{"paper_id":"aaccace0-ed0d-4c17-9ed1-d655246da3e9","body_text":"Granulosa cells are somatic cells surrounding\nand supporting the\noocytes in mammalian ovarian follicles. At the antral follicle stage\nduring which a fluid-filled cavity called the antrum is formed, the\ngranulosa cells that originally enclose the oocyte differentiate into\n2 distinct subtypes under the control of oocyte-secreted factors:\nthe cumulus granulosa cells (cGCs) that are in intimate metabolic\ncontact with the oocyte via gap junctions and the mural granulosa\ncells (mGCs) that line the wall of the follicular antrum. 1  Notably, there is not only anatomical heterogeneity\nbut also functional differences between these two subtypes of cells\nthroughout follicle development. Generally, cGCs play a major role\nin oocyte growth, development, and meiotic maturation, while mGCs\nprimarily execute an endocrine function and engage in mitosis activity\nleading to follicular growth. 2  Distinct\ngene expression profiles between these two cell types reflect the\ndifferent physiological functions. For instance, genes encoding steroidogenic\nenzymes, as well as a range of growth factors and hormone receptors\nwere differentially expressed between mGCs and cGCs in rodents as\nwell as in humans. 3 , 4\nA physiological level of\nreactive oxygen species (ROS) plays a\nkey role in the development of oocytes and follicles. In the ovarian\nfollicles ROS are fundamental for oocyte meiotic maturation. 5  Free radicals, including some of the most reactive\nROS molecules, can act as secondary messengers for cellular signaling\nand are involved in the regulation of ovarian physiological processes,\nsuch as ovulation. 6  A certain amount of\noxygen is also required in oocyte meiotic maturation. 7  However, overabundance of free radicals can lead to oxidative\nstress, which is detrimental to oocyte meiotic maturation 8  and associated with reduced female fertility. 9  Although ROS are inevitable products of aerobic\nmetabolism, lifestyle factors such as obesity and pathological conditions\nsuch as endometriosis may also contribute to oxidative stress. The\n“free radical theory of aging” proposed more than half\na century ago has been implicated to be associated with fertility. 10  Emerging evidence supports the central role\nof oxidative stress in age-related oocyte quality decline such as\ndisturbed meiotic spindle formation that is responsible for chromosomal\nsegregation leading to a higher chance of aneuploid oocytes. 11  Conceivably, as cells surrounding and supporting\nthe oocytes, free radical levels of cGCs and mGCs and their responses\nto oxidative stress may be associated with oocyte quality. However,\nwhether these two subtypes of granulosa cells respond differently\nto oxidative stress remains unknown: on the one hand, these cells\nare derived from the same origin, and some certain oxidative stress\nbiomarkers in both kinds of cells have been indicated to strongly\nassociate with oocyte developmental competence and even embryo quality; 12  on the other hand, the heterogeneity between\nmCGs and cCGs in many different biological aspects is increasingly\nrecognized. 2 , 3\nTo date, free radical detection in\nbiological samples remains a\ngreat challenge in practice due to the short lifespans and low abundance. 13  Electron spin resonance (ESR), which is considered\nas the gold standard for direct free radical detection, is still faced\nwith the problem that free radicals in biological samples are naturally\nat a low steady-state concentration. 14  Although\ndifferent indirect techniques have been developed and applied for\nthe measurement of free radical levels in granulosa cells, 15  a temporospatial measurement with single cell\nresolution has never been achieved. By measurement of the signal generated\nfrom redox interaction or oxidative cell damage instead of the radicals\nthemselves, indirect detection is not able to obtain any spatial information\nor single-cell resolution. In addition, some indirect approaches are\nbased on a free-radical dye reaction and subsequent fluorescent molecule\ngeneration. As a result, these dye-based methods suffer from the risks\nof photobleaching over time and, thus, are not suitable for real-time\nand longer measurements. Moreover, they reveal the history of free\nradical generation in samples rather than the current levels. 16\nWe use a method based on negatively charged\nnitrogen vacancy (NV – ) defects of fluorescent nanodiamonds\n(FNDs). 17  Due to their stable fluorescence,\nthey can be\nused for long-term tracking and labeling. 18  NV centers also change their optical properties in response to their\nmagnetic surrounding. 19  Importantly, these\nFNDs are also excellently biocompatible. 20  This method offers a new way for direct free radical measurement\nin real-time and at a subcellular level. NV center-based sensing has\nalready been used for applications in 2-dimensional materials or magnetic\ncharacterization of materials under high hydrostatic pressures, sensing\nof nanoscale temperature, 21  magnetic nanostructures, 22  or paramagnetic ions; 23 , 24  NV centers are traditionally utilized in physics while their application\nin biological fields is less explored. NV centers can “feel”\nmagnetic noise from free radicals and convert it into an optical signal.\nWith this method free radical sensing on a subcellular level has been\ndemonstrated in a variety of mammalian cells 25 − 27  as well as\nyeast or bacteria, 28  suggesting its potential\napplication in granulosa cells.\nIn the current study, we aim\nto investigate if diamond-based relaxometry\ncan be applied to measure free radicals in human mGCs and cGCs at\nsubcellular levels in real-time. Further, we aim to test whether cGCs\nand mGCs respond differently to induced oxidative stress at a subcellular\nlevel.\n\nFigure  1  shows an\noutline of the quantum sensing experiments that were conducted in\nthis study on cGCs and mGCs isolated from the preovulatory follicles\nof females during an IVF procedure.