Introduction

Studies from the past few decades have shown the apparent role of ROS in influencing various biological processes^1–3. Generally, the sources of ROS are compartmentalized in co-occurrence with their targets mediating important signalling events⁴. Mitochondria, a major source of oxidant species^5,6, localize dynamically towards nucleus, effecting oxidation induced reshaping of gene expression profiles⁷. Localized ROS production are contributed by NADH oxidases distributed across different cellular regions⁸. Intracellular hydrogen peroxide gradients maintained by thioredoxin system allow intercompartmental exchanges among endoplasmic reticulum (ER), mitochondria and peroxisomes^9–11.

The process of redox relays is conserved and plays a fundamental role in self-renewal and differentiation of stem cell populations¹². Stem cells whether embryonic or adult maintain a low redox profile, characterized by subdued mitochondrial respiration^13–17. Levels of ROS are tightly regulated and elevated amounts promote early differentiation and atypical stem cell behaviour¹⁸. However, leading evidences suggest that many transcription factors require oxidative environment for maintenance of pluripotent states. For instance, physiological ROS levels play a crucial role in genome maintenance of embryonic stem cells¹³. Multipotent hematopoietic progenitors and neural stem cells require relatively high baseline redox for their maintenance^14,16,19. Suppression of Nox system or mitochondrial ROS compromises the maintenance of these self-renewing populations and promote their differentiation or death¹⁸. These evidences point towards essentiality of a perfectly tuned redox state for balancing the pluripotent and differentiated states. But how the stem cells harmonize their redox potential by possessing a restrained oxidant system is not very clear.

We addressed this fundamental question in multi-stem cell lineage-based niche architecture in Drosophila testis. The testicular stem cell niche is composed of a central cluster of somatic cells called the hub which contacts eight to eleven GSCs arranged in a round array^20,21. A pair of cystic stem cells (CySCs) originating the somatic lineage cyst cells, enclose each GSC and make their independent connections with the hub and GSCs via adherens junctions^22–26. The hub cells secrete Unpaired family of cytokines that defines niche occupancy and self-renewal of both GSCs and CySCs through Jak-Stat signalling^27,28. CySCs also repress GSC differentiation by suppressing transcription of bag-of-marbles (Bam)^29,30. Concerted asymmetric division of both GSC and CySC is essential for proper cyst formation³¹. A developing cyst contains a dividing and differentiating gonialblast encircled by cyst cells that provide nourishment to developing spermatids^32,33. However, the maintenance of GSCs in the spatially controlled microenvironment is still not very clear.

We found the existence of a perfectly balanced redox differential between GSC and CySC essential for niche homeostasis. CySCs by virtue of their higher redox threshold and clustering of mitochondria at GSC-CySC interface, generate an intercellular redox gradient that maintain physiological ROS levels in GSCs. Alterations in CySC redox state affected self-renewal and differentiation propensities of the GSCs. Intercellular redox-imbalance induced niche disequilibrium through precocious differentiation of GSCs and deregulated CySC proliferation, due to activation of pro-proliferative transductions and loss of cell-cell contact signalling. Our results in multi-stem cell lineage-based germline system indicate a nuanced division of labour where a higher redox state in CySCs not only maintains their self-renewing processes but also non-autonomously foster GSCs maintenance.

Results

Redox state of GSC is maintained by asymmetric distribution of CySC mitochondria

The apical region of the Drosophila testis incorporates a cluster of differentiated cells the hub surrounded by GSCs in a rosette arrangement with each GSC being enclosed by two CySCs (Fig. 1A). The mitochondria morphology and distribution in this cluster was checked by immunostaining ATP5A subunit of mitochondrial ATPase. We observed fragmented mitochondria in Vasa⁺ GSCs as compared to a fused organization in Traffic jam (Tj)⁺ CySCs (Fig. 1B#, C, D). Distribution profiles of mitochondria showed its higher propensity to overlap with Tj, indicating CySCs to be a mitochondria dense region compared to GSC (Fig. 1F); and its mitochondria to be predominantly clustered near the GSC-CySC boundary, illustrated by intensity profile (Fig. S1A’’’) and X-Z and Y-Z sectioning (Fig. S1B-B’). The ATP5A labelling overlapped with TFAM-GFP, a mitochondrial transcription factor³⁴, further confirming the patterning observed between these two stem cell populations (Fig. S1C’).

Figure 1.

