Systemic transport of saturated fatty acids lessens in regenerating Drosophila
Intestinal cells are exposed to chemical and mechanical stress, and damaged cells are constantly replaced by proliferation of intestinal stem cells (ISC). One well studied model of intestinal regeneration is adult Drosophila flies fed with food supplemented with the drug dextran sodium sulfate (DSS)6,7. In Drosophila, feeding behavior is sex-specific, in brief, males feed at irregular intervals alternating with roaming behavior (Supplementary Movie 1), while mated females show a constant food intake and less mobility (Supplementary Movie 2). In this study, we use mated females to exclude metabolic differences created by the feeding behavior. Continuous intestinal tissue damage could influence feeding behavior and nutritional input, thus, we monitored the food intake of DSS-treated flies using the CApillary FEeder assay (CAFE)8 and found no changes in feeding behavior with respect to controls (Supplementary Fig. 1a). In addition, blood sugar and protein content of regenerating flies were comparable to fed wild type controls but noticeably higher than in starving wild type animals (Supplementary Fig. 1b, c), indicating that the function of the interstitial epithelium is not compromised by DSS administration. Further, DSS treatment of adult flies did not impede viability (Supplementary Fig. 1d). However, our assay induces a proliferative response, validating the use of this assay as a regenerative model (Supplementary Fig. 1e).
Although Drosophila flies are sterol auxotrophs, they are able to synthesize saturated and mono-unsaturated fatty acids9. We expected that our experimental flies feeding on lipid-free food (see material and methods) will be starved of sterols and poly-unsaturated fatty acids. Further, we speculated that with the loss of dietary lipid (DL) flows, we would easily detect if a systemic re-distribution of essential lipids in regenerating flies takes place. To investigate changes in circulatory lipid contents, we isolated blood, named hemolymph, of adult flies and probed for lipoprotein particles, the principal systemic lipid carriers. In fruit flies, the lipoprotein Lipophorin (LPP) is responsible for carrying the bulk of diacylglycerols (DAGs) and sterols to supplement target cells with lipids10,11. First, we measured the number of LPP particles and their lipid load. Quantification of total hemolymph LPP by western blotting showed no differences between regenerating and control flies (Fig. 1a). To investigate the LPP density distribution, we isolated hemolymph and separated LPP in density gradients and found no difference in the LPP density profile in regenerating flies with respect to untreated controls (Fig. 1b). To assess the lipid load of LPP particles we extracted lipids from our samples and quantified glycerophospholipids and DAGs using mass spectrometry (Supplementary Data 1). We found no changes in the glycerophospholipid distribution of lipoproteins (Supplementary Fig. 2e, lower panels). However, regenerating animals transport less DAGs compared to non-regenerating animals (Supplementary Fig. 2f), and recorded a significant decrease in three of the most prominent DAG species: DAGC30:1 (n = 6, regenerating = 75.38 pmol/vol, control = 95.25 pmol/vol, p = 0.049 unpaired t-test with Welch correction), DAGC32:1 (n = 6, regenerating = 96.99 pmol/vol, control = 119.0 pmol/vol, p = 0.019 unpaired t-test with Welch correction) and DAGC32:2 (n = 6, regenerating = 60.78 pmol/vol, control = 74.48 pmol/vol, p = 0.027 unpaired t-test with Welch correction) (Fig. 1c). These DAG species consist mostly of saturated and mono-unsaturated fatty acids. In contrast, we did not find a significant reduction of DAGs composed of poly-unsaturated fatty acids (PUFAs). Lastly, we measured by thin layer chromatography (TLC) the sterol and sterol ester content in hemolymph of regenerating flies. Sterols form the lipid shell of LPP particles, whereas sterol esters resemble cargo lipids. We detected a decrease in the sterol:sterol ester ratio in regenerating flies indicating that the proportion of transported sterol esters is increased (Fig. 1d). Thus, our results indicate that regenerating flies are either incapacitated in their synthesis of fatty acids or the use of saturated fatty acids is exceeding the mobilization capacity from lipid stores. Further, the proportional increase in sterol esters points to compensatory transport possibly required to maintain cellular sterol homeostasis.