\nSchematic summary of applying relaxometry\nto probe free radicals\nin human granulosa cells by using fluorescence nanodiamond (FND).\n(A) Oocyte retrieval was performed 36 h after luteinizing hormone\n(LH) in females who were planned for  in vitro  fertilization\n(IVF) for fertility problems. Granulosa cells are collected during\nthis procedure. (B) Cumulus granulosa cells (cGCs) and mural granulosa\ncells (mGCs) were isolated from preovulatory follicles separately,\nfollowed by culture and incubation with FNDs. Two subtypes of FNDs\nwere used, bare-FNDs (directed to cytoplasm) and aVDAC2-FNDs (directed\nto mitochondria) for 24 h before relaxometry; (C) The raw data for\nrepresentative T1 relaxation curves are shown. These were generated\nfrom different dark times plotted against the fluorescence intensity.\nThe inset presents the pulsing sequence used in relaxometry. The green\nblocks indicate when the laser was on, while the red blocks indicate\nwhen the photoluminescence (PL) from the FND was read out.\nTo measure free radicals on a subcellular\nlevel, two kinds of particles were applied: uncoated FNDs (bare-FNDs)\nthat are expected to be in the cytoplasm at the timing of the measurement,\nand FNDs coated with physically adsorbed anti-VDAC2 antibodies which\nbind to voltage-dependent anion channel isoform 2 (aVDAC2-FNDs) that\nare targeted to the mitochondrial outer membrane as previously described. 26\nThe sizes and zeta potentials of bare-FNDs\nand aVDAC2-FNDs are shown in  Figure S1 .\nTo exclude a potential contribution of contaminating cells, such as\nnot fully removed red blood cells or immune cells from human follicular\nfluid samples, identification of human primary granulosa cells was\nperformed by flow cytometry. Follicle stimulating hormone receptor\n(FSHR) was used as a granulosa cell biomarker due to its high specificity\nfor granulosa cell demonstrated by previous studies. 29  As flow cytometry plots showed, proportions of FSHR  + cGC were as high as 99.46% after isolation ( Figure S2 ). Proportions of FSHR  + mGC were 19.19%\nbefore Percoll purification and strainer filtering but reached as\nhigh as 97.3% afterward ( Figure S3 ), suggesting\nhigh purity of both cGCs and mGCs.\nTo\nconfirm that FNDs do not affect metabolic activity and therefore cell\nviability, an MTT assay (3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl\ntetrazolium bromide) was performed. mGCs and cGCs were incubated with\ndifferent concentrations of bare-FNDs (1 and 5 μg/mL), aVDAC2-\nFNDs (1 and 5 μg/mL), or HCl for 24 h, respectively. HCl (0.1\nM) was used as a positive control, as it induces cell death. There\nare no differences between untreated cells and the groups exposed\nto FNDs either at a concentration that we later applied in this study\n(1 μg/mL) or a concentration that is relatively high for relaxometry\n(5 μg/mL) ( p  > 0.05,  Figure  2 A–B), suggesting a good biocompatibility\nof FNDs in human primary GCs. To test if FNDs induce changes in the\nintracellular ROS level, a 2′,7′-dichlorodihydrofluorescein\ndiacetate (DCFHDA) assay was performed. Specifically, cGCs and mGCs\nwere incubated with bare FNDs (1 μg/mL), aVADC2 coated FNDs\n(1 μg/mL) or menadione (10 μM) for 24 h. Menadione was\nused as a positive control, as it induces intracellular ROS generation.\nThere are no differences between the negative controls and the cells\nexposed to bare or aVADC2 coated FNDs ( p  > 0.05,  Figure  2 C–D), indicating\nthat bare and aVADC2 coated FNDs do not affect intracellular ROS levels\nin cGCs and mGCs and thus can be used for different measurements in\nthese cells.\nEffects on FNDs on cell viability and intracellular reactive\noxygen\nspecies (ROS). Cell viabilities were determined by a thiazolyl blue\ntetrazolium bromide (MTT) assay after incubation with low bare-FND\nand aVDAC2-FND concentration (1 μg/mL), high bare-FND and aVDAC2-FND\nconcentration (5 μg/mL) and HCl (0.1 M) respectively in cGCs\n(A) and mGCs (B). DCFHDA assay shows intracellular ROS generation\nafter incubation with bare-FNDs (1 μg/mL), aVDAC2-FNDs (1 μg/mL),\nand menadione (5 μM) for 24 h in cGCs (C) and mGCs (D). 100%\nrepresents a control without any stimuli exposure. The experiment\nwas repeated for cells from six patients, and error bars represent\nthe standard deviations. The data were analyzed by using one-way ANOVA\nfollowed by a Tukey post hoc test in comparison to the control groups.\n**** p  < 0.0001.\nBefore relaxometry\nexperiments, the uptake of FNDs or aVDAC2-FNDs (1 μg/mL) by\nGCs after incubation for 2 and 24 h was evaluated with confocal z-scans.\nTypical images of bare-FNDs and aVDAC2-FNDs uptake by cGCs and mGCs\nfollowing different incubation times are shown in  Figure S4 . More aVDAC2-coated FNDs are found inside cells\nin comparison to bare-FNDs after incubation for 24 h in cGCs ( p  < 0.05) and mGCs ( p  < 0.001) ( Figure S5A ). Although the uptake of bare-FNDs\nby cGCs and mGCs was comparable ( p  > 0.05), the\nnumber\nof aVDAC2-FNDs was significantly higher in mGCs compared to that of\nbare-FNDs ( p  < 0.01) ( Figure S5A ).\nTo explore the localization of bare FNDs and aVDAC2-FNDs,\nthe colocalization of bare FNDs or aVDAC2-FNDs with  translocase\nof the outer mitochondrial membrane 20  (Tom20) was observed\nby confocal microscopy. As shown in  Figure  3 , bare-FNDs colocalize less with TOM20 but\nare always in the proximity of actin filaments, whereas aVDAC2-FNDs\nare prone to colocalize with TOM20, suggesting localization at mitochondria.\nSubcellular\nlocation of FNDs revealed by confocal microscopy. Bare-FNDs\n(1 μg/mL) and aVDAC2-FND (1 μg/mL) were incubated with\ncGCs (A–B) and mGCs (C–D) for 24 h. Tom20 antibody,\nan outer mitochondrial membrane biomarker, was used to show mitochondria.\nColor code: blue, Tom20; green, Phalloidin-FITC, staining actin filaments\n(also known as F-actin); red, bare-FND or aVDAC2-FND. The scale bar\nis 10 μm.\nAs shown\nin the fluorescent images acquired by our homemade relaxometer ( Figure S5B ), the FND (red arrow) is very bright\nin comparison to the autofluorescence of the cell.