Mitochondria localization near the GSC-CySC boundary, together with its dispersion pattern was in accordance with the intensity variance of ROS reporter line gstD1-GFP ³⁵ at the niche (Fig. 1H and S1D-D’). The redox gradient intensifies at the Tj-Vasa boundary and then gradually diffuses into Vasa⁺ GSCs, suggesting an establishment of GSC redox state through CySCs (Fig. 1Q; left panel). A GSC intrinsic oxidative environment should have resulted in a constant continuum of gstD1-GFP intensity across the Vasa-stained structure. We confirmed the greater ROS gradient at the GSC-CySC interface by 2D-Plot profile of Vasa, Tj and gstD1-GFP across the different cell types (Fig. S1E).

To affirm the role of CySC in influencing GSC redox state, we asked if disrupting dismutases in CySCs would influence the redox state of GSCs. We depleted the expression of superoxide dismutase 1 (Sod1) in CySCs and early differentiating cyst cells using Tj-Gal4 driver (Tj>Sod1i). Higher CySC redox state resulted in corresponding increase in GSC ROS profile, as indicated by a greater degree gstD1-GFP intensity and DHE fluorescence at the Vasa-gstD1GFP (VՈG) intersection (Fig. 1K, P, R and S1F), suggesting non-autonomous maintenance of GSC redox gradient by CySCs (Fig. 1Q; right panel). The surge in GSC ROS levels upon lowering of antioxidant defences in CySCs, was additionally confirmed through enhanced Vasa:gstD1-GFP colocalization intensities in Sod1i populations (MCC^mean = 0.482) with respect of controls (MCC^mean = 0.276) (Fig. 1S). These observations hinted towards a potential role of CySCs and its GSC polarized mitochondria in elevating the cellular ROS gradient at the interface between two stem cell populations.

Balanced CySC redox profile is crucial for its proliferation and GSC maintenance

Along with the shift in the redox gradient, we observed that high ROS in Tj>Sod1i had a striking effect on the increase in DAPI⁺-nuclei at the tip of testis (Fig. 1G and L). This overcrowding was attributed to an increase in number and positional shift in Tj⁺ CySCs and early differentiating cyst cells (Fig. 2A’, B’ and E). The enhancement in the number of CySCs was also reflected by ∼2-fold change in Zfh1⁺ cells (Fig. 2F). Vasa⁺ cells flanking the hub however, substantially reduced in number (Fig. 2C’’-D’’ and G), which was validated through lower detectable levels of Vasa even when its expression per cell remained unchanged (Fig. S2O). The observations were further verified using Sod1i localized in different chromosome to avoid any chromosome or balancer-based biasness (Fig. S1G-L). However, the phenotypic effect of higher ROS was limited to its origin in CySCs only because ablation of GSC redox status did not cause any marked change in niche composition. Nos-Gal4 driven knock-down of Sod1 in GSCs did not result in significant change in CySC number (Fig. S2A’, B’ and E) but effected a considerable reduction on Vasa⁺ cells (Fig. S2C’, D’ and F), aligning with previous observations¹⁷. Although, Sod1 is the predominant enzyme which also localizes in the intermembrane space of mitochondria, we observed a similar result upon depleting the matrix localized Sod2 in both GSCs and CySCs (Fig. S2G-M), indicating involvement of mitochondrial ROS for the observed cellular phenotypes.

Figure 2.

To confirm active proliferation of cells and rule out the possibility of arrested growth, we utilized fly-Fluorescent Ubiquitination-based Cell Cycle Indicator (FUCCI) system³⁶ which comprises of two reporter constructs marking G1/S transition (green), S (red) and G2/early mitosis (yellow) (Fig. 2H). Tj>Sod1i testes presented a ∼2-fold increase in cell number across all stages of cell-cycle (Fig 2I). The enhanced number of dividing cells was contributed mainly by progressive division of CySCs, showing >2-fold difference in accumulation of Zfh1⁺ nuclei in S and G2/M phases (Fig. 2J, S2PV-VII and QV-VII). The number of Zfh1 labelled cells in G1/S transition was also substantially more (Fig. S2PIII and QIII), indicating dysregulated redox ensued increased mitotic index of CySCs. Downregulation of Sod1 in GSCs through Nos-Gal4 demonstrated no significant change in G1/S, S and G2/M phase numbers, confirming our previous observation that altering ROS levels in GSCs does not have any non-autonomous effect on CySCs (Fig. S2N).