a Total amount of Lipophorin (LPP) in Drosophila hemolymph in control and regenerating animals. (n = 8, 9 flies pooled per sample, 3% DSS=regeneration). Unpaired Student’s t-test. Mean and SD are indicated. b LPP particles present in hemolymph, separated in a density field. (n = 4, 27 flies pooled per sample). Arrow points to high density. Multiple unpaired Student’s t-tests. Mean and SD are indicated. c DAG yields in hemolymph (n = 6, 3 flies pooled per sample, 3% DSS=regeneration). Multiple unpaired t-tests with Welch correction, *p < 0.05. Mean and SD are indicated. d Circulating Sterols in hemolymph normalized to the amount of sterol esters (SE). n = 6, 5 flies pooled per sample, unpaired Student´s t-test, **p < 0.01, mean and SD. Mean and SD are indicated. e TAG yields in head samples of control and regenerating flies. Identical lipid species are connected using a line to visualize their individual change between the mean of the two experimental groups. Wilcoxon test for total TAG changes, p** < 0.01. n = 6, 3 heads pooled per sample. f Saturation profile of all measured TAG species from Drosophila head samples. DB double bonds. Mean and SD are indicated.
Fatty acid stores increase in fat body cells
In mammals and insects, liver-like cells are the main provision stock of low-density lipoproteins (LDL)11,12,13, and intestinal cells or white adipocytes represent the source for the majority of circulating lipids11,14. In Drosophila, fat body cells (FB) form functional hybrids of vertebrate hepatocytes and white adipocytes. To investigate if the reduction of saturated DAGs in circulation is correlating with changes in respective fatty acids in FBs we measured the lipid content of these cells using mass spectrometry (Supplementary Data 1). We analyzed the lipid composition of head samples due to their distance from the regenerating intestine. The majority of triacylglycerols (TAGs) measured in head samples originate from FB cells. In head samples, we found that total TAG levels are significantly higher in regenerating flies (Figs. 1e and 2b, wild type panel). Interestingly, the increase in TAG was driven by TAG species likely composed of FAs with 14 or 16 carbons (Fig. 1e, TAG species indicated). However, regeneration did not induce a change in the general saturation profile of TAGs (Fig. 1f).
a Expression of insulin-like peptides (Dilps) measured by qPCR. The expression in regenerating (3%DSS, circles) animals of Dilp2, Dilp3 and Dilp5 is shown as fold change to control (dots). (n = 5, each sample = 5 animals pooled, unpaired t-test, *p < 0.05). Mean and SD are indicated. b TAG yields normalized to PE and the average corresponding control signal in fat body samples of WT (black), FB-InRKD (lpp > InRRNAi, brown) and FB-PI3KKD (lpp > PI3KRNAi, purple) flies during control conditions (dots) and regeneration (3% DSS, circles). -Welch’s t-test, *p < 0.05). n = 15 (WT control), 18 (WT 3% DSS), 12 (InRKD, control and 3% DSS), 6 (PI3KKD control) and 10 (PI3KD 3% DSS), 3 animals pooled per sample. Mean and SD are indicated. c Sterol yields normalized to PE in fat body samples. Sample description, see (b). n = 15 (WT control), 18 (WT 3%DSS), 12 (InRKD, control and 3% DSS), 6 (PI3KKD control) and 10 (PI3KD 3% DSS), 3 animals pooled per sample. Mean and SD are indicated. d Sterol yields normalized to PE in midgut samples. Sample description, see (ba)., Welch’s t-test **p < 0.01. n = 16 (WT control), 15 (WT 3% DSS), 17 (InRKD, control), 14 (InRKD 3% DSS), 12 (PI3KKD control and 11 (PI3KD 3% DSS). 2 animals were pooled per sample. Mean and SD are indicated. e Quantification of pH3 positive cells per midgut in WT and InRKD flies in control condition and upon regeneration. Welch’s t-test, **p < 0.001, *p < 0.05. Wild type: n = 7 (Control), n = 8 (3% DSS); InRKD: n = 6 (both conditions). Mean and SD are indicated. f Graphic overview depicting the InR knock-down strategy in the fat body cells. DSS treatment in the InRKD results in markedly reduced intestinal stem cell division rates in regenerating animals.