\nBefore relaxometry\nmeasurements, toxicity of menadione at different concentrations was\nevaluated by MTT assay, and HCl served as a positive control. As shown\nin  Figure S6A–B , incubation with\n10 μM of menadione for 30 min was within the safe range for\nboth cGCs and mGCs. To exclude the potential effects of DMSO-dissolved\nmenadione on FNDs, relaxometry measurement of FNDs was performed in\nthe absence of cells. As shown in  Figure S6C , no significant T1 change was observed when DMSO-dissolved menadione\nwas added to FNDs in the absence of cells.\nBoth types of nanodiamonds\nwere used in relaxometry measurements\nto detect real-time free radical level changes after menadione treatment\nin cGCs from 4 patients ( Figure  4 ). For each type of nanodiamond, 4–6 particles\ninside the cGCs were selected. For each particle, we tracked the free\nradical change for 30 min, and time dependent T1 reductions were observed\nfor both FND variants. Among the 4 patients, cGCs from 3 patients\npresent significant changes of T1 values from 10 min on ( Figure  4 A,B,D), and an earlier\nsignificant change of T1 (from 5 min on) is observed in 1 patient\n( Figure  4 C) using bare-FNDs.\nA decrease in the T1 value corresponds to an increase in the free\nradical concentration near the nanodiamond sensor. To estimate the\nradical concentrations equivalent with T1 values, a calibration of\n*OH radical measurement with known concentrations was obtained from\nprevious work. 24\nBox-whisker plots shows\nreal-time free radical change determined\nby T1 after menadione treatment in cGCs from 4 different patients.\nFor each patient, T1 of 4–6 bare-FNDs or aVDAC2-FNDs followed\nby menadione treatment at different time points (0, 1, 5, 10, 20,\n30 min) were measured. Bare-FNDs (left side, boxes of color green)\nand aVDAC2-FNDs (right side, boxes of color purple) measured in Patient\n1 (A–B), Patient 2 (C–D), Patient 3 (E–F), Patient\n4 (G–H). The right Y axis represents the estimated radical\nconcentration obtained from previous work. 24  Each particle is represented by one color, and each curve represents\nmeasurements performed on the same particle at different time. The\ndata were analyzed by using a paired  t  test in comparison\nto the control groups. *  p  < 0.5, **  p  < 0.01, ***  p  < 0.001, ****  p  < 0.0001.\nTo compare T1 measurements with traditional methods,\nintracellular\nROS and mitochondrial superoxide detection assays were performed.\nThe classical intracellular ROS probe, DCFH-DA, was subsequently applied\nto validate the oxidative stress induced by menadione. A significant\nintracellular ROS change was observed from 20 min on ( p  < 0.0001,  Figure S7A ). For the results\nobtained using aVADC2-FNDs, cGCs from all the patients exhibited a\nsignificant T1 decrease from 5 min, and the reduction is time-dependent.\nA commercial assay kit, MitoSox, was subsequently applied to quantify\nthe mitochondrial superoxide induced by menadione. As shown in  Figure S7B , a significant change in mitochondrial\nsuperoxide production was observed after incubation with menadione\nfor 20 min ( p  < 0.0001).\nSimilarly,\nboth types of nanodiamonds were used in relaxometry measurements to\ndetect real-time free radical level changes after menadione treatment\nin mGCs from 4 patients ( Figure  5 ). For each type of nanodiamond, 4–6 particles\ninside the mGCs were selected according to the criteria mentioned\nin  Methods . For each particle, we tracked\nthe free radical change for 30 min, and time dependent T1 reductions\nwere observed for both FND variants. In mGCs from all patients we\nobserved significant changes of T1 values from 5 min on either using\nbare-FNDs or aVADC2-FNDs. Consistent with cGC, the reduction of T1\nis time dependent. The classical intracellular ROS probe, DCFH-DA,\nwas applied to validate the oxidative stress induced by menadione.\nSignificant changes of intracellular ROS levels were observed from\n10 min on ( p  < 0.05,  Figure S7C ). MitoSox was applied to quantify mitochondrial superoxide\ninduced by menadione. As shown in  Figure S7D , significant changes in mitochondrial superoxide production were\nobserved after incubation with menadione for 20 min ( p  < 0.0001).\nBox-whisker plots shows real-time free radical change\ndetermined\nby T1 after menadione treatment in mGCs from 4 different patients.\nFor each patient, T1 of 4–6 bare-FNDs or aVDAC2-FNDs followed\nby menadione treatment at different time points (0, 1, 5, 10, 20,\n30 min) were measured. Bare-FNDs (left side, boxes of color green)\nand aVDAC2-FNDs (right side, boxes of color purple) measured in Patient\n1 (A–B), Patient 2 (C–D), Patient 3 (E–F), Patient\n4 (G–H). The right Y axis represents the estimated radical\nconcentration obtained from previous work. 24  Each particle is represented by one color, and each curve represents\nmeasurements performed on the same particle at different times. The\ndata were analyzed by using a paired  t  test in comparison\nto the control groups. *  p  < 0.5, **  p  < 0.01, ***  p  < 0.001, ****  p  < 0.0001.\nTo investigate if menadione induces free\nradical changes in the cytoplasm and the mitochondria, we compared\nthe T1 values measured by bare-FNDs and aVADC2-FNDs in the GCs. In\nboth cGCs and mGCs, there were no significant T1 differences between\nthe measurement of bare-FNDs and aVADC2-FNDs in the absence of menadione\n( p  > 0.05 at 0 min) ( Figure  6 A–B). This suggests that cytoplasmic\nand mitochondrial free radical levels are indistinguishable in the\nphysiological state. However, in the presence of menadione, T1 values\nmeasured by aVADC2-FNDs were significantly lower than those measured\nby bare-FNDs from 1 min on in cGCs ( p  < 0.05 at\n1 min and  p  < 0.0001 at 5, 10, 20, 30 min) ( Figure  6 A) and from 10 min\non in mGCs ( p  < 0.01 at 10, 20 min and  p  < 0.05 at 30 min) ( Figure  6 B). The percentage of T1 reduction was significantly\nbigger from 5 min on when the FNDs were targeted to mitochondria in\nboth cGC ( p  < 0.0001) ( Figure  6 C) and mGC ( p  < 0.001)\n( Figure  6 D). These\nresults indicate that mitochondria are the main sites of menadione-induced\nfree radical generation. In addition, in cGCs, the long shape of the\npurple halves of the violin graph indicated a bigger variance in the\nmitochondria than that in the cytoplasm, the free radical of which\nwere presented as the green half-violin graph. In general, both bare-FNDs\nand aVDAC2-FNDs showed bigger T1 variations at physiological states\nwhich decreased in time after exposure to menadione. This suggests\na nonlinear change in T1 upon oxidative stress. A smaller variation\nof the T1 values beyond a certain level of oxidative stress may be\na result of saturation.