CySC induced ROS couples precocious differentiation of stem cell lineages

Since, in Sod1i testis the number of Tj⁺ CySCs and early differentiating cells³⁷ were elevated and more than CySC-specific Zfh1⁺ cells (Fig. 2I and J), we sought to label these cells with another differentiation marker Eya (Eyes absent). We found that Eya⁺ cells clustered towards the hub along with a significant increase in their number in Sod1 depleted CySCs (Fig. 3A, B and D). The supposed premature expression of Eya resulted in a population of differentiating cyst precursor cells co-expressing Eya-Zfh1 (Fig. 3A’’, B’’ and D). This population was found to be deviating from the normal partitioning of wild type representative populations (Fig. 3C). On contrary, the same cells would not show a similar phenotypical arrangement when Sod1i depletion was driven in the germline lineage by Nos-Gal4 (Fig. S3A-D). These observations hint towards an obvious change in redox state of GSCs affecting their number, even when generated non-autonomously from CySCs.

Figure 3.

Together with reduction in Vasa⁺ GSCs shown earlier (Fig. 2G), we also detected gonialblast expressing differentiation-promoting factor Bam to be much closer to the hub in Tj>Sod1i testes (Fig. 3E’’ and F’’). The mean distance between the hub and Bam⁺ cells was found to be shortened by ∼7 microns (Fig 3G). The relative advancement of spermatogonial differentiation towards the hub was also ascertained by tracking the transformation of GSC-specific spectrosome into branched fusomes connecting the multi-cell gonialblast (Fig. 1A). The branching fusomes were found more proximal to hub than controls (Fig. 3H, I’’ and J’’), along with parallel reduction in number of spectrosomes, implying a decline in early-stage germ cells (Fig. S3F). This loss could possibly be attributed to a higher oxidizing environment in GSCs and loss of intercellular contacts, particularly with hub cells (Fig. 2C# and D#). The observation corroborated with decreased expression pattern of E-cadherin in Tj>Sod1i particularly near the hub (Fig. 3K-M and S3G) and could be one of the reasons for GSCs dissociation and differentiation (Fig. 2D#), due to absence of self-renewing signals. Indeed, disengagement of GSCs from the hub caused reduction in Socs36E and Ptp61F transcripts that act as drivers of E-cadherin expression and are induced by maintenance factor STAT originating from the hub (Fig. 3N and O). These results suggest that disruption of intercellular redox gradient effected premature differentiation of stem cells in the niche.

Disrupted niche architecture compromises GSC-CySC communication

GSCs and CySCs intercommunicate through redox dependent factors such as EGFR, PI3K/Tor to support CySC maintenance³⁸. EGF ligands secreted by GSCs maintain differences in cell polarity and segregates self-renewing from differentiating populations³⁹. Tj>Sod1i cells showed lower levels of cell-polarity marker Dlg (Fig. 4A”, B” and S4A), that expanded more into the differentiated zone (Fig. S4B). Since, Dlg expression is dependent on EGFR⁴⁰, impairment in its reception was reflected in net reduction in pErk levels across sections in Sod1i testis (Fig. S4C). Due to loss of contact, this reduction was mainly limited to regions surrounding the hub, in CySC zone (Fig. 4C’’-D’’ and E). Differentiated cyst cells presented relatively higher levels of pErk expression (Fig. 4E, S4D’), particularly in Tj>Sod1i tissues where the pErk area was significantly extended, mapping with that of Dlg (Fig. 4F) and located comparatively closure to the hub (Fig. S4D), supporting its redox-dependent enhancement to promote differentiation. The deregulated proliferation of cystic cells reconciled with higher levels of its self-renewing inducers PI3K effector, p4E-BP (Fig. 4G’, H’ and I) and Hedgehog (Hh) (Fig. 4M’ and N’) ⁴¹ in Tj>Sod1i testis. The noticeably high p4E-BP was associated with Cyclin-D accumulation (Fig. 4J). Similarly, depletion of Sod1 in CySC resulted in elevated expression of Hh receptor Patched (Ptc) (Fig. 4O) which extended to regions farther from the hub, overlapping with expanded Tj⁺ cells (Fig. 4M-N). Since, Ptc itself is a transcriptional Hh target, its expression in CySCs suggest active Hh signalling⁴², further supported by higher levels and similar pattern of Hh transcriptional effector Cubitus interruptus (Ci) in Sod1i testis (Fig. 4P and S4E’’’-F’’’). Since, Ptc, Ci and a GPCR-like signal transducer Smoothened (Smo) are all transcriptionally driven by Hh, and Hh itself is regulated by higher levels of Ci^43,44, we expectedly found higher levels of transcripts for these pathway genes (Fig. S4G-J), indicating ROS-mediated upregulation of Hh activity. The overt proliferation of CySCs under high ROS conditions was suppressed by reducing gene dosage either by single copy of PI3K^DN (dominant-negative allele) or Hh-RNAi (Fig. 4L and Fig. S4K-N) which was also associated with an equivalent maintenance of GSC population (Fig. 4K and S4O-R). Together these observations suggest that intercellular redox gradients balance the stem-cell populations in the niche through CySC maintenance factors.