Our data rejects the idea that FA synthesis is reduced in regenerating flies. Although correlative, our results indicate that especially systemic traffic and hepatic storage saturated/mono-unsaturated medium-chained FAs are regulated by regeneration. Thus, we speculate that the reduction of circulating saturated DAGs (Fig. 1c) is likely based on increased storage of respective FAs in cells. Earlier, we have measured increased sterol ester levels in circulation (Fig. 1d). Unfortunately, the presence of wax in the tracheal cuticle does not permit a clean sterol ester measurement in Drosophila tissues. Nevertheless, we found sterol levels not affected in FB of regenerating flies (Fig. 2c, wild type panel). Taken together, we speculate that regenerating flies remove saturated and mono-unsaturated FAs selectively from the system by storing these lipids in FBs.
Insulin signaling in fat body cells regulates lipid content in regenerating tissue
Insulin signaling is uniquely important for storing energy as fat in humans. In Drosophila, FB cells are reported to change lipid turnover and the storage of TAG in response to insulin15. To investigate if the expression of insulin-like peptides (Dilp) is affected by regeneration, we measured relative mRNA of Dilp2, 3 and 5 by qPCR. Our results show upregulation of Dilp2 expression while Dilp3, and 5 remain unchanged (Fig. 2a). Insulin is required for ISC division in Drosophila7. We found that regenerating female flies increase specific TAGs in fat body cells (Fig. 2b) and less sterol esters are transported systemically (Fig. 1d). Insulin signaling is regulating TAG storage and thus, we speculated that impaired insulin signaling in FB cells could affect lipid homeostasis. We focused on the fact that regenerating animals decrease systemic transport of saturated FAs and increase sterol traffic. One rationale for modifying lipid transport may arise from the specific needs of regenerating tissues at the local level. To test the idea, we knocked down the insulin receptor specifically in FB cells (FB-InRKD). We found that sterol levels are not affected in the FB of intact or regenerating flies (Fig. 2c, mid panel). However, when we measured sterol levels in intestinal cells, we found a profound decrease in sterols in midguts from regenerating FB-InRKD flies (Fig. 2d mid panel). To confirm the result, we impaired the expression of a downstream factor of the insulin signaling cascade in fat body cells, the phosphoinositide 3-kinase (PI3K). We found that FB-PI3KKD cells mimic intestinal sterol changes found in FB-InRKD (Fig. 2d right panel). This suggests that insulin signaling in the FB, required to maintain intestinal sterol homeostasis, is regulated by the metabolic branch of the insulin signaling pathway. To investigate if preventing the systemic regeneration-driven lipid adaptations affects intestinal regeneration, we assessed the ISC division rate in midguts. We found that ISC in regenerating FB-InRKD flies decreased their proliferation rate by about 50% compared to the wild type condition (Fig. 2e). While DSS treatment in wild type flies induced an increase in proliferation (Fig. 2e left panel), this increase in proliferation was significantly reduced in the FB-InRKD condition (Fig. 2e right panel). Taken together, induced intestinal regeneration requires insulin signaling in FB cells to regulate sterol dynamics of midgut cells (Fig. 2f). However, we cannot rule out that changes in insulin signaling may affect cell numbers in the FB, thus reflecting different lipid content. Therefore, we normalized our sterol readings to membrane lipids (PE). Adaptations of sterol metabolism should not be surprising, since cellular sterol levels are critical to allow cell functionality or proliferation16. Nevertheless, Drosophila as a sterol auxotroph, could potentially reflect insect-specific lipid managements in response to continuous regeneration. Further work is needed to quantify the regenerative success in intestines upon impaired insulin signaling in FBs.