\nHalf-violin plots showing differences between\ncytoplasm and mitochondrial\nfree radical change in response to menadione at different time points.\nT1 measurements from all the bare-FNDs and aVDAC2-FNDs were compared\nto show the differences between cytoplasm and mitochondrial absolute\nfree radical levels at 0, 1, 5, 10, 20, 30 min after menadione treatment\nin cGCs (A) and mGCs (B). Percentages of T1 changes from all the bare-FNDs\nand aVDAC2-FNDs were compared to show the differences between cytoplasm\nand mitochondrial free radical changes at 1, 5, 10, 20, 30 min when\ncompared to the control in cGCs (C) and mGCs (D). Significance was\ntested by using a  t  test. *  p  <\n0.5, **  p  < 0.01, ***  p  <\n0.001, ****  p  < 0.0001.\nTo investigate if menadione induces\nfree radical changes in cumulus and mural granulosa cells, we compared\nthe absolute T1 values at different time points as well as the T1\nchange of these two subtypes of cells after exposure to menadione.\nUsing bare-FNDs which measure the cytoplasm free radical levels, we\nfound no significant differences between cGCs and mGCs at any time\npoint ( p  > 0.05) ( Figure  7 A). However, from 5 min on, the percentage\nof T1 reduction in mGCs was significantly bigger compared to that\nin cGCs ( p  < 0.01) ( Figure  7 B). These results suggest that although the\ncytoplasmic free radical levels are comparable between cGCs and mGCs\neither in the basal state or after exposure to oxidative stress, the\ncGCs are more resistant to oxidative stress. When aVADC2-FNDs were\nused to measure the mitochondrial free radicals, the absolute T1 values\nwere significantly higher in mGCs at the basal state ( p  < 0.05) and during the early period (at 1 and 5 min) of oxidant\nexposure ( p  < 0.01 and  p  <\n0.05, respectively) ( Figure  7 C). However, at later time points (at 10, 20, and 30 min),\nthe absolute T1 values between the two types of cells become comparable\n( p  > 0.05) ( Figure  7 C), and no significant percentage of T1 reduction at\nany time point between cGC and mGCs was observed ( p  > 0.05) ( Figure  7 D). These results implicate that although the mitochondrial free\nradical levels are lower in mGC physiologically, upon oxidative stress\nat first, the mitochondrial free radical response to the oxidant between\nthese two types of cells was similar.\nHalf-violin plots showing differences\nbetween cGCs and mGC free\nradical levels and changes in response to menadione at different time\npoints. (A) T1 values measured by all the bare-FNDs in cGCs and mGCs\nwere compared to show the differences between cGC and mGC absolute\ncytoplasm free radical levels at 0, 1, 5, 10, 20, 30 min after menadione\ntreatment. (B) Percentages of T1 changes measured by all the bare-FNDs\nin cGCs and mGCs were compared to show the differences between cGC\nand mGC cytoplasm free radical changes at 1, 5, 10, 20, 30 min when\ncompared to the control. (C) T1 measured by all the aVDAC2-FNDs in\ncGCs and mGCs were compared to show the differences between cGC and\nmGC absolute mitochondrial free radical levels at 0, 1, 5, 10, 20,\n30 min after menadione treatment. (D) Percentages of T1 changes measured\nby all the aVDAC2-FNDs in cGCs and mGCs were compared to show the\ndifferences between cGC and mGC mitochondrial free radical changes\nat 1, 5, 10, 20, 30 min when compared to the control. Significance\nwas tested by using a  t  test. *  p  < 0.5, **  p  < 0.01, ***  p  < 0.001, ****  p  < 0.0001.\nTo exclude the possibility that T1 values can be affected by the\nsurrounding environment, such as temperature or stress induced by\nthe measurement conditions themselves during the 30 min measurement,\nT1 were measured in both kinds of cells using bare-FNDs and aVDAC2-FNDs\nwithout menadione. As shown in  Figure S8 , no significant T1 changes were observed in both cGC and mGCs, either\nincubated with bare-FNDs or aVDAC2-FNDs for 30 min.\n\nIn this study, we aimed to use a quantum\nsensing approach to probe\nfree radical generation in human cGCs and mGCs. To achieve this goal,\nwe first confirmed that the diamond relaxometry allows for temporospatial\nfree radical measurement in both types of granulosa cells with high\nbiocompatibility and sensitivity. Further, similarities as well as\ndifferential free radical responses on subcellular levels were revealed:\nmitochondria may serve as the main sites of menadione induced free\nradical generation in both kinds of cells; cGCs may be more resistant\nto oxidative stress compared to mGCs. Several previous studies have\ndemonstrated that menadione induces intracellular ROS, especially\nsuperoxide via one-electron transfer reactions at multiple cellular\nsites. 30 , 31  In this study, menadione is used as an external\noxidant to trigger the generation of free radicals in both mGCs and\ncGCs. The temporospatial property of relaxometry provides us with\nsome interesting biological findings based on the differential ROS\ngeneration between cytoplasm and mitochondria as well as cGCs and\nmGCs upon oxidative stress.\nSeveral cellular organelles, including\nmitochondria, lysosome,\nendoplasmic reticulum, are responsible for ROS generation. 32  Among these, mitochondria are energy powerhouses\nin most mammalian cells and are considered as the major source of\nROS, which results from electron escape from the internal mitochondrial\nmembrane as a natural byproduct of mitochondrial oxidative phosphorylation\n(OXPHOS) during ATP generation. 