Figure 4.

Elevating CySC antioxidant defence promotes GSC self-renewal

To further substantiate the role of ROS in coordinating the multi-lineage stem populations, we strengthened the cellular defences against oxidants by overexpressing Sod1 in CySCs and monitored the relative GSC/CySC populations. Indeed, we observed an increase in the number of Vasa⁺ cells (Fig. 5A and S5A’’-B’’), together with a slight reduction in Tj⁺ cells (Fig. 5B and S5E’’-F’’). Any morphological anomalies associated with cell-cell adhesion or abnormal cellular dispersion, was not observed (Fig. S5A-B and E-F). Enhancing the levels of Sod1 in GSCs promoted the growth of GSC-like cells (Fig. S5C’’, D’’ and I) but did not have any prominent effect on CySCs (Fig. S5G’’, H’’ and J). The increment in Vasa⁺ GSCs under CySC-induced low redox conditions was also represented by parallel increase in spectrosome number (Fig. 5C-E). The uptick in GSC number due to scavenging of ROS in CySC, might be a result of delayed differentiation as indicated by displacement of Bam⁺ zone away from hub; thus, confirming non-cell-autonomous role of CySC ROS in maintaining GSC fate (Fig. 5F-H). The data suggest that reducing redox profile of CySCs differentially affected their own self-renewing propensities along with GSC maintenance and optimum redox state of both the stem-cell populations is controlled by CySCs.

Figure 5.

Discussion

ROS is a crucial mediator of stem cell maintenance where these species act as second messengers to induce different post-translational protein modifications thereby, affecting cell fate^45,46. Effectuation of ROS is mainly considered to be autocrine in nature where generation of these oxidative species is compartmentalized with their site of action. Recent studies have reported non-autonomous generation of ROS by NOX enzymes or upon stimulation by growth factors, to play a critical role in regenerative growth^47–49. However, in either of the cases, the target cell serves as the source and consequent response. Our observations in multi-lineage Drosophila stem cell niche suggested a previously unaccounted transmission of redox signals inter-cellularly that act as an organizer of niche homoeostasis. The higher ROS threshold of CySCs and resultant CySC-GSC oxidative gradient maintained the physiological redox state of GSC. Reduced ROS in GSC was accompanied with its proliferation⁵⁰, phenocopied when Sod1 was overexpressed in its neighbours (Fig. 6).

Figure 6.

The GSCs are attached to the hub via adherens junction to receive the maintenance signals like, Upd that trigger the JAK-STAT pathway^27,28, suppression of which activates Bam, acting as a switch from transit amplifying cells to spermatocyte differentiation^51,52. E-cadherin levels ensure only undifferentiated stem cells to persist within the niche, while differentiating GSCs lose the competition for niche occupancy²⁵. Elevated ROS in CySC possibly affected STAT phosphorylation through S-glutathionylation^53,54, causing reduced presence of dependent transcripts, Socs36E and E-cadherin thereby, resulting in detachment of GSCs from the neighbours as under E-cadherin depleted conditions^55–58. This resulted in compromised GSC-CySC paracrine receptions such as EGFR which play a crucial role in maintenance of cell-polarity that segregates self-renewing CySC populations from the ones receiving differentiation cues ^39,59,60, leading to accumulation of cells co-expressing both stemness and differentiation markers. In contrast to GSCs, ROS imbalance in CySCs induced accelerated differentiation due to deregulation of PI3K/Tor and Hedgehog pathways. Although, redox modulation of PI3K pathway components has been previously demonstrated⁶¹, in this study we found Hedgehog to be probably susceptible to redox regulation.

