MeeY and YybP, two proteins regulated by manganese-sensing riboswitches, are required for Bacillus subtilis biofilm formation Authors: Ankita J Sachla, Miguel Piñeros, John D Helmann Published: 2026 ABSTRACT AbstractIn the rhizosphere, Bacillus subtilis forms biofilms on plant roots and contributes to improved plant growth and disease resistance. Manganese stimulates biofilm formation in B. subtilis 3610, but the underlying mechanism is unknown. Here, we show that YybP and MeeY (formerly YkoY), two proteins regulated by the eponymous manganese-sensing yybP-ykoY riboswitches, are required for biofilm formation. MeeY is a TerC family membrane protein implicated in manganese export and YybP is of unknown function. Inactivation of either meeY or yybP leads to reduced synthesis of matrix-associated exopolysaccharide and a loss of biofilm-associated divalent cations.Deletion of sinR restores exopolysaccharide but does not restore normal metalation of the biofilm matrix. We conclude that the YybP and MeeY membrane proteins are previously unappreciated regulators of biofilm formation important for the production and metalation of exopolysaccharide.Subject terms: Biotechnology, Microbiology Introduction Biofilm formation is important for the ability of Bacillus subtilis and related plant-growth promoting rhizobacteria (PGPR) to function as biocontrol agents1,2. B. subtilis NCIB 3610 (hereafter, 3610) is an important model system for studies of biofilm formation in the Bacilli3,4. Strain 36105 is ancestral to the B. subtilis 168 type strain6, the model organism for this species7, and displays complex multicellular behaviors not seen in the 168 strain. These include swarming motility and the ability to form well-differentiated biofilms4,8. B. subtilis 168 strains fail to form complex biofilms due to four mutations (affecting sfp, epsC, swrA, and degQ) and the loss of a plasmid-associated phosphatase (rapP)9. B. subtilis biofilms are complex, differentiated communities with several distinct cell types exhibiting spatial heterogeneity4. Biofilm cells are enclosed within a self-produced extracellular matrix (ECM) comprising TasA and TapA protein fibers, extracellular DNA, γ-polyglutamic acid, and exopolysaccharides (EPS). A surface layer containing the non-polar hydrophobin BslA protects the growing structure from solvents, heat, desiccation, water, and selectively conducts gas exchange10. Mature biofilms have a highly rugose, wrinkled appearance that forms over several days of incubation. The most common model systems for analysis of Bacillus biofilms are the formation of a complex, structured colony on agar plates (colony biofilm) and the formation of a dense, wrinkled mat on the surface of liquid medium (a pellicle biofilm)4. Two parallel regulatory pathways control the transcriptional repressors that regulate biofilm matrix production11. The biofilm master regulator (SinR) and the transition state regulator (AbrB) bind to the operator regions of biofilm matrix operons3,12. The initiation of biofilm formation is governed by the phosphorylation of the transition state regulator Spo0A in response to declining nutrient availability and decreasing growth rate3,13,14. Multiple histidine-protein kinases phosphorylate Spo0A with KinC and KinD active during biofilm formation15–18. Phosphorylated Spo0A (Spo0A-P) activates the SinI-SinR-SlrR regulatory network that regulates the transition from a motile, planktonic lifestyle to biofilm formation19,20. The SinI-SinR-SlrR regulatory network is centered on the tetrameric SinR repressor and two antagonist proteins, SinI and SlrR13. As Spo0A-P levels rise, increased expression of SinI leads to sequestration of SinR in an inactive SinI:SinR heterodimer19,21,22. As SinR-mediated repression abates, the second SinR antagonist SlrR is produced and forms SinR2·SlrR2 heterotetramers. The SinR2·SlrR2 complex is unable to repress slrR expression, leading to a positive feedback loop that reinforces the switch from a genetic program supportive of planktonic growth to one that generates biofilms. Inactivation of SinR allows the RemA activator to increase expression of the epsA-O and tapA-sipW-tasA operons that encode the ECM12,23,24. In parallel with the upregulation of genes associated with the biofilm lifestyle, the SinR2·SlrR2 complex also reduces motility by downregulating autolysins, flagellin, and SigD20,25. Collectively, these processes contribute to phenotypic bistability resulting in matrix expression by a subpopulation of cells in the biofilm20,26. Several interrelated effects contribute to phenotypic heterogeneity, including stochastic variations in Spo0A activation27 and SinI levels28, changes in growth rate17,29, and chromosome position-dependent gene dosage effects30. As levels of Spo0A-P increase further, matrix genes are repressed, and sporulation is induced29,31. Robust biofilm formation is strongly stimulated by the combination of manganese ion (Mn) and glycerol15. Sensing of glycerol requires the extracellular CACHE domain of KinD15, which also senses malate, a common constituent of plant root exudate that supports soil microbiome and colonization of plant root18. However, the molecular basis for the effect of Mn on biofilm formation is still obscure. Previously, Mn supplementation was shown to elicit expression of epsA-O and other biofilm associated genes as well as the downregulation of motility functions32. The stimulatory effect of Mn on biofilm formation likely requires import into the cell. In B. amyloliquefaciens, Zn inhibits biofilm formation by blocking the MntABCD Mn transporter, and the inhibited cells have reduced expression of the epsA-O and tapA-sipW-tasA operons33. Elevated levels of Mn are known to have wide-ranging effects on gene expression34–36. In B. subtilis, Mn binds to MntR to regulate Mn importers and exporters37,38. In addition, the MeeY (formerly YkoY) and YybP proteins are regulated by yybP-ykoY type riboswitches39. These riboswitches are similar in proposed structure to the Lactococcus lactis yoaB Mn-sensing riboswitch that binds a single manganese (Mn) ion to stabilize an antitermination structure40. The functions of the two Mn riboswitch-regulated proteins in B. subtilis are still poorly defined. MeeY is a TerC family membrane protein41 and functions in Mn export in support of metalation of extracytoplasmic proteins42,43. YybP