33  In addition,\nthe role of mitochondria in granulosa cell functions has been highlighted\nin the literature. For instance, mitochondria ATP is the primary source\nof energy for the FSH-dependent proliferation and differentiation\nof mouse granulosa cells during folliculogenesis. 34  Impaired mitochondrial OXPHOS function in granulosa cells\nwas also associated with maternal aging and oocyte incompetence. 35  Thus, in our model of ROS challenge, mitochondrial\nROS changes were compared with cytoplasmic ROS after menadione induction.\nInterestingly, we found that although cytoplasm and mitochondrial\nfree radical levels are indistinguishable in the physiological state\nin both kinds of cells, significantly bigger percent changes were\nobserved in mitochondria compared to cytoplasm after 5 min of oxidant\ninduction. This finding suggests that mitochondria may be the major\nsites of menadione-induced free radical generation, which is in accordance\nwith previous work showing that menadione causes rapid superoxide\naccumulation in neuronal cells. The authors speculated that this accumulation\noccurred preferentially in mitochondria of hippocampal neuronal cells. 30\nAs described earlier, cGCs are in direct\ncontact with and metabolically\ncoupled with the oocyte through gap junctions. Thus, cGCs have long\nbeen believed to be the gatekeepers for the oocyte from its surroundings\nand play a variety of essential roles in the growth and meiotic maturation\nof oocytes. It is also known that the oocyte relies on the surrounding\ncGCs for providing protection against excessive ROS since it does\nnot have the capacity on its own to mobilize all the necessary antioxidant\ndefense mechanisms. 36  On the other hand,\nmGCs execute more endocrine rather than defense functions. 2  Thus, it is not surprising that cGCs are more\nresistant to oxidative stress after exposure to oxidative stress than\nmGCs.\nThere are also some results different from those from\nprevious\nstudies of diamond relaxometry in other cell types. In contrast to\nprevious studies showing higher T1 variability in the macrophage cytoplasm, 26  we found a bigger variance in the mitochondria\nthan that in the cytoplasm in cGCs. This could be attributed to the\nfact that cells differ in ROS generation on subcellular levels. In\nmacrophages, free radical generation in mitochondria is generally\nhigher than in this case, which results in T1 values being closer\nto saturation and thus a lower variability. The mitochondria in cGC\nare taking part in a variety of free radical generating pathways,\nand thus free radical generation can differ a lot in ways from what\nactivities they are involved. In contrast, free radical values in\nthe cytoplasm are more stable probably due to less free radical-generating\npathway involvement. In comparison, the distributions of T1 measured\nby bare-FND and aVDAC2-FND were rather comparable in mGCs, indicating\na relatively equivalent variation of the free radical response between\nmitochondria and cytoplasm in mGCs.\nThere are several advantages\nof this technique. Relaxometry enables\ntemporospatial measurements of the free radical load, which means\nthe detection can be performed in real-time and on subcellular levels.\nTo date, different methods, either indirect or direct, have been utilized\nin several studies measuring ROS levels in cumulus and/or mural granulosa\ncells. 15  Indirect detection of ROS in biological\nsamples by measuring ROS-induced lipid, protein and DNA modification\nis gaining popularity due to its stability and reliability. 13  For example, markers of ROS-induced lipid modification\nsuch as malondialdehyde (MDA), which is based on the principle that\nfree radicals induce lipid peroxidation by attacking lipids containing\ncarbon–carbon double bonds, has been used in a recent study\nto investigate the correlation between cGC ROS and oocyte quality. 37  Although these fluorescent dye-based methods\nhave a high sample throughput within several minutes and appear stable\nand reliable in clinical settings, it is a one-time measurement not\ncapable of detecting temporospatial changes upon oxidative stress\ninduction. Additionally, this method serves as a bulk assay where\ninformation from many cells is averaged, and it is limited in spatial\nresolution, since dye molecules can diffuse freely in cells. It is\nalso a potential drawback that no further treatment or analysis of\nthat group of cells is possible. In the current study, two traditional\nassays (DCFH-DA and MitoSox) were applied for comparison to measure\nthe cytoplasm and mitochondria. Also, the results should be cautiously\ninterpreted due to photobleaching with time. 38  In addition, to eliminate the background effects, a control group\ncontaining cells without treatment is always required, which means\nthat the results are always a percentage change over the control group\nand are not suitable for basal ROS measurement across different patients.\nFurther, due to different detected radicals and reaction principle,\nassay results from different kits cannot be compared directly, and\nthus, the main source of ROS production is hardly tracked. Finally,\nconsistent to previous findings, 27  our\nresults show that free radical change occurs earlier than those in\neither of the two other methods, suggesting higher sensitivity of\nT1 measurements compared to these traditional methods. With respect\nto clinical practice, the single-cell resolution property of diamond\nrelaxometry provides an ideal solution to ROS detection with high\nresolution in mGCs and cGCs, even with low cell numbers, and their\nassociation with oocyte competence and embryo development can be explored\nin assisted reproductive techniques.