential bias, as all participants included in this study were softball club members; the results may be affected by group‐specific characteristics. Next, the questionnaire has a self‐reporting nature, and recall bias and over‐reporting of their condition might occur. These biases could distort the true prevalence of reported symptoms. Next, the imbalance of hormones and menstrual phases is also one of the limitations. To compare the menstrual symptom group to the non‐symptom group, hormone levels that regulate the menstrual cycle should be aligned. This study was designed as a prospective study. However, capturing continuous samples during the menstrual cycle in such a design is challenging, as hormone levels fluctuate unpredictably and vary between individuals. In addition, even if some participants are in the same menstrual phase at one sampling point, hormone imbalance may lead to different physiological events among participants. A final limitation is the sample size. This study did not focus on a few specific symptoms but instead targeted a total of 46 menstrual‐related symptoms. Since participants answered their various experienced symptoms through the questionnaire, there was an imbalance of sample numbers between the symptom group and the non‐symptom group on some question items. The present study recruited 46 participants who have completed all sampling. However, only 14 participants out of 46 were included in the analysis due to hormone imbalances. As a result, some MDQ items have had sample size bias on either the symptom group or the non‐symptom group. It is necessary to conduct further study that address these limitations. Based on the current study, the sample size needs at least 90 participants, assuming coefficient of determination ( R 2 ) is 0.15 with the same multiple regression model as used in this study (the number of explanation variables are three). Menstrual symptoms are recurring features of cycle, sampling should therefore be conducted over at least two cycles per individual participants at least to ensure reproducibility. Although this study did not use ovarian hormones level directly, these hormones may play a key role in the mechanisms underlying menstrual related symptoms due to their diverse physiological functions. A major problem in using hormone levels is the considerable inter‐individual variability among women. Therefore, it may not be appropriate to use absolute hormone levels in the analysis, as the result could be cofounded by this variability. Instead, using relative hormone values—such as those based on the first day of menstruation or intra‐individual change rates—may be effective. A study by Shultz et al., which targeted women engaged in regular physical activity, demonstrated the reproducibility of hormone profiles across two consecutive menstrual cycle. The authors also noted that intra‐individual variability was smaller than inter‐individual variability [ 68 ]. It is also well established that both the length of menstrual cycle and the duration of each phase vary among individuals [ 69 ]. Therefore, the sampling interval is a critical consideration. In this study, sampling was conducted once a week over 5 weeks. However, this approach posed some challenges, such as missing samples from a complete cycle for some participants and the absence of data from specific phases. To ensure comprehensive phase coverage, more frequent sampling, such as continuous sampling over three to 4 days per week for several months, may be necessary, depending on research target. Taken together, the present study revealed potential relationships between specific proteins and menstrual symptoms through a 2DE approach. However, transferrin was the only protein that showed significant changes across multiple MDQ items, while other proteins were associated with individual symptoms. This finding may suggest that each menstrual symptom is associated with distinct protein‐specific mechanisms, although some common pathways may also be involved. Therefore, investigating these proteins is crucial to clarify the mechanisms of menstrual symptoms and develop symptom‐relief treatments. Given the exploratory nature of this work, these associations should be validated in independent cohorts of women with diverse characteristics to confirm their robustness.

Unlike major research which focuses on the relationship between ovarian hormone and menstrual symptoms, this study tried to demonstrate the association between proteins and menstrual symptoms with a proteomic approach. It is revealed that six out of 46 items in the MDQ are significantly associated to eight proteins. Among them, only one protein showed common significance for several questions. This suggests that the onset mechanisms of various menstrual symptoms may differ from one symptom to another. Each protein which showed significance in this study may serve as a potential stepping stone toward uncovering the mechanisms of menstrual cycle‐related changes in women's condition and could assist in the development of new diagnostic or therapeutic approaches. This is the first study to analyze 46 menstrual‐related symptoms by proteomics, revealing that protein changes vary depending on each menstrual symptom. It is hoped that these findings will contribute to the advancement of health management for improving women's quality of life.