is a membrane-associated lipoprotein with a predicted RNA-binding domain of the AmiR and NasR transcription antitermination regulators (ANTAR) family44, but is of unknown function. Here, we show that MeeY and YybP are both critical for the normal development of B. subtilis 3610 biofilms. Mutant strains lacking either protein have reduced EPS, are compromised in surface-associated hydrophobicity, and fail to develop mature biofilms. Pellicle formation and the hydrophobicity of colony biofilms can be largely corrected by mutations in sinR or by overexpression of slrR. Although loss of SinR restores EPS production, it does not restore normal metalation of the biofilm with Mn and calcium (Ca) ions. These results suggest that MeeY and YybP reinforce biofilm maturation by supporting both the synthesis of EPS and its metalation with divalent cations. Results ΔmeeY and ∆yybP cells are defective in colony biofilm morphology The two most widely adopted media for studies of biofilm formation are LBGM (LB medium amended with 1% glycerol and 100 μM Mn)15 and a minimal medium with glutamate and glycerol (MSgg) containing 50 μM Mn8. Since TerC family proteins are linked to Mn transport42,43, and the biofilm ECM accumulates divalent cations45,46, we set out to determine if TerC family proteins in B. subtilis 3610 (MeeY, MeeF, and YjbE) are important for colony biofilm formation. We grew B. subtilis 3610 and isogenic meeF, meeY, and yjbE mutants on LBGM agar plates and monitored biofilm formation over 5 days. As expected, B. subtilis 3610 developed highly rugose colony biofilms. Biofilm formation was grossly perturbed only in ∆meeY, which grew as a featureless colony that spread over the plate surface, similar to B. subtilis 168 (Fig. 1A). We conclude that MeeY is unique among the three TerC family proteins in being required for biofilm formation. MeeY and YybP are essential for biofilm formation. A 4 µl of mid-log cells were spotted onto freshly prepared LBGM plates and B onto defined MSgg agar. All plates were incubated at 30 °C and were imaged periodically. Like MeeY, expression of YybP is regulated by a Mn-sensing riboswitch. The ∆yybP mutant was also defective in biofilm formation and formed featureless colony biofilms similar to ∆meeY on LBGM (Fig. 1A). A ∆meeY ∆yybP (YP) double mutant also formed featureless colony biofilms, albeit with less cell spreading (Fig. 1A). The biofilm defects of the ∆meeY and ∆yybP strains were also apparent from the decreased biomass of the resulting colony biofilms (Fig. S1). Similarly, on MSgg medium both growth and biofilm formation were greatly reduced for ∆meeY, ∆yybP, and YP (Fig. 1B, Fig. S2). In a previous analysis of TerC family proteins, we found that a B. subtilis 168 ∆meeF ∆meeY double mutant (but not the single mutants) grew poorly on LB due to defects in the expression of feeding proteases and the metalation of extracytoplasmic enzymes42. Here, we tested the effects of ∆meeY and ∆yybP deletions on protease production in B. subtilis 3610. Consistent with the results in strain 168, the 3610 ∆meeF ∆meeY double mutant grew poorly on LB agar and was severely defective in protease production (Fig. S3). In contrast, the ∆meeY and ∆yybP single mutants, as well as the YP double mutant, still secreted proteases, albeit at reduced levels (Fig. S3C). Further, we observed no difference in export of the model secretion substrate AmyQ-His for WT, in the single ∆meeY and ∆yybP mutants, or a YP double mutant (Fig. S3D). Thus, the biofilm defects of the ∆meeY and ∆yybP mutants are not correlated with large defects in protein secretion. ΔmeeY and ∆yybP mutants are defective in forming pellicle biofilms We next tested the ability of ∆meeY and ∆yybP to form pellicles on LBGM liquid medium (Fig. 2). The ∆meeY strain developed only a thin, fragile, and easily fractured pellicle and the ∆yybP strain lacked a well-differentiated pellicle (Fig. 2, Fig. S4). Both the ∆meeY and the ∆yybP mutations can be complemented by expression of the missing gene from an ectopic locus (Fig. S5). Remarkably, we also observed cross-complementation where expression of MeeY can increase pellicle formation in ∆yybP, and expression of YybP can increase pellicle formation in ∆meeY (Fig. S5 vs. Fig. 2). However, efficient cross-complementation was not seen with colony biofilms (Fig. S5). This suggests that, at least in pellicle biofilms, induction of MeeY or YybP can compensate for the lack of the other protein. The meeY and yybP deletion mutants exhibited defective pellicle biofilm formation in LBGM broth. For pellicle development, cells were incubated in 2.5 ml of medium at 30 °C under static conditions and imaged at the indicated time points. ∆meeY and ∆yybP mutants are defective in biofilm hydrophobicity The formation of a 3610 biofilm depends on the production of an elaborate ECM comprising EPS, the TapA-TasA proteins, and an interfacially active protein (BslA) that confers surface hydrophobicity4. We used a droplet test to assess the surface hydrophobicity of colony biofilms (Fig. 3A), a property attributed to the BslA hydrophobin10,47. As expected, we saw a typical bead-like behavior for a water droplet spotted on mature WT biofilms and a similar behavior for mutants lacking either of the two TerC family proteins (ΔmeeF and ΔyjbE) not implicated in biofilm formation (Fig. 3B, C). In these cases, the contact angle between the water droplet and biofilm surface was always greater than 90° (~125°) (Fig. 3C, D). In contrast, the surface of the ΔmeeY and ∆yybP colonies was much more hydrophilic with a mean contact angle of less than 30° (Fig. 3C, D). The surface of the YP double mutant had further reduced hydrophobicity, but not to the extent seen in B. subtilis 168 (Fig. 3C). Since surface hydrophobicity depends on secretion of BslA, we also tested the ∆meeF ∆meeY (FY) mutant, which is defective in protein secretion in both strains 16842 and 3610 (Fig. S3). The FY mutant was severely defective in surface hydrophobicity (Fig. 3D) and absorbed the droplet, similar to the 168 strain. MeeY and YybP are important for surface hydrophobicity. A 10 µl of saffranin solution was air-dropped with a pipette onto 5-day biofilms on LBGM medium. B A top view and C a side view were imaged. D The contact angle was measured using the Adobe Photoshop ruler tool on images gathered from multiple biological