\nHowever, quantum sensing\nalso has points that should be approached\ncautiously. First, there is considerable variability among T1 values\nmeasured by different FNDs. This can be explained by differences between\nparticles in size, shape, and exact surface area. In addition, nanodiamonds\ncan be in a different environment within the cells, and thus free\nradical concentrations can vary within a few nanometers due to their\nshort-life nature. However, since nanodiamonds allow long-term measurements,\nit is possible to follow a specific particle and thus differentiate\nbetween the original variability and changes due to the intervention,\nsuch as a menadione challenge in our case. The T1 changes induced\nby the oxidant are sensitive and robust, suggesting that this nanoscale\nMRI may perform better in the research of free radical change than\ndetecting free radicals in bulk samples, where a large amount of cells\nand medium is measured at once. Second, diamond relaxometry requires\na FND inside cells, which means the uptake of FNDs by cells is a premise\nof T1 measurement, and it takes some time for the FND to be taken\nup by cells and directed to the organelle of interest. However, this\nalso applies to dye-based methods. Since with FNDs extremely small\namounts (single particles) are needed for a measurement, they are\nusually tolerated better than conventional dyes. Additionally, it\nneeds to be noted that our measurements are very local. This is an\nadvantage for spatial resolution but might lead to relevant stress\nresponses that we miss since they occur in a location that is not\naccessible for nanodiamonds. Finally, we conducted measurements in\na specific cell type, but there might be other cell types that could\nbe interesting to study as well.\nIn conclusion, this study demonstrates\nthe feasibility of diamond\nrelaxometry as a novel, sensitive method for measuring free radical\nchanges in human granulosa cells in real-time and at subcellular levels.\nIn addition, time-dependent free radical generation in response to\noxidants differs on a subcellular level as well as between the cumulus\nand granulosa cells. Studies of cGC and mGC free radical in patients\nof different infertility factors can be expected by using this diamond\nrelaxometry technique and would be interesting and additive to the\ncurrent work.\n\nPrimary granulosa\ncells were obtained from preovulatory follicles from individual healthy\nwomen between 20 and 35 years old undergoing ovum pick-up for  in vitro  fertilization (IVF) at the department of Reproductive\nMedicine University Medical Centre Groningen, The Netherlands from\nOctober 2022 to February 2023. Inclusion criteria were as follows:\n(1) 25–35 years old; (2) normal menstrual cycle; (3) standard\nlong hyperstimulation protocol; (4) at least 3 follicles with diameter\n>18 mm at the day of follicle triggering; (5) intracytoplasmic\nsperm\ninjection (ICSI). Women with polycystic ovarian syndrome, endometriosis,\ndiminished ovarian reserve, chromosome abnormality, or hydrosalpinx\nwere excluded since these ovarian factors might affect follicle growth,\nand granulosa cells may thus behave very differently under oxidative\nstress.\nEthical approval from the Institutional Review Board\nwas requested and waived since anonymized waste material (granulosa\ncells that routinely become available after oocyte retrieval) was\nused. An informed consent form was signed by all patients, and their\nmaterial was processed anonymously. (All patients agreed on the use\nof their cumulus granulosa cells and mural granulosa cells, which\nroutinely become available after oocyte retrieval and otherwise would\nbe discarded as waste material.)\nGenerally, during the use of\noral contraceptive pills (OCP) hormonal\ndownregulation was started with daily injections of subcutaneous triptorelin\n0.5 mg of GnRH analogue mg (Decapeptyl, Ferring Pharmaceuticals, The\nNetherlands). After 12 days, patients received human menopausal gonadotrophin\n150–225 international unit (IU) per day (Menopur, Ferring Pharmaceuticals,The\nNetherlands) or Follitropine alfa rec FSH 150–225 IU (Gonal\nF, Merck Serono, Italy). Oocytes were collected 36 h after injection\nof 250 μg of recombinant human chorionic gonadotropin (hCG)\n(Ovitrelle Merck Serono, Italy). During oocyte retrieval, follicular\nfluid containing mGCs was collected in 50 mL centrifuge tubes. Parts\nof the cumulus cell clusters were mechanically separated from the\ncumulus-oocyte complex and stored in 15 mL centrifuge tubes in a G-MPOS\n(Vitrolife). Samples of cGCs and mGCs were then brought to a cell\nculture hood for further purification and culture.\nThe isolation of cGCs and\nmGCs was performed as previously described. 39  In brief, cGC clusters were dispersed by gently pipetting before\ncentrifugation in HBSS (Life technologies, USA) for 4 min at 1400\nrpm. For mGCs isolation, the follicular fluid was centrifuged 7 min\nat 600 g , and the pellet was resuspended in phosphate-buffered\nsaline PBS. Blood cells were removed by layering the cell pellet using\na 40% Percoll gradient (Fisher Scientific, cat. no. 10607095) and\n20 min centrifugation at 600 g. Cells from the interface were collected\nand washed in PBS through 4 min centrifugation at 1400 rpm, followed\nby resuspension of pelleted cells in 1 mL of trypsin and 3 min incubation\nat 37 °C and pipetting to disperse the clustered cells. Both\ntypes of cells were then passed through a Falcon 40 μM strainer\n(Corner, cat. no. 352340), followed by being cultured in Dulbecco’s\nModified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12) (Life Technologies,\nUSA, cat.no. 11320033) supplemented with 10% FCS, 1% penicillin-streptomycin-amphotericin\nB at 37 °C, and 5% CO 2 . For T1 measurements and confocal\nmicroscopy, cells were plated in 35 mm culture dishes (CELLview Culture\ndish, nontreated, 4 compartments, glass bottom, Greiner Bio-One) at\na density of 4000 cells/compartment. For DCFHDA, MTT, and MitoSOX\nassays, cells were plated into culture at a density of 10000 cells/well\nof 96-well plates. After 48–72 h, the culture medium was refreshed.\nCells were used in experiments within 1 week. The purity of cGCs and\nmGCs was confirmed by FSHR fluorescence microscopy.\nFNDs with a hydrodynamic diameter of 70 nm\ncontaining >300 NV centers purchased from Adámas Nanotechnologies\n(Raleigh, NC, USA) were selected since they are suitable for T1 measurement\nfor several reasons. 