Regenerating axolotl show normal feeding behavior and nutrition
To test the translational potential of our findings in fruit flies, we studied limb regeneration in axolotls. The process of regeneration after limb amputation poses important differences from the Drosophila intestinal injury model, such as the complexity and multi-tissue content of the structure. In mice, findings indicate sex specific systemic lipid transport and lipid turnover17, therefore, we included both sexes in our study. We used axolotl larvae, 6 days after injury (dpa) when a transitional structure begins to form, the blastema, containing an initial rush of progenitor cells that will give rise to the new limb. During earlier timepoints, the main events are characterized by cell death and the initial immune response, with less proliferation. At later time points, a major remodeling event results in skeletal resorption and the beginning of differentiation5. For this study, we considered that the high proliferation rates at 6 days within the blastema and adjacent tissue in the stump, reflects the cell activation in response to the injury that we report in Drosophila and offers a point of interspecies comparison. Furthermore, at 6 dpa the expression of regeneration-specific markers can already be observed18. At this developmental stage, axolotls are fed with freshly hatched Artemia franciscana nauplii, rich in PUFAs and cholesterol (see materials and methods) – representing a homogenous, lipid-rich food (Supplementary Data 2). Thus, unlike Drosophila on lipid-free food, regenerating axolotls can rely on dietary lipid (DL) flows to control their lipid homeostasis. First, we controlled for the feeding behavior and nutrient content of regenerating axolotls. We videotaped foraging regenerating (5 dpa) and intact animals, n = 3. After 1 h, both experimental cohorts ingested the same amount of artemia as observed in these representative videos (Supplementary Movie 3−6, Supplementary Movie 3 = Male intact, Supplementary Movie 4=Male 5 dpa, Supplementary Movie 5=Female intact, Supplementary Movie 6=Female 5 dpa). Next, we measured circulating sugars and proteins. Similar to our findings in fruit flies, regeneration in axolotl larvae does not result in changes in blood sugar or protein levels (Supplementary Fig. 1f, g). In addition, we tested if regenerating axolotls experienced starvation by quantifying the expression of the transcription factor Forkhead box protein O1 (Foxo1) in the liver, which is known to increase its own expression profile in starving cells19. We did not detect significant mRNA changes in regenerating animals, indicating that limb amputation and subsequent recovery time did not induce malnutrition of experimental axolotls (Supplementary Fig. 3d).
Systemic lipid transport is adjusted in a sex-specific manner
Regenerating tissues rely on high cellular proliferation rates and thus, we speculated that the flow of DLs are suitable resources to build cellular membranes. We investigated if lipid transport and compositions were different in 6 dpa animals with respect to intact axolotls. The molecular organization of systemic lipid transport in axolotls is largely uncharted. Therefore, we searched within the recently sequenced axolotl genome20 to discover which proteins axolotls encode that resemble lipoproteins of other species. We used reported protein sequences of identified lipoproteins from mice, frogs, and humans as templates and found homologous genes in axolotls, including ApoB and others (Supplementary Data 3). Thus, axolotls possess a more complex lipid transport than fruit flies and produce low- and high-density lipoprotein particles like other studied vertebrates21. Mass spectrometry showed that axolotl lipoproteins, similar to mammals, mainly transport TAGs, ~47 mol% (Supplementary Data 4).
Given the more complex lipid transport, we additionally investigated if sex-specific differences in lipid transport exist, based on mice data showing sex-specific lipid turnover17. We speculated that adaptive metabolic changes equalizing sex specific turnover help synchronize tissue repair. Here, we demonstrate by TLC that intact and regenerating male and female axolotls predominantly transport sterol esters and sterols in LDL particles (Supplementary Fig. 2a−d). In addition, we determined plasma cholesterol levels using a colorimetric approach (Fujifilm DRI CHEM NX600V), and detected a consistent 30% reduced cholesterol in pooled samples of regenerating females with respect to intact siblings, whereas males show a slight increase (Fig. 3a). Males rely exclusively on LDL particles to traffic TAGs systemically (Fig. 3c). These lipoprotein density profiles in males did not change during regeneration, suggesting a consistent approach to TAG transport that is independent of the injury status. In contrast, in female axolotls at homeostatic conditions and during regeneration, TAG transporting lipoproteins are spread across a wide range of densities, indicating they are transported by multiple different lipoprotein classes (Fig. 3b). In addition, mass spectrometry revealed that the content of trafficked TAGs in axolotls is also sex-specific: Female axolotls have more TAGs in their circulatory system than males and in consequence, traffic more polyunsaturated fatty acids (PUFAs)(Fig. 3d). Following amputation, regenerating female axolotls do not change systemic TAG transport and show TAG profiles similar to non-regenerating controls. Astonishingly, the difference between females and males in intact conditions, recedes during regeneration as males increase circulating TAG yields and the proportion of PUFAs (Fig. 3e). This increase aligns male TAG levels with those of females, potentially indicating that such levels are optimal for regeneration. In both sexes, the majority of TAG species belong to TAGC52 and TAGC54, which represent about 78% of all plasma TAGs (Supplementary Data 4, Supplementary Fig. 2g). Further, we detected the glycerophospholipid phosphatidylserine (PS) in the plasma of intact females but not in males. Interestingly, following amputation, PS were not found in regenerating female samples, but did appear in circulation of regenerating males (Supplementary Fig. 2e top panel). As PS are found primarily on high density particles (HDLs)22, these findings further underline the existing substantial sex-specific differences in lipid transport. In contrast to TAGs and PS, sterol and SE levels in circulation remain unchanged between all experimental axolotl groups (Supplementary Fig. 2h).