replicates. ∆meeY and ∆yybP mutants are defective in sporulation Both biofilm formation and sporulation depend on phosphorylation of Spo0A, a transcription factor that governs several distinct developmental processes in Bacillus13,31,48. In response to nutrient depletion, low levels of Spo0A-P lead to the expression of cannibalism toxins and intermediate levels to biofilm formation (Fig. 4). If nutrient depletion persists, higher levels of Spo0A-P trigger sporulation. In mature 3610 biofilms, spores are enriched in aerial structures known as fruiting bodies8. MeeY and YybP support biofilm-associated spore formation. A 2xSG media plate was used to demonstrate physical difference (rugose vs. smooth appearance) in strains lacking spore formation. B Enumeration of colony-forming unit (CFU) for cells treated with and without heating for estimation of spore numbers. C A rudimentary depiction of Spo0A and the processes it affects based on its ~P threshold levels. High Spo0A~P induces sporulation genes including (including spoIIGA)31,49. The PspoIIGA-luciferase fusions were tested in the following B. subtilis 3610 strains: D WT, E meeY::kan, and F yybP::kan at 30 °C under mild shaking condition in different media (buffered LB, LBM, LBG, and LBGM) for 30 h. The in vivo, time-resolved luminescence was normalized to OD600, n = 3. Consistent with prior work8, we observed an abundance of heat-resistant spores in 3610 biofilms. However, spore levels were greatly reduced in the ∆meeY and ∆yybP strains (Fig. 4 and Table 1). To determine if this was due to a lack of sufficient Spo0A-P to drive sporulation gene expression, we monitored expression of the early sporulation gene spoIIGA during growth on LBGM49 (Fig. 4D). While a PspoIIGA-luciferase fusion was well expressed in LBGM colony biofilms, it was not expressed in the ∆meeY and ∆yybP mutants (Fig. 4D, E, and F). These results suggest that the ∆meeY and ∆yybP mutants fail to accumulate the high levels of Spo0A-P needed for sporulation. Table 1Biofilm characteristics among various strainsBiofilm featuresWT3610meeFmeeYsrfAAyybPyjbEmeeY yybP (YP)Colony Biofilm FormationYesYesNoPartialNoYesNoPellicle Biofilm FormationYesYesNoN/ANoYesNoHydrophobicity Angle (degrees)a123°±4°119°±428°±8°N/A30°±9°120°±3°24°±13°Sporulationa75%100%0.03%N/A0.003%66%0.12%LC-MS and spectroscopic analysisBioactive compounds (AU)WT3610meeFmeeYsrfAAyybPyjbEBlankSurfactin A2.98E + 066.25E + 061.10E + 043.19E + 031.13E + 063.91E + 079.05E + 02Surfactin B3.93E + 067.29E + 062.07E + 046.01E + 031.20E + 065.04E + 078.61E + 02Surfactin C6.73E + 061.11E + 076.24E + 041.86E + 045.17E + 069.65E + 072.27E + 03Surfactin D1.76E + 053.04E + 051.20E + 03N.D3.86E + 042.07E + 06N.D.Pulcherriminic acid100%60%2%N/A2%76%N/AAU Arbitrary units were used to express the relative peak area of lipopeptides detected by LC–MS.N.D non-detectable refers to the measurements below the limit of detection.N/A not applicable as strains were not included in the analysis.Sporulation was calculated as a percentage in relation to non-heated cells from the same sample.aObservations with colony biofilm. Biofilm characteristics among various strains AU Arbitrary units were used to express the relative peak area of lipopeptides detected by LC–MS. N.D non-detectable refers to the measurements below the limit of detection. N/A not applicable as strains were not included in the analysis. Sporulation was calculated as a percentage in relation to non-heated cells from the same sample. aObservations with colony biofilm. ∆meeY and ∆yybP mutants lack pulcherrimin and surfactin Biofilm formation is accompanied by the production of secondary metabolites, including pulcherrimin and surfactin. Pulcherrimin is an iron-chelating pigment that protects growing biofilms against oxidative stress50–52. Pulcherrimin was visible from day 1 as a reddish-brown pigment emanating from colony biofilms of 3610, ΔmeeF, and ∆yjbE strains and was also seen in the liquid phase of standing cultures used to test pellicle formation (Figs. 1–3). In contrast, the ∆meeY and ∆yybP strains (and the YP mutant) lacked pigmentation, with an estimated >98% reduction in pulcherrimin levels in the pellicle supernatant (Table 1). Pellicle cultures of the ∆meeY and ∆yybP strains were also defective in the production of surfactin, with levels reduced between 50 and 100-fold relative to 3610 (Table 1). This is consistent with our prior report that CU1065 ∆meeY mutants are defective in surface motility due to reduced surfactin production53. Further, supplementation with surfactin (Fig. S6) or with supernatant from 3610 cultures (Fig. S7) did not improve biofilm maturation of ∆meeY under our conditions. Thus, a lack of surfactin is unlikely to be the sole cause of the observed biofilm defect. Indeed, the role of surfactin in biofilm formation has been controversial54,55. Exogenous metals fail to restore biofilm formation in ∆meeY or ∆yybP mutants The ECM is known to associate with divalent cations, with Ca in particular proposed to contribute to biofilm structure46. Since MeeY and related TerC family proteins have been implicated in Mn export, we hypothesized that perhaps ∆meeY and ∆yybP mutants were deficient in the metalation of the ECM. Therefore, we tested whether amending LBGM with metals could help restore more normal colony biofilm growth. Amendment of LBGM medium with additional Mn (Fig. S8), Fe (Fig. S9), or Mg (Fig. S10) had little effect on the appearance of colony biofilms. In contrast to LBGM alone, amendment with Ca led to a reduction in surface spreading, as previously reported56. However, Ca supplementation did not restore rugosity to biofilms formed by the ∆meeY, ∆yybP, or YP mutants (Fig. S11). These findings suggest that simply amending the medium with metal ions is not sufficient to rescue biofilm formation in cells lacking MeeY and YybP. ΔmeeY and ∆yybP mutants have reduced epsA-O expression Biofilm formation is the culmination of a complex regulatory cascade initiated by the activation of membrane-associated sensor kinases (principally KinC and KinD) that lead through a phosphorelay to a rise in Spo0A-P (Fig. 5A). We used the biofilm-specific transcriptional reporter fusions PepsA-GFP and PtapA-YFP to evaluate the impact of meeY and yybP mutations on expression of the promoters for the ECM-associated EPS (epsA-O) and protein fiber (tapA-sipW-tasA) operons. In