24  First, it takes\na longer time for FNDs with smaller diameters to obtain a good signal-to-noise\nratio due to their lower brightness. Second, it is comparatively more\ndifficult to track smaller FNDs since they move faster. Larger particles\nalso have the advantage that they contain more NV centers, and the\nmeasurement is already an average of all these NV centers, which greatly\nimproved reproducibility. However, NV centers in the core of these\nlarger particles are too distant from the surface to sense the spin\nnoise from radicals. The particles we used are produced by high pressure\nand high temperature synthesis followed by grinding and size separation\nto the desired size. To increase the NV center yield nanodiamonds\nare irradiated with 3 MeV electrons at a fluence of 5 × 10 19  e/cm 2  and annealed at the temperature exceeding\n700 °C by the manufacturer. 40  As a\nlast step of their synthesis by the manufacturer, FNDs are cleaned\nin oxidizing acid, resulting in oxygen terminated particles. These\nare widely used in the field and have been characterized before. 41  Anti-VDAC2 antibody ([C2C3], C-term, catalog\nno. GTX104745) was obtained from GeneTex (The Netherlands). aVDAC2-FNDs\nwere freshly prepared before use as previously described, 3  followed by size and zeta potential measurements\nusing the Malvern ZetaSizer Nanosystem (Dynamic Light Scattering;\nMalvern Instruments Ltd., Malvern, UK;  www.malvern.com ). Menadione (cat.\nno. M5625-25G) was purchased from Merck. Tom20 antibody (rabbit, catalog\nno. sc-11415) was purchased from Santa Cruz Biotechnology (USA). Goat-α-rabbit\nAlexa 405 secondary antibody (cat. No. A-31556) and MitoSOX Red mitochondrial\nsuperoxide indicator (cat. no.  M36008 ) were bought from Thermo Fisher.\nPhalloidin–fluorescein isothiocyanate (FITC) was obtained from\nSigma-Aldrich, The Netherlands. A cellular ROS assay kit (DCFDA/H2DCFDA,\nab113851) was purchased from Abcam. FSH receptor Polyclonal Antibody\n(rabbit, cat. no. bs-0895R) was bought from Bioss (USA).\nAfter isolation, both kinds of granulosa\ncells were separately fixed with 3.7% paraformaldehyde rather than\nbeing plated. Then cells were first incubated with rabbit-FSHR antibody\ndiluted in 500 μL of 0.1% BSA for 3 h at room temperature, followed\nby incubation with 1:200 of goat-α-rabbit Alexa 488 secondary\nantibody for 45 min at room temperature protected from light. Cells\nwithout any staining were regarded as negative control groups. Cells\nstained with only a secondary antibody were also set to exclude unspecific\nbinding. mGC samples before Percoll purification were also set for\ncomparison. Cells were subjected to Quanteon Flow Cytometer Systems\n(Agilent Technologies, US) using a laser at 488 nm. Data acquisition\nand analyses were performed using NovoExpress software and gated for\na high level of FITC expression.\ncGCs and mGCs were seeded as 6 × 10 5  per well in a 96-well cell culture plate (Tissue Culture-Treated,\nFlat-Bottom with lid, Corning) and incubated for 48 h to allow attaching\nto the bottom. For intracellular ROS measurement, cells were first\nwashed with PBS and then incubated with 10 μg/mL DCFDA prevented\nfrom light at 37 °C for 45 min. Then the DCFDA staining solution\nwas removed, and cells were washed and replaced with PBS. Next, cells\nwere either treated with 10 μM menadione and the fluorescence\nintensity was measured. As a control we used cells with PBS only.\nAs experimental groups, 5 time points were evaluated: 1, 5, 10, 20,\nand 30 min after menadione treatment. For experiments in which menadione\nserved as a positive control of intracellular ROS induction, cells\nwere first incubated with 5 μM menadione in 37 °C for 24\nh before 10 μg/mL DCFDA was added and incubated at 37 °C\nfor 45 min. All the fluorescence intensities were measured by a plate\nreader (Bio Tek, Santa Clara, CA) at Ex/Em = 485/535 nm prevented\nfrom light. Cells without staining with DCFDA were recorded for background\nsubtraction.\nMitochondrial superoxide\nwas detected using the MitoSOX Mitochondrial Superoxide Indicator\n(Introgen,  M36008 ) following the manufacturer’s protocol. Briefly,\nall samples except the control groups were treated with 10 μM\nmenadione, and the medium was removed after 1, 5, 10, 20, and 30 min.\nAfter washing with PBS, 1 μM of the superoxide detection compound\nwas added followed by incubation for 30 min at 37 °C. After staining,\ncells were washed in their cell culture medium twice, and the fluorescent\nproduct was measured by a plate reader (Bio Tek, Santa Clara, CA)\nat an excitation of 396 nm and an emission of 610 nm.\nAn MTT assay was carried out to evaluate\nthe viabilities of cGCs and mGCs following nanodiamond incubation\nas well as determining a safe menadione concentration, which does\nnot affect cell viability after incubation for 30 min. This assay\nprovides an evaluation of cell metabolic activity by detecting nicotinamide\nadenine dinucleotide phosphate (NAD(P)H) dependent oxidoreductases\nactivity. Cells cultured in 96-well plates were treated with 0.75\nμg/mL MTT dissolved in DMEM/F12 medium. After 3 h of incubation\nat 37 °C, the reagent was removed, and 2-propanol was added to\nthe samples to dissolve the formazan formed inside the cells. To confirm\nthat FNDs do not affect cell viabilities, both kinds of cells were\nincubated with FNDs (1 and 5 μg/mL) for 24 h. To find a safe\nmenadione concentration for cGC and mGC viability, both kinds of cells\nwere pretreated by menadione of 2, 10, 50, 100 μM for 30 min.\nIn all experiments, cGCs and mGCs treated with 0.1 M HCl were used\nas positive controls, while cells without any treatment were used\nas a negative control. The absorbance of the colored solution was\nmeasured by using a plate reader (Bio Tek, Santa Clara, CA) at 570\nnm. All experiments were performed in six replicates. Samples were\nnormalized against the mean absorbance value of the negative control,\nrepresented as a line at the value 1.\nFor FND uptake experiments, cGCs\nand mGCs were incubated at 37 °C and 5% CO 2  with both\nbare-FNDs and aVDAC2-FNDs (1 μg/mL) for 2 and 24 h, respectively.