However, we do not rule out the possibilities of hub cells playing an equivalent role in GSC maintenance, given their parallel effect in suppressing GSC differentiation²⁸. We had tried using E-132 driver to express Sod1i lines in hub cells but were challenged by substantial adverse effect on fly viability. We also avoided usage of Gal80 dependent clonal populations to maintain a homogenous genetic background and prevent false readouts due to diffusion of ROS signals in unaltered neighbourhood.

The proposed concept of intercellular ROS communication can be of importance in deciphering the biochemical adaptability and plasticity of different niches influencing stem cell fate, ensuring niche size and architecture to prevent stem cell loss and aging. This has been observed in neural stem cells where inflammation-induced quiescence recovers the regenerating capacity of aging brain⁶². In addition to metabolic variations, dependence of the germline on somatic neighbours for its redox state might be one the reasons behind its presumed immortal nature and renitence to aging⁶³. The same can also be applied to cancer stem cells which maintain niche occupancy and resist oxidative stress induced apoptosis probably, by receiving redox cues from the environment. However, further work is required to elucidate the myriad of driver mechanisms intersecting at the realm of redox regulation that extend beyond the present system into broader translational areas.

Materials and Methods Fly strains

Fly stocks were maintained and crosses were set at 25 °C on normal corn meal and yeast medium unless otherwise indicated. All fly stocks (BL-24493) UAS SOD1 RNAi, (BL-32909) UAS SOD1 RNAi, (BL-29389) UAS SOD1 RNAi, (BL-32983) UAS SOD2 RNAi, (BL-24754) UAS SOD1, (BL-32489) UAS hhRNAi, (BL-8288) UAS PI3K^DN, (BL-25751) UAS Dcr-nos, (BL-55122) UAS-FUCCI, were obtained from Bloomington Drosophila Stock Centre. Nos-Gal4, Tj-Gal4, Tj-Gal4-BamGFP, gstD1-GFP, TFAM-GFP were kind gift from U. Nongthomba, P. Majumder, K. Ray, B. C. Mandal, Hong Xu labs respectively. For optimal Gal4 activity parent crosses were shifted to 29 °C till the eclosion of adults. 3-5 days old males were taken for all the experiments. The control lines for all experiments are the corresponding Gal4-line crossed with wild type OregonR⁺.

Dissection and immunostaining

Anesthetized flies were dissected in 1X Phosphate Buffer Saline (PBS). All incubations were carried out at room temperature (25 °C) unless otherwise mentioned. Testes were fixed in 4% paraformaldehyde for 30 minutes, followed by multiple washes with 0.1% PBTX (PBS + TritonX 100). Post-blocking in 0.5% Bovine Serum Albumin, testis was incubated overnight at 4°C for primary antisera, followed by washing in 0.1% PBTX 3 times 15 minutes each before incubating with secondary antibody. The tissues were counterstained with DAPI (4,6-diamidino-2-phenylindole) for 20 minutes, followed by three washes in 0.1% PBTX 10 minutes each. For PTC staining, a modified protocol utilizing PIPES-EGTA buffer was used as described⁶⁴. The dpErk was labelled by dissecting fly testes in Schneider’s media, followed by fixation in 4% PFA for 30 min, 3x wash in 0.1% PBTX, blocked and incubated in primary antibody overnight; each step supplemented with phosphatase inhibitor cocktail 2 (1:100, Sigma, cat#P5726). Samples were mounted in DABCO anti-fade medium prior imaging.

The following primary antibodies were used in the experiments – anti-FasIII (7G10) (1:120, DSHB (Developmental Studies Hybridoma Bank)), anti-Eya (1:50, DSHB), anti-DEcad (DCAD2) (1:50, DSHB), anti-α-Spectrin (3A9) (1:20, DSHB), anti-Ptc (1:100,DSHB), anti-Ci (1:50, DSHB), anti-βtubulin (1:300, DSHB), anti-pERK (1:100, Cell Signaling (4370)), ATP5A (1:700, Abcam (ab14748)), anti-Tj (1:5000), anti-Vasa (1:4000), anti-Zfh1(1:2000). The primary labelling was detected using appropriate Alexa-Flour tagged secondary antibody (ThermoFisher Scientific).