the 3610 parental strain, PtapA-YFP and PepsA-GFP expression was initiated from 24 h. However, both tapA and eps expression were delayed in the ∆meeY and ∆yybP strains, with the expression of eps drastically reduced at all growth stages (Fig. 5B). In addition, the expression of eps was lost over time in the center of the biofilm in 3610, but this effect was not as apparent in the ∆meeY and ∆yybP strains (Fig. 5B). We also tested a srfAA-GFP transcriptional fusion, but this fusion was poorly expressed under our conditions (Fig. 5B). eps operon expression is reduced in ΔyybP and ∆meeY strains. A A simple schematic of the phosphorelay cascade for Spo0A~P accumulation and downstream derepression of SinR from eps and tap operators for biofilm formation. B 4 µl of cultures were spotted onto freshly prepared LBGM plates and plates were incubated at 30 °C and fluorescence was imaged periodically. C Monocultures (wells outside of the pink box) and co-cultures (wells inside the pink box) were performed in LBGM broth. For co-culture 1:1 cell volume of different strains were inoculated and pictures were taken after 3 days at 30 °C. The wells inside the yellow box show the cooperative establishment of biofilm by two strains in co-culture, which failed to show a high degree of reticulation in monoculture. At the cellular level, the loss of eps gene expression in ∆meeY and ∆yybP was confirmed by microscopic evaluation using the PepsA-GFP reporter. As expected, expression of ECM-related genes was preferentially observed in the population of chained cells3,20. Whereas eps expression was compromised, we still observed a signal for the PtapA-YFP reporter in ∆meeY and ∆yybP under biofilm conditions (Fig. S12). Consistent with the strong defect noted for expression of the PepsA-GFP reporter, the ∆meeY and ∆yybP strains are phenotypically similar to ∆epsA-O mutants when grown on LBGM (Figs. 1A, S13). These strains display featureless colonies15 and easily fractured pellicles57. These results are consistent with the hypothesis that MeeY and YybP are required for optimal expression of epsA-O. Next, we tested the ability of co-cultures containing two different strains to form biofilms. Prior work has shown that mixtures of strains lacking ∆epsA-O and ∆tapA-sipW-tasA can extracellularly complement each other to form mature pellicles by sharing public goods to build a complex architecture57. Compared with strains grown in monocultures (Fig. 5C, top row), the reticulation observed in pellicle biofilms formed by ∆tasA and ∆tapA mutants was enhanced when mixed with either ∆meeY or ∆yybP, but not with the YP double mutant (Fig. 5C, wells in pink square). This supports the idea that ∆meeY or ∆yybP strains fail to produce EPS, which can be provided by the ∆tasA and ∆tapA mutant strains. As expected, the ∆meeY or ∆yybP strains do not complement the ∆epsA-O strain, nor do they complement each other (Fig. 5C). Biofilm defects of ∆meeY and ∆yybP are reduced in the absence of SinR SinR is a central repressor of biofilm gene expression during exponential growth11. Therefore, we introduced a sinR mutation into ∆meeY and ∆yybP. The resulting ∆sinR∆meeY and ∆sinR∆yybP double mutants had greatly increased surface hydrophobicity (Fig. 6A, bottom panel and quantified in Fig. 3D) and increased eps expression (Fig. 5B) compared to the ∆meeY and ∆yybP single mutants. However, compared with deletion of sinR alone, colony biofilms of both the ∆sinR ∆meeY and ∆sinR ∆yybP double mutants had an increased surface spreading and a loss of complex architecture (Fig. 6A). The ∆sinR ∆meeY and ∆sinR ∆yybP double mutants also formed more robust pellicles with complex reticulation in a 24-well plate pellicle assay (Fig. 6B). sinR deletion partially suppresses biofilm defects of meeY and yybP. A 4 µl of cultures were spotted onto freshly prepared LBGM plates and plates were incubated at 30 °C and images captured periodically. Top and side view panels at the bottom show the test of hydrophobicity. B 2.5 µl of cultures were inoculated in 2.5 ml LBGM broth and 24-well plates were incubated at 30 °C for 5 days. C slrR-epsA-O gene organization and its control by SinR are depicted. D A xylose-inducible copy of slrR was expressed in the ∆meeY and ∆yybP strains and colony biofilms were monitored on LBGM media supplemented with 1% xylose. The epsA-O operon is transcribed divergently from slrR, which encodes SlrR, a second SinR antagonist (Fig. 6C). Consistent with the rescue observed with deletion of SinR, ectopic overexpression of SlrR bypassed the biofilm-associated defects for both ∆meeY and ∆yybP with a dramatic increase in colony rugosity (Fig. 6D), colony biofilm mass (Fig. S1C), and a restoration of complex, reticulated pellicles (Fig. 6B). Biofilm defects of ∆meeY, but not ∆yybP, are rescued by induction of RemA We next tested the effects of other regulators that control some of the same operons regulated by SinR. AbrB is a transition state regulator that also plays a role in repression of ECM genes11. However, deletion of abrB had no effect on the structure and rugosity of ∆meeY or ∆yybP colonies grown on LBGM (Fig. 6A, B). SinR functions to block epsA-O expression by occluding RemA (regulator of extracellular matrix), a transcription activator that is necessary and sufficient for activation of EPS expression23,58. Intriguingly, expression of an IPTG-inducible copy of remA rescued the biofilm defects of the ∆meeY but not the ∆yybP strain in both colony biofilm (Fig. 7A) and pellicle formation assays (Fig. 7B). RemA overexpression partially suppresses biofilm defects of meeY. An IPTG-inducible copy of remA was expressed in the ∆meeY, ∆yybP, and YP strains and biofilm formation was tested on LBGM A agar or B broth supplemented with or without 0.5 mM IPTG in final concentrations. All plates were incubated at 30 °C for the designated time. LBGM induces meeY and yybP expression Biofilm formation in B. subtilis 3610 is strongly stimulated by Mn by still unknown mechanisms. The meeY and yybP genes are preceded by Mn-sensing riboswitches (Fig. 8A) related in predicted structure to the L. lactis riboswitch40. We therefore hypothesized that meeY and yybP would be up-regulated in Mn-rich, biofilm-conducive media. To monitor their expression, we introduced PmeeY- and PyybP-luciferase transcriptional fusions (Fig. 8A) into B. subtilis CU1065 (a 168 