a Cholesterol content measured in axolotl plasma. Plotted as a percentage of the deviation from wild type cholesterol content (indicated as zero). Mean and SD are indicated. n female = 3 pools, from a total of 14 intact females or 14 6dpi females; male n = 4 pools from a total of 18 intact males or 18 6dpi males. Pools were created from animals of the same batch. Mann−Whitney test with *p < 0.05 or indicated. b, c TAG density distribution of female (a) and male (b) of axolotl plasma plotted as percentages of total TAG yields. Density gradient indicated by the arrow in the x axis. n = 6 per group. Mean and SD are indicated. Female (F), male (M). d, e Saturation profile of TAGs in axolotl plasma. n = 5, (d) comparison of intact F and intact M. e comparison of 6 dpa F and 6 dpa M. Statistical significance for each unsaturation group was determined via unpaired t-test with *p < 0.05 or indicated. Mean and SD are indicated. Female (F), male (M).
We show that in homeostasis, axolotls possess a systemic sex-specific, lipoprotein-based, lipid transport, and we conclude that systemic traffic of fatty acids and sterols are likely not directly linked. Further, lipid transport is also dynamically different in regenerating males and females however, equaling the initial homeostatic difference.
Liver serves as systemic sterol buffer in females
To assess if changes in lipid transport exert influence in lipid storage, we first established where fat is stored in axolotl larvae. Histological analysis shows that at this developmental stage, white adipose tissue is largely scarce23 (and own observations). However, we found that liver cells are strongly enriched with TAGs and sterol esters (Supplementary Fig. 3a, b), concentrating lipid storage in one organ, thus facilitating a parallel to FB cells in fruit flies. The hepatic lipid profile of intact animals showed that female livers contain lower amounts of unsaturated TAGs, compared to males (Fig. 4a) (Supplementary Data 5) and that one of the most abundant species, TAGC52, is significantly different between females and males (Fig. 4a′). However, limb regeneration in females induced an increase in TAG and PUFAs with respect to controls (Fig. 4b, c). In addition, we found an increase in hepatic cholesteryl esters (CEs) of females, while they remain unchanged in males, resulting in a similar profile for high and low abundant CEs in both sexes during regeneration (Fig. 4d). CEs represent a biological storage form of sterols and on demand, CEs can be converted into cholesterol and free FA. An increase in CEs could point to raising cholesterol levels, however, there is the possibility that the axolotl sterol profile is not dominated by cholesterol24. Thus, we used TLC to assess total sterol levels in the axolotl liver, which showed that sterol yields of female or male hepatocytes are not increasing in regenerating animals (Supplementary Figure 3c). Hence, we suggest that regenerating axolotls are capable of buffering increasing hepatic cholesterol levels by converting super-numerical sterol molecules into sterol esters.
a TAG yields in liver samples of control and regenerating female and male axolotls. Same lipid species are connected with a line to visualize their individual change between the two experimental groups. Most abundant lipid species are labeled. n = 6. Student’s t-test to compare between intact F and 6 dpa F and between intact M and 6 dpa M. a’ Data in (a) graphed by TAG species including all groups of DB. n = 6, per species one-way ANOVA and Tukey´s multiple comparison test. *p ≤ 0.05. Mean and SD are shown. Female (F), male (M). b, c Quantification of stored lipids in the axolotl liver by the degree of unsaturation via mass spectrometry. Comparison between (b) intact male and intact female, (c) regenerating male and female animals. n = 6 per group. Statistical significance for each unsaturation group was determined via unpaired t-test. For all graphs, the mean and SD are shown. d Cholesteryl ester species yield from female and male axolotl samples (n = 6). Kruskal-Wallis and Dunn’s multiple comparison test, *p < 0.05 or indicated. SD and mean are indicated. Female (F), male (M).