\nAt each time point, the cell culture medium with FNDs was removed.\nAfter washing with 1× phosphate-buffered saline (PBS), cells\nwere fixed with 3.7% paraformaldehyde for 10 min at room temperature.\nAfter fixation, cells were either covered with PBS and stored in at\n4 °C for later staining or immediately stained. For FND uptake\nexperiments, cells were first treated with 1% Triton X-100 for 3 min\nto permeabilize the cell membranes. Next, 5% PBSA was applied and\nincubated for 30 min to block the nonspecific background, followed\nby adding staining solution mixed by 4 μg/mL 4′,6-diamidino-2-phenylindole\n(DAPI) and 2 μg/mL phalloidin–fluorescein isothiocyanate\n(FITC) in PBSA for visualization of nuclei and F-actin, respectively.\nFinally, 500 μL of PBS was added to cover the sample for imaging.\nImages were taken using a 63× 1.30 GLYCEROL objective in a Leica\nSP8X DLS confocal microscope (Leica microsystems, Wetzlar, Germany)\nwith a 405 nm laser to detect DAPI, a 488 nm laser to measure phalloidin-FITC,\nand a 561 nm laser to detect FNDs. Z-stacks were performed to determine\nthe numbers of particles inside the cells at both time points. Three\nindependent experiments were performed, and at least 30 cells were\nquantified for each time point.\nFor colocalization experiments,\ncGCs and mGCs were incubated at 37 °C and 5% CO 2  with\nboth bare-FNDs and aVDAC2-FNDs (1 μg/mL) for 24 h. After fixation,\ncell membrane permeation, and nonspecific background blockage as described\nabove, cells were first incubated with rabbit-Tom20 antibody diluted\nin 500 μL of 0.1% BSA for 3 h at room temperature or overnight\n(1:500) at 4 °C. Then, cells were incubated with 1:200 of goat-α-rabbit\nAlexa 405 secondary antibody and 2 μg/mL phalloidin–FITC\nin 500 μL of 0.1% BSA for 45 min at room temperature protected\nfrom light. Finally, 500 μL of PBS was added to cover the samples\nfor imaging. Images were taken using a 63× 1.30 GLYCEROL objective\nin a Leica SP8X DLS confocal microscope (Leica microsystems, Wetzlar,\nGermany) with a 405 nm laser to detect TOM20, a 488 nm laser to measure\nphalloidin-FITC, and a 561 nm laser to detect FNDs. All images were\nprocessed by using the FIJI software.\nCells were first washed with PBS, and\nthen PBS was replaced with DMEM/F12 medium after incubation with 1\nμg/mL bare-FNDs or aVDAC2-FNDs for 24 h. After particle identification\nand localization, we performed T1 (relaxometry) measurements using\na laser pulsing sequence in a custom-made magnetometry setup, which\nis in principle a confocal microscope with some modifications 42  ( Figure  1 C).\nIn a typical T1 measurement, we pump NV centers\ninto the bright ms = 0 state of the ground state. We then probed after\ndifferent dark times if the NV centers remained in this state or returned\nto the darker equilibrium between ms = 0 and ms = + −1. In\nthe presence of free radicals this process occurs faster and can be\nused to quantify free radical generation. 43  To extract the magnetic noise level from these plots, we used a\ndouble exponential fit of the form: 1 This fit is different from\nthe single exponential fits that are used for single NV center measurements.\nAfter the observation was made that the single exponential model\ndoes not represent the data well in ensembles, this model was determined\nempirically. This fit considers that there are different NV centers\nwith different T1 values within each particle. While both constants\nrespond to changes in magnetic noise, the longer constant is more\nsensitive to changes in magnetic noise. This was found earlier by\nmeasuring different known concentrations and observing how the different\nconstants respond. Thus, to quantify free radical generation we use\nthe longer time constant Tb which we call T1. 24  A detailed discussion of the biexponential model as well as comparison\nwith other models can be found. 44\nThis measurement reveals a signal that is equivalent to T1 in conventional\nMRI. However, since NV centers only detect their local environment\n(up to a few tens of nm), this method offers nanoscale resolution. 45\nThe laser we used is a 532 nm laser at\n50 μW at the location\nof the sample (measured in continuous illumination). The measurement\nsequence consisted of 5 μs long laser pulses separated by variable\ndark times τ from 0.2 to 1000 μs. To conduct the pulsing\nsequence, an acousto-optical modulator (Gooch & Housego, model\n3350-199) and a magnification oil objective (×100) (Olympus,\nUPLSAPO 100XO, NA 1.40) were applied. Under a bright field camera\n(Thorlabs), the following criteria were checked and confirmed before\nan FND particle was selected: 1. The FND was well located inside a\ncell; 2. the brightness was around 3 million photon counts/s; 3. The\nfluorescence is stable since bleaching structures are background fluorescence\nrather than FNDs. Then, the first T1 measurement of the selected FND\nwas performed. After this measurement, menadione (10 μM) was\ngently added to the DMEM/F12 medium to trigger free radical generation.\nT1 measurements on the same FND were recorded at 1, 5, 10, 20, 30\nmin after menadione addition. Before each measurement, it was again\nconfirmed that a particle was an FND by tracking its location, photon\ncount, and stable fluorescence. All T1 measurements were conducted\nat room temperature and under ambient air. Due to the relatively low\nlaser power and the fact that the laser is mostly off during a T1\nmeasurement, we did not observe any measurable heating. 26\nQuantitative data were presented\nas the mean ± standard deviation (SD). All statistical tests\nwere conducted using R programming language. Significance was tested\nby using one-way ANOVA followed by a Tukey post hoc test or  t  test and is specifically indicated in the legend of each\nfigure. All statistical tests were compared to the control group and\ndefined as ns  P  > 0.05, * P  ≤\n0.05, ** P  ≤ 0.01, *** P  ≤\n0.001, and **** P  ≤ 0.0001.","source_license":"CC-BY-4.0","license_restricted":false}