derivative) and B. subtilis 3610. The domesticated CU1065 cells (Fig. 8B, C) showed little induction compared with the 3610 strain (Fig. 8D, E). B. subtilis 3610 demonstrated a clear induction of meeY and yybP by ~12 h in the LBGM conditions that support biofilm formation, but not in media amended with Mn or glycerol alone. This induction was lost in biofilm defective strains with deletions of kinC or kinD (Fig. 8F-I). These results suggest that induction of meeY and yybP in 3610 under biofilm-conducive conditions requires both Mn and glycerol and the kinases that initiate the regulatory cascade that controls biofilm formation. PmeeY and PyybP respond to biofilm-inducing conditions. A The genetic context is shown for selected yybP-ykoY family riboswitches and open reading frames (yellow represents the Mn-sensing element). The Lactococcus lactis (L. lactis) riboswitch is upstream of yoaB encoding a calcium-transporting ATPase 1 and has been structurally and functionally characterized40. In B. subtilis, the eponymous yybP-ykoY family riboswitches are upstream of yybP and ykoY(meeY). YybP is a 148 amino acid lipoprotein, and MeeY is a TerC family Mn transporting and metalating protein42. The intergenic regions for yybP and meeY were cloned into pBS3K-lux for measurement of gene expression during growth at 30 °C under mild shaking condition in different media (buffered LB, LBM, LBG, and LBGM) for 30 h. The in vivo, time-resolved luminescence was normalized to optical density (OD600). The strains tested were: B B. subtilis CU1065 with PmeeY and C PyybP-luciferase fusions. The PmeeY and PyybP-luciferase fusions were tested in the following B. subtilis 3610 strains: D, E WT 3610 parent strain, F, G ∆kinD, H, I ∆kinC, J, K ∆sinR, L, M espA-O::tet, and N, O ΔtapA-sipW-tasA::spec. LBGM induction of meeY and yybP depends on ECM production A deletion of sinR leads to robust, wrinkled biofilms even under conditions that normally do not induce biofilm formation12. In a ΔsinR background, we observed a 2-fold increase in the maximal PmeeY -luciferase signal (Fig. 8J vs. D) and a greater than 100-fold increase in the PyybP signal (Fig. 8K vs. E). This is unlikely to be a direct effect of SinR, since ChIP-Seq studies did not reveal any SinR binding for either meeY or yybP promoter regions25. Since meeY or yybP expression is induced specifically in LBGM medium, which is known to support synthesis of the ECM, we hypothesized that ECM synthesis might be needed for induction. Indeed, strains carrying either epsA-O::tet or ∆tapA-sipW-tasA deletions, and therefore unable to synthesize ECM, no longer displayed typical biofilm (Fig. S13) and the PmeeY and PyybP luciferase fusions were no longer induced in LBGM (Fig. 8L-O). We conclude that biofilm-related signals, and in particular ECM synthesis, are needed for induction of meeY and yybP. Conversely, MeeY and YybP are required for epsA-O expression, which is suggestive of a positive feedback loop. Biofilms formed by ∆meeY and ∆yybP mutants have reduced levels of Mn and Ca We next used ICP-OES to quantify total biofilm-associated metal ions in 5 day colony biofilms grown on LBGM agar (Table 2; Fig. S14). The ∆meeY, ∆yybP, and YP mutants exhibited reduced levels of Mn and Ca compared with WT biofilms. There was also a small decrease in Mg, but no significant decrease for K, Fe, or Zn. The level of metal ions noted in these colony biofilms is much higher than typical for planktonic cells59 and is likely associated with the binding of metal ions to the ECM. Consistently, metal accumulation in colony biofilms was decreased in epsA-O mutants, but not in tasA-sipW-tapA mutants (Table 2). The sinR mutant, which overproduces EPS12, had a milder decrease in Mn and Ca associated with the biofilms (Table 2). These results suggest that it is the EPS that is primarily responsible for metal accumulation, but that this process may be impaired in the sinR mutant, where EPS is overproduced. Table 2Metal content among colony biofilms grown on LBGMICP-OES MS analysisMetal (μg/g)WT3610meeYyybPYPepsA-OtapA-sipW-tasAsinRMn370 ± 130163 ± 5672 ± 30134 ± 7053 ± 14290 ± 76190 ± 8Ca3746 ± 1216906 ± 248463 ± 211870 ± 423504 ± 1312890 ± 7652260 ± 157Zn36 ± 1231 ± 537 ± 1464 ± 3410 ± 234 ± 922 ± 1Mg405 ± 143138 ± 48163 ± 76160 ± 68214 ± 38233 ± 60216 ± 19Fe13 ± 313 ± 1210 ± 524 ± 1710 ± 316 ± 47 ± 1K4570 ± 17304808 ± 14335347 ± 30708720 ± 25305357 ± 12954280 ± 8922016 ± 155 Metal content among colony biofilms grown on LBGM To determine if ECM-associated metal accumulation is correlated with EPS production, we turned to pellicle biofilms, which are easier to isolate from the growth medium than colony biofilms. In early pellicle biofilms, both Mn and Ca were substantially reduced in meeY and yybP mutants to levels compared to those seen in epsA-O mutants (Table 3 and Fig. S15). In contrast, Mg and potassium (K) levels were relatively stable across strains. Table 3Metal content among pellicle biofilms grown on LBGMICP-OES MS analysisMetal (μg/g)WT3610meeYyybPmeeY sinRyybP sinRepsA-OsinRMn27 ± 19.4 ± 34.4 ± 2.36 ± 1.52.6 ± 0.32.8 ± 1.65 ± 2Ca161 ± 846 ± 1622 ± 734 ± 816 ± 2.526 ± 1439 ± 15Zn8.4 ± 1.33.4 ± 0.42 ± 0.42.25 ± 0.71.8 ± 0.23 ± 1.42 ± 0.7Mg80 ± 388 ± 3062 ± 1660 ± 1871 ± 716 ± 561 ± 24Fe5.3 ± 0.86 ± 0.92.4 ± 1.32 ± 0.44.6 ± 53.7 ± 22.3 ± 1.2K5089 ± 20275292 ± 22083939 ± 12053811 ± 17404663 ± 5284844 ± 16423969 ± 1131 Metal content among pellicle biofilms grown on LBGM Next, we asked whether a sinR mutation would restore EPS production and metalation in meeY and yybP mutants. The EPS layer can be visualized using negative staining with India ink as visualized by differential interference microscopy60. Deletion of sinR restored production of the EPS layer to both the meeY and yybP mutants (Fig. 9). Although the meeY sinR and yybP sinR strains had regained EPS (Fig. 9), the pellicle biofilms failed to accumulate Mn and Ca (Table 3). Similarly, the sinR mutants had reduced metal-loading, particularly in early-stage pellicles (Table 3). A sinR mutation restores EPS in meeY and yybP mutants. B. subtilis strains were recovered from early pellicle biofilms grown in MSgg at 37 °C and EPS was visualized by negative staining with India ink and DIC microscopy. The images are representative of three or more independent cultures for each genotype. The polysaccharide capsules were clearly visible around