AKT-dependent signaling is increased in the liver of regenerating females
We speculated that insulin signaling is regulating lipid turnover in axolotl hepatocytes analogous to our findings in FB cells of fruit flies. Thus, we predicted, based on observed lipid changes, to find profound adjustments of the insulin signaling cascade between regenerating females and males. To test our idea, we first identified the kinases AKT and annotated MAPK in the axolotl genome (Supplementary Data 6). AKT regulates the metabolic branch of the insulin cascade, whereas MAPK regulates the mitogenic pathway. To compare intact to regenerating animals we loaded equal amounts of protein (1 mg/ml) on a Western blot, and confirmed loading with a corresponding protein stain (Supplementary Fig. 4a’, b’). We show that intact female axolotls express a low baseline of AKT protein when compared to male siblings (Supplementary Fig. 4a). However, in regenerating females we observed a dramatic increase of AKT levels. Conversely, MAPK is expressed in females and males, but MAPK protein levels remain unchanged in 6 dpa hepatocytes with respect to intact animals (Supplementary Fig. 4b). This protein detection is not specific to the activated form of the proteins, therefore, we proceeded to investigate a downstream effector, Foxo1. Our data shows that in intact female and male axolotls, Foxo1 expression correlates to the tendency of higher AKT protein content in males. However, we did not detect significant mRNA changes in the injured group, indicating that during regeneration, Foxo1 expression does not report AKT activity or alternatively, AKT targets other effectors25 than FOXO (Supplementary Fig. 3d).
State of sterol content in limb blastema is regeneration specific
In the growing blastema, we found a significantly lower sterol: sterol ester ratio (driven by the increase of sterol esters) with respect to undamaged tissue. This sterol ratio is not replicated if animals had only a skeletal injury (fracture) instead of an amputation (Fig. 5a, b). To investigate if sterols are mobilized from tissues with high sterol content, we sampled dorsal and ventral muscle wall from regenerating animals and found no significant differences between groups (Supplementary Fig. 4c). Using existing data sets26, we investigated if lipid and sterol metabolism enzymes are correspondingly modified at the injury site. We found that sterol regulatory element-binding factor 2 (Srebf2) is upregulated at 24 h after amputation, and after a moderate down regulation, it remains above control levels up to 28 post amputation (Fig. 5c). This trend is mimicked by a target of SREBF2, the ATP citrate lyase (Acly), a fatty acid biosynthetic enzyme. Fatty acid synthase (Fasn), doubles its expression by 24 h after amputation, and at 3 dpa returns to control levels. In contrast, Srebf1 is also detected in the blastema, but remains unchanged during regeneration. Since Srebf expression can be induced by insulin growth factors (IGFs)27,28,29, we investigated if IGFs would change during the course of regeneration locally in the injury site. Using existing data sets26, we found that only Igf-2 is upregulated at the injury side in the hours prior to the Srebf peak (Fig. 5d). Taken together, we propose that axolotl blastema cells require a specific sterol homeostasis to drive efficient proliferation and differentiation. Further, we postulate that the sterol homeostasis in blastema cells is aided by systemic sterol transport regulated by hepatocytes, however, there is an activation of the fatty acid-based synthesis machinery at the injury site.
a Graphical depiction of the experimental set-up for (b). b Sterol/SE ratio in intact limb and regenerating blastema, or in limbs with a fracture. A.U. arbitrary unit. n = 8, Kruskal-Wallis and Dunn´s multiple comparison test, *p ≤ 0.05. SD and mean are indicated. Female (F), male (M). c Expression of Srebf1 and 2, Acly and Fasn in limb blastema during regeneration. Data obtained from Stewart et al.26. Counts per million (CPM) transcripts, hours post amputation (hpa), days post amputation (dpa). d Expression of Igf1-3 in limb blastema during regeneration. Data obtained from Stewart et al.26. Counts per million (CPM) transcripts, hours post amputation (hpa), days post amputation (dpa).
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.nature.com/articles/s41536-024-00375-x