To assess superoxide levels, testes were dissected in Schneider’s medium (SM) and immediately incubated in 30 µM Dihydroxyethidium (DHE) at 25 °C in dark. Testes were washed thrice with SM for 7 minutes each before quantifying the emitted fluorescence using SpectraMax iD5 (Molecular Devices).

Quantitative Reverse transcription–PCR

For semi-quantitative RT-PCR and qRT-PCR analyses, total RNA was isolated from testes of 3-5 days old male flies using Qiagen RNA extraction kit. RNA pellets were resuspended in nuclease-free water and quantity of RNA was spectrophotometrically estimated. First strand cDNA was synthesized from 1-2 µg of total RNA as described earlier. The prepared cDNAs were subjected to real time PCR using forward and reverse primer pairs as listed below, using 5 µl qPCR Master Mix (SYBR Green, ThermoFisher Scientific), 2 pmol/µl of each primer per reaction for 10 µl final volume in ABI 7500 Real time PCR machine. The fold change in expression was calculated through 2^-ΔΔCt method. The primers used are listed in Supplementary Table S1.

Imaging and image analysis

Confocal imaging was carried out using Zeiss LSM 900 confocal microscope, using appropriate dichroics and filters. Images were further analyzed and processed for brightness and contrast adjustments using ImageJ (Fiji). Mean intensity measurements were carried out using standard Fiji plug-ins, 2D surface plot quantification or Zeiss Zen software as indicated. Quantification of cell number/nuclei in single or co-stained populations (Tj, Vasa, Zfh1, Eya, FUCCI) were performed from individual slices of z-stacks using semi-automated Cell Counter plugin in Fiji for all the mentioned genotypes. The relative distance from the two reference points in the images was estimated through Inter-edge Distance ImageJ Macro v2.0 from Github. All images were assembled using Adobe Photoshop version 21.2.1.

Immunoblot Analysis

Protein was extracted from the dissected testes using RIPA buffer as described previously and quantified using Bradford reagent (Biorad). Equivalent concentration of lysate was denatured using 1X Laemelli Buffer with 1 M DTT at 95°C, separated in 10% SDS-PAGE and transferred onto PVDF membranes. Membranes were blocked with commercial blocking solution in TBST base before sequential primary antibody incubation with anti-Vasa and anti-β-Tubulin (DSHB, E7). Secondary detection was performed using HRP-tag anti-mouse antibody (GE Amersham).

Statistical analyses

All data points represent at least three biological replicates and the sample sizes carrying adequate statistical power are mentioned in the figure legends. Statistical significance for each experiment were calculated using two tailed Student’s t-test unless otherwise mentioned, using GraphPad Prism 8 software. The P values were calculated through pairwise comparison of the data with wild type or driver alone and driven RNAi lines. Significance values for sample sizes mentioned in figure legends were represented as *P < 0.01, **P <0.001, or ***P < 0.0001.

Author contribution

D.S. and O.M conceived the study and designed the experiments. O.M, A.C and T.C performed the experiments. D.S and O.M analyzed the data. D.S, O.M. and A.C. wrote the manuscript.

Acknowledgements

We thank P. Majumder, U. Nongthomba, K. Ray, G. Ratnaparkhi, B. C. Mandal, Hong Xu lab and S. C. Lakhotia for kindly sharing some of the transgenic fly lines. Christian Bökel for sharing experimental protocols. Anti-Tj antibody was a generous gift from Prof. D. Godt, University of Toronto. Anti-Vasa and Anti-Zfh1 were kindly gifted by Prof. R. Lehmann, New York University. We also thank Dr. Bama Charan Mandal for generously sharing his Confocal microscope imaging system and Sudeshna Majumder for preliminary data. We acknowledge BDSC for fly stocks and DSHB for antibodies. This work is supported by Innovative Young Biotechnologist Award by Department of Biotechnology (BT/12/IYBA/2019/01), Early Career Research Award by Science and Engineering Research Board (ECR/2018/000009), BHU-Institute of Eminence grant (R/Dev/IoE/Incentive/2021-22/32452). The authors also acknowledge financial support from DST-INSPIRE fellowship (to O.M.), UGC-CAS doctoral fellowship (to T.C.).

Declaration of interest

The authors declare no competing interests

Data availability

The data that support the findings of this study are available within the main text and its Supplementary Information file. The lead contact will share all raw data associated with this paper upon request. This study did not utilize or generate any unique datasets or codes.