the WT, sinR, meeY sinR, and yybP sinR mutants. The green scale bars represent 10 μm. Discussion The decision to build a biofilm begins within a complex signal transduction cascade that converges to regulate the phosphorylation of Spo0A, a master regulator with major impacts on Bacillus physiology. Prior work suggests that many regulators control production of the biofilm matrix13,20. Spo0A-P is critical for production of SinI and thereby sequestration of SinR. This regulatory switch is reinforced by a positive feedback loop involving derepression of SlrR, a second SinR antagonist. A decrease in SinR-mediated repression allows binding of the RemA activator, which induces the epsA-O and tapA-sipW-tasA operons (Fig. 10). MeeY and YybP are integral to biofilm signaling and metal homeostasis. MeeY and YybP are two membrane-associated proteins. In their absence, the synthesis of EPS is reduced and mature biofilms do not form. This defect in eps expression is upstream of SinR and likely affects the phosphorelay. Expression of meeY and yybP in biofilm conducive conditions (0.1 mM Mn and 1% glycerol) may be regulated, in part, by Mn-sensing riboswitches upstream of both meeY and yybP. EPS functions to retain divalent cations (Ca, Mn, and Mg) within the biofilm matrix in a process that involves MeeY and YybP. Here, we show that MeeY and YybP are both required for proper maturation of colony and pellicle biofilms. In the absence of either MeeY or YybP the expression of the epsA-O operon is greatly reduced (Fig. 5) as is biofilm-associated sporulation (Fig. 4) and hydrophobicity (Fig. 3). These phenotypes are related, since a lack of ECM contributes to defects in sporulation and in the expression of BslA, which can account for the loss of hydrophobicity. Further, measurement of the phosphorylation status of Spo0A using a spoIIGA-luxABCDE reporter revealed a large reduction in expression in the absence of MeeY or YybP (Fig. 4). The biofilm defects of ∆meeY and ∆yybP can be partially rescued by deleting SinR or by inducing SlrR (Fig. 6). These results suggest that the biofilm defect in ∆meeY and ∆yybP is correlated with decreased activity of the phosphorelay that culminates in Spo0A-P and is required for alleviating the SinR-mediated repression of the epsA-O genes (Fig. 10). It is still unclear why the ∆meeY and ∆yybP mutants have strongly reduced expression from the epsA promoter, but a lesser effect on the tapA promoter. One difference is that tapA, but not epsA, requires SlrR for its expression61. It is also not yet clear why biofilm formation can also be rescued by overexpressing the activator RemA in the ΔmeeY, but not in the ∆yybP, strain. The meeY and yybP genes are induced under biofilm conducive conditions (LBGM), but not in LB amended with glycerol or Mn alone (Fig. 8). The Mn requirement can be attributed, at least in part, to the regulation of MeeY and YybP by Mn-sensing riboswitches. Further, induction of meeY and yybP seems to require the ECM, since ∆epsA-O and ∆tapA-sipW-tasA strains did not show induction in LBGM (Fig. 8). ECM is already posited to have a signaling role during biofilm formation, where it is required for sporulation62. Since the levels of Mn and Ca in the biofilms are much higher than in planktonic cells59, we infer that much of the biofilm-associated metal is associated with the anionic ECM. We currently favor a model in which MeeY and YybP are critical for the proper metalation of the ECM. Analysis of colony and pellicle biofilms using ICP-OES indicates that Ca and Mn levels are greatly reduced in the weak biofilms formed by the ∆meeY and ∆yybP mutants. The loss of biofilm-associated metals is similar to that seen with an ∆epsA-O mutant. This is consistent with the observation that ∆meeY and ∆yybP mutants are deficient in expression of epsA-O and the cells lack a visible EPS capsule (Fig. 9). Strikingly, deletion of sinR restores EPS encapsulation to the ∆meeY and ∆yybP strains (Fig. 9), but these cells do not recover normal metalation of the ECM (Tables 2, 3). The metal distribution within complex colony biofilms has been visualized using X-ray diffraction, which revealed a striking co-localization of Mn and Ca associated with the visible wrinkles63. Although these authors suggest that TapA fibrils may play a central role in metal accumulation, our results are more consistent with a primary role for the polyanionic EpsA-O polysaccharide60. High levels of Ca, including as mineralized CaCO3, have also been associated with biofilm maturation46. This process of mineralization required activation of the Spo0A phosphorelay, which was primarily dependent on the KinA and KinB kinases64, and may additionally involve a putative Ca-translocating ATPase, YloB46. Other metals also interact with the anionic ECM. For example, excess Zn and Cu can induce high levels of extracellular γ-polyglutamic acid as a protective shield65. While the critical role of MeeY and YybP in biofilm formation is clear, the underlying mechanisms are not yet understood. MeeY is a TerC family membrane protein66 implicated in Mn transport in support of metalation of extracytoplasmic proteins and surfactin42,43,53. TerC proteins belong to the uncharacterized protein family 16 (UPF0016), which includes membrane transporters active with both Mn and Ca ions. Representative members include yeast Gdt1p, human TMEM165, and plant PAM7167. Both Gdt1p and TMEM165 are thought to translocate Ca and/or Mn ions into the Golgi to assist in protein metalation. Therefore, we speculate that MeeY may be important in metalation of EPS by the direct transport of Mn from the cytosol to the exterior of the cell (Fig. 10). Translocation of Ca may also involve MeeY in addition to YloB46. YybP is predicted to be a membrane-associated lipoprotein with an extracellular domain of unknown function (DUF6376) that has similarity to predicted RNA-binding domains of the ANTAR family44 (EMBL Smart). YybP is unlikely to be a transporter but could function in sensing extracellular signals such as EPS or Ca and communicating this information to other membrane proteins. Ultimately, understanding the precise roles of MeeY and YybP in supporting the formation of biofilms will require biochemical analyses. Methods Bacterial strain construction, media, and growth conditions To generate various genotypes in Bacillus subtilis 3610 and DK1042 we