Menstrual symptoms are significant health problems for women, causing both physical and mental effects affecting their quality of life. Premenstrual syndrome (PMS) is characterized by both physical and mental symptoms which appear during the luteal phase [ 1 ]. The prevalence of PMS is high, although it varies across studies due to the lack of standardized diagnostic criteria, and many women suffer from its symptoms [ 2 , 3 , 4 ]. In particular, it has been reported that 95% of Japanese women, which is the target of this study, experienced PMS, including mild symptoms [ 5 ]. Primary dysmenorrhea (PD) is defined as lower abdominal pain before or during menstruation without any pelvic pathology. In addition to pain, other associated symptoms may include nausea, vomiting, insomnia, diarrhea, and mood changes such as depression or irritability [ 6 , 7 ]. The prevalence of PD is 70%–80% and it is especially common in young women, with the rate decreasing with aging [ 8 , 9 , 10 ]. Both PD and PMS impact the lower quality of life for women. Some studies have reported that women with PMS experience more stress and lower quality of life than those without PMS, which further interferes with daily life, decreases productivity in adults and lower grades for students [ 11 , 12 , 13 ]. Similarly, PD can also further cause distress in relationships with family and friends, as well as affect performance at the workplace and in school [ 14 ]. In addition, some studies have reported that menstrual symptoms negatively affect athletic performance in female athletes, the population targeted in the present study [ 15 , 16 ]. Menstrual symptoms significantly affect women's life; however, their molecular mechanisms are poorly understood. In the case of PMS, ovarian hormone (estradiol and progesterone) levels and their sensitivity in women may be related to occurrence of symptoms [ 17 , 18 ]. On the other hand, current evidence of PD supports that excess prostaglandin leads to PD, contributing to vasocontraction and uterine contractions which cause lower blood and oxygen supply to the uterus. In addition, it has been reported that progesterone is related to the production of prostaglandin and prostaglandin level in the PD group is higher than the control group [ 14 , 19 , 20 ]. Thus, some studies show evidence of the pathogenesis of menstrual symptoms. However, most of these studies are linked to biochemical substrates such as ovarian hormones, with few studies focusing on proteins. Proteins perform various functions, including roles in transporter and catalysis [ 21 , 22 ]. Therefore, it is important to explore the functions of proteins in relation to menstrual symptoms, as changes in protein levels directly affect our body condition. Interestingly, various factors such as aging and diets affect protein states including their post‐translational modifications (PTMs) [ 23 , 24 ]. For example, serum protein carbonyl levels increase under the exam in female students and IgG glycosylation changes during the menstrual cycle [ 25 , 26 ]. Thus, PTMs may be an important factor affecting women's health. This study adopted a high‐performance two‐dimensional electrophoresis (2DE) to elucidate protein changes by menstrual symptoms in female athletes. 2DE is a proteomic approach that allows direct detection of PTMs. The technique used in this study is high‐throughput, which is demonstrated to have superior resolution in protein profiling compared to conventional methods [ 27 ]. This study aimed to explore proteome changes associated with menstrual symptoms in female athletes living in the same dormitory. The proteins identified in this study, which are related to menstrual symptoms, may support the development of new diagnostic and therapeutic approaches based on the pathology of these symptoms.

The authors declare no conflicts of interest.

Figure S1: Changes in female hormone levels by menstrual phase and menstrual symptoms. Figure S2: Two‐dimensional electrophoresis images acquired in this study. Figure S3: Full blot images of Western blotting for validation experiments. Table S1: Sample information collected in this study. Table S2: Estradiol levels over 5 weeks (pg/mL). Table S3: Progesterone level over 5 weeks (ng/mL). Table S4: Identification result of proteins spots by LC–MS/MS. Appendix S2: fba270109‐sup‐0002‐AppendixS2.zip.