used generalized transduction and transformation, respectively. Initially, all the mutations were created in 168 background using transformation performed in a modified competence media37 and selected on appropriate antibiotics [chloramphenicol (10 μg/ml), macrolide-lincosamide-streptogramin B - MLS is the mixture of erythromycin (1 μg/ml) plus lincomycin (25 μg/ml), and kanamycin (15 μg/ml)] and these mutations were then introduced into 3610 using SPP1 phage. SPP1 phage transduction: for packaging desired donor DNA mutations in the phage, serial dilution of SPP1 lab stock was performed. Next, 10 μl of each SPP1 phage dilution was used to infect 0.2 ml of 168 bacterial cells grown overnight in TY broth [Lennox Broth (RPI) supplemented with 10 mM Mg and 0.1 mM Mn]. This mixture was inverted once to mix and incubated at 37 °C for 20 min under static conditions. Then the mixture was added to a molten soft TY agar of 50 °C (3 ml). The phage-bacterial suspension in the agar was mixed and overlayed on top of the 15 ml of hard TY agar supplemented with appropriate antibiotics for selection. These plates were incubated at 37 °C overnight for plaque development. We observed a range of plaque formation, for our parent SPP1 phage stock usually 10−3 and 10−4 with multiple plaques selected. These plates were flooded with 4 ml of TY broth and plaques were scraped off to dislodge the top layer using a sterile tip and a glass spreader. The phage solution was collected in a 15 ml falcon tube and centrifuged at 4000 rpm for 2 min at room temperature. The supernatant was filter sterilized using 0.2 µm filter (Corning) and stored at 4 °C for further use as a phage lysate. To 0.9 ml of 3610 recipient cells, which were grown in TY broth to OD600 of 1.0 of stationary phase, were infected with 10 µl phage lysate in a 15 mL falcon tube. To this we added 5 ml of TYC (TY supplemented with 10 mM sodium citrate) broth and further incubated at 37 °C for 20 min. These tubes were centrifuged at 5000 rpm for 5 min at room temperature. The supernatant was carefully removed, and the tubes were completely emptied of liquid on a paper towel. To this pellet, we added 0.1 ml of fresh LB broth and plated on TYC plates containing appropriate antibiotic selection. These plates were further incubated at 37 °C for an overnight period for colony development. Colonies with active infections were picked and patched on LB agar consisting of an appropriate antibiotic. All oligos and strains used in this study are listed in Table S1 and S2, respectively. We performed the whole-genome resequencing to confirm selected strains (3610, ∆meeY, ∆yybP, YP, ∆meeY ∆sinR, and ∆yybP ∆sinR), which contained mutations in the respective gene and not in any other loci. Promoter-riboswitch-luciferase assay The promoter and initial transcribed region (including the riboswitch element) of meeY and yybP were fused to luciferase in pBS3K-lux plasmid68 and were integrated at sacA. Cells harboring these constructs were grown in LB broth to an OD600nm of 0.6 and then 2 µl of cells were inoculated into 198 µl of either LB or LB + 0.2 mM Mn (LBM) or LB + 1% glycerol (LBG) or LB + 1% glycerol + 0.2 mM Mn (LBGM). All media were buffered with 50 mM MOPS (pH 7.0) and cell growth monitored in a plate reader at 30 °C with shaking for 30 h. (BioTek Synergy H1; Agilent, VT). The relative luminescence was calculated and plotted by normalizing against the OD600. Biofilm assay Colony biofilms were monitored both in the complex and defined media. For complex media we used LBGM (LB agar supplemented with 1% glycerol and 0.1 mM MnCl2)15 and for a defined media we employed MSgg69 as a biofilm-inducing and supporting conditions in the laboratory. 4.5 ml of semi-solid media was dispensed into 8 well rectangle plate with 128 mm by 86 mm (cat#167064, Nalgene) dimensions. Where indicated, an appropriate amount of surfactin (Sigma) or xylose or sterile cell supernatant was mixed. The plates were allowed to dry for 15 min in a laminar air flow hood and 4 µl of overnight grown culture was air dropped in a center position. These plates were further incubated under similar conditions to dry the culture and then incubated at 30 °C for 5 days. For pellicle biofilms, we used either 125 ml of Erlenmeyer flask containing 25 ml of LBGM and or 24 well containing 4.5 ml of LBGM broth with either 200 or 10 µl of inoculum, respectively. For pellicle growth, standard 24 well polystyrene plates (cat#7338, Corning) were used. As a defined medium, we used minimal salts of glycerol glutamate (MSgg) media prepared as per Camba et al.70. Basically, 100 mM MOPS [pH 7.0], 5 mM phosphate buffer, 0.5% (w/v) glycerol, 0.5% (w/v) glutamate, 700 µM CaCl2, 2 mM MgCl2, 100 µM FeCl2, 50 µM (for pellicle)-250 µM (for colony) MnCl2, 1 µM ZnCl2, and 2 µM thiamine-HCl was prepared in milli-Q water with 1.5% agar in final concentration and after sterilization 4.5 ml was dispensed per well in the 8 well plates. These plates were air dried in laminar flow hood for 10 min and then 10 µl of overnight-grown bacterial culture was air dropped till it dried out. For 2xSG base was prepared with 16 g/L nutrient broth (Difco), 0.5 g/L MgSO4.7H2O, and 2 g/L KCl adjusted to pH 7.0 (17 g/L Agar) autoclaved and was supplemented with sterile stocks of 1 mM Ca(NO3)2, 0.1 mM MnSO4, 1 μM FeSO4, and 0.1% glucose71. Sporulation assay Cells were grown under biofilm conditions on LBGM and sporulation was determined as described16. Briefly, biofilms were collected, resuspended in 0.5 mL of water and vortexed followed by a mild sonication. Further, OD600 was determined and equal volume of samples were spilt into two vials. One was subjected to 80 °C heating for 20 min while other kept without heating. These samples were then serially diluted in LB broth and were plated on LB agar for viable CFUs. Imaging Cultures grown in or on LBGM and MSgg were incubated under static conditions at 30 °C for 5 days. Images were captured periodically using a Nikon DX SWM VR aspherical camera mounted on a tripod with fixed magnification or iPhone (1.7× magnification) depicting an arial/side view. For GFP & YFP fusion imaging ChemiDoc MP imager (Bio-Rad) with Alexa 488 filter setting was employed. All the fusions were tested together and also individually for the optimal exposure and resolution. The yellow and green false coloring were applied to signal intensities of cells expressing YFP and GFP fusions, respectively. For spore visualization a NIKON Ti2-E Inverted Motorized Microscope was used with 40× phase contrast. For GFP and YFP reporter fusion 100X magnification with either FITC or YFP filters were used in a fluorescence setting along with the phase contrast in a microscope. For EpsA–O visualization, B. subtilis 3610 strains WT, ∆eps, ∆sinR, ∆meeY, ∆yybP, YP, ∆meeY ∆sinR, and ∆yybP ∆sinR were grown overnight in MSgg liquid medium at 37 °C. Two microliters of overnight culture were mixed with an equal volume of Indian ink on a microscope slide as described60. A coverslip was placed on the slide, and excess fluid was removed by applying gentle thumb pressure. Negatively stained samples were observed using differential interference contrast (DIC) microscopy with an oil-immersion objective. The presence of EpsA–O was indicated by the exclusion of Indian ink particles surrounding individual bacterial cells or bacterial aggregates. The images shown are representative of three independent biological experiments. Hydrophobicity test and weight measurements The 10 mg/mL of Safranin stain stock was prepared in water and 10 μL was spotted onto various biofilms without touching them. Images were immediately captured for arial and side view. To measure total cell mass, colonies were collected from an agar plate and weighed using a precision balance scale (Mettler Toledo) (n = 6). Pulcherrimin release estimation All strains were grown as liquid pellicle biofilms using LBGM (25 mL) and monitored at 30 °C for 5 days. After 5 days pellicle was separated from liquid broth using cellulose acetate filter sterilization unit. The filtrate was then read at 410 nm for pulcherrimin estimation. LBGM blank was subtracted from the readings. Inductively coupled argon plasma-optical emission spectrometry (ICP-OES) ICP-OES analysis was performed using modifications of a previously described procedure59. Briefly, colony biofilms were grown in LBGM agar for 5 days at 30 °C and for pellicle biofilm, cultures were established in LBGM broth after an overnight at 37 °C. The biofilms were then transferred to a microcentrifuge tube, weighed, and washed with Chelex-treated phosphate-buffered saline (PBS), resuspended in 0.5 mL PBS, and transferred to ICP-glass tubes for drying the samples. Samples were digested in double distilled HNO3 in a carbon heat block using a Vulcan 84 automated sample digestion unit (http://www.qtechcorp.com/), followed by the addition of 60/40 nitric/perchloric acid and incubation at 150 °C. Samples were cooled down and brought to a constant final volume with deionized 18 MOhm water with 5% HNO3. ICP analysis was run on an Agilent 7700 series ICP-ES housed at the US Department of Agriculture—ARS Robert W. Holley Center for Agriculture and Health in Ithaca, NY. Data were normalized to the weight of the biofilms [mean ± standard deviation (SD); n = 4 independent biological replicates]. Liquid chromatography-mass spectrometry (LC-MS) identification Sample preparation Cell-free supernatants from pellicle biofilms (post 5 days) were collected from WT, ∆meeF, ∆meeY, ∆yybP, ∆yjbE, YP, and ∆srfAA strains. For each sample, 200 µL of supernatant was transferred to a 1.5 mL microcentrifuge tube and mixed with 600 µL of ice-cold 100% methanol (final concentration 75% MeOH). Samples were vortexed for 20 s, incubated at 4 °C for 60 min, and centrifuged at 18,000 × g for 10 min at 4 °C. The clarified supernatant was recovered quantitatively and dried in a speed-vac, then stored at –20 °C. Dried residues were reconstituted in 300 µL of 30% acetonitrile with 0.1% formic acid, and 10 µL was injected for LC-MS analysis. HPLC Conditions: chromatographic separation was performed on a Shimadzu Nexera LC-20ADXR using a ThermoScientific Hypersil Gold aQ C18 column (100 × 2.1 mm, 1.9 µm). The flow rate was 0.20 mL/min, column temperature 35 °C, and mobile phases were: A: 0.1% formic acid in water and B: 0.1% formic acid in 95% acetonitrile/4.9% water. An injection volume of 10 µL was used for samples, blanks, and 1 µM of Surfactin (Sigma) standards. LC–MS/MS (IDA) conditions Data were acquired on a SCIEX 7600 ZenoTOF using an IDA method optimized for surfactin C. Electrospray ionization was performed in positive mode (5.5 kV) with the following parameters: GS1 40 psi, GS2 60 psi, curtain gas 35, CAD 9, and source temperature 450 °C. MS1 spectra were collected from m/z 100–2000 (0.2 s accumulation). The top 10 ions were subjected to ddMS² with CE 35 ± 15 V, collecting fragments from m/z 200–2000. Calibration was performed using a negative calibrant (CDS system). Data Processing: all data were processed and interpreted using SCIEX OS v3.3. Supplementary information SI Source data Source Data Acknowledgements This work was supported by NIH grant R35GM122461 awarded to J.D.H. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We acknowledge Janice J. Im for technical assistance. We acknowledge Austin Shambles of IDI-GEMS Bacillus Genetic Stock Center (BGSC) at OSU, Prof. Yunrong Chai, Prof. Dr. Jörg Stülke, and Dan Kearns for the strains. We acknowledge Andy Lui (LC-MS Cornell BRC) and Eric Craft (ICP-OES) for assistance with the identification of lipopeptides and metals, respectively. We mention Tobias Doerr, Akos Kovács, Nicola Stanley-Wall, and Yunrong Chai for their comments on the initial draft of this paper. Author contributions Conception: A.J.S. and J.D.H. Experimental design and all experiments performed: A.J.S., ICP-OES analysis: M.P. Figures and illustrations: A.J.S. Manuscript drafted and edited: A.J.S. and J.D.H. Data availability The data supporting the findings of this study are available within the article and its supplementary files. Any additional requests for information can be directed to and will be fulfilled by the corresponding authors. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Ankita J. Sachla, Email: ajs588@cornell.edu John D. Helmann, Email: jdh9@cornell.edu Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41522-026-01045-0. References Associated Data Supplementary Materials SI Source Data Data Availability Statement The data supporting the findings of this study are available within the article and its supplementary files. Any additional requests for information can be directed to and will be fulfilled by the corresponding authors.