Differential Regulation of CsrC and CsrB by CRP-cAMP in Salmonella enterica

Post-transcriptional regulation mediated by regulatory small RNAs (sRNAs) has risen as a key player in fine-tuning gene expression in response to environmental stimuli. Here, we show that, in Salmonella enterica, the central metabolic regulator CRP-cAMP differentially regulates the sRNAs CsrB and CsrC in a growth phase-dependent manner. While CsrB expression remains unchanged during growth, CsrC displays a growth phase-dependent expression profile, being weakly expressed at the logarithmic growth phase and induced upon entry into stationary phase. We show that CRP-cAMP contributes to the expression pattern of CsrC by repressing its expression during the logarithmic growth phase. The CRP-cAMP mediated repression of CsrC is independent of SirA, a known transcriptional CsrB/CsrC activator. We further show that the sRNA Spot 42, which is derepressed in a Δcrp strain, upregulates CsrC during logarithmic growth. We propose a model where the growth-dependent regulation of CsrC is sustained by the CRP-cAMP-mediated repression of Spot 42. Together, our data point toward a differential regulation of the sRNAs CsrB and CsrC in response to environmental stimuli, leading to fine-tuning of gene expression via the sequestration of the RNA-binding protein CsrA.


INTRODUCTION
Bacteria need to adapt rapidly to changing environmental conditions, which is particularly crucial for pathogenic bacteria during the process of infection. While transcription plays a major role in the regulation of gene expression, bacteria display a plethora of post-transcriptional mechanisms that allow the fine-tuning of gene expression in response to environmental cues. Salmonella enterica has been investigated extensively with respect to gene regulation and has become a model for the study of post-transcriptional RNA-mediated regulation.
Among the described post-transcriptional regulation mechanisms, a prominent role has been attributed to sRNAs that modulate gene expression primarily via binding to target mRNAs (Hör et al., 2020). However, sRNAs can also regulate gene expression at the post-translational level, such as the two sRNAs CsrB and CsrC. These sRNAs contain several stem loop structures with GGA motifs in their loop regions, enabling them to bind CsrA. CsrA is a widely conserved RNA-binding protein that inhibits translation by binding to GGA motifs around the ribosome-binding site of target mRNAs (Romeo and Babitzke, 2018). Binding of CsrB and CsrC to CsrA titrates CsrA away from its target mRNAs, thereby counteracting its inhibitory activity (Liu et al., 1997;Weilbacher et al., 2003).
CRP is a transcription factor that acts as a metabolic sensor and becomes active upon binding to the intracellular second messenger cAMP (cyclic adenosine monophosphate) (Görke and Stülke, 2008). Interaction of CRP-cAMP with DNA leads to activation or repression of its target genes. In E. coli, it has been shown that CRP-cAMP represses the expression of both CsrB and CsrC (Pannuri et al., 2016). While repression of CsrB occurs through an indirect mechanism, the repression of CsrC occurs by direct binding of CRP-cAMP to the promoter region of csrC, where it competes with the csrC activator UvrY (Pannuri et al., 2016). In Salmonella, the homolog of the two-component system BarA-UvrY is BarA-SirA (Johnston et al., 1996), which, as in E. coli, it also positively regulates the expression of CsrB and CsrC (Teplitski et al., 2003;Fortune et al., 2006;Martínez et al., 2011). Expression studies using S. enterica cultures on solid media indicates that CRP-cAMP, in contrast to its role described in E. coli, positively regulate the expression of CsrB and CsrC via upregulation of sirA (Teplitski et al., 2006). In S. enterica other sRNAs, such as CyaR and Spot 42, are also regulated by CRP-cAMP, acting as an activator for CyaR and as a repressor for Spot 42 (Papenfort et al., 2008;El Mouali et al., 2018).
In this study we investigated the role of CRP-cAMP in the expression of the sRNAs CsrB and CsrC in the model organism S. enterica using liquid cultures. We describe that CsrC, but not CsrB, display a growth-dependent expression pattern. CsrC expression is silenced during logarithmic growth and highly expressed upon entry into stationary phase, while CsrB expression seems to be constitutive through the growth curve. CRP-cAMP plays a relevant role in this growth-dependent regulatory network. CRP-cAMP differentially regulates the levels of CsrB and CsrC as it does not affect CsrB expression but represses CsrC expression during logarithmic growth. Remarkably, while SirA is required for full expression of CsrC, the CRP-cAMP-mediated repression during logarithmic growth seems to be independent of SirA-mediated regulation. Our data further indicates that Spot 42 contributes to the CRP-cAMPinduced repression of CsrC, suggesting that CRP-cAMP and Spot 42 converge into the differential regulation of CsrB and CsrC in Salmonella.

CRP-cAMP Represses CsrC Levels at the Logarithmic Growth Phase
In order to study the CRP-cAMP mediated regulation of CsrB and CsrC in Salmonella, the expression of both sRNAs was monitored using transcriptional reporter fusions. The regulatory elements controlling CsrB and CsrC expression in Salmonella have been identified in silico and characterized experimentally (Martínez et al., 2014). Accordingly, the described regulatory regions of CsrB (−403, +18) and CsrC (−347, +60) were cloned as lacZ transcriptional fusions in the pQF50 vector (Farinha and Kropinski, 1989), allowing to monitor the levels of CsrB and CsrC during growth. The contribution of CRP-cAMP was assessed by determining the expression level of CsrB and CsrC in wild-type (WT) and in a crp mutant strain, lacking the transcriptional factor CRP. In rich media, no differences in growth rate could be observed in WT and crp strains carrying either csrB-lacZ or csrC-lacZ ( Figure 1A). Transcriptional expression was monitored during logarithmic growth (OD 600 nm 0.4) and upon entry into stationary phase (OD 600 nm 2.0). In the WT, CsrB expression was apparently identical in the two growth phases ( Figure 1B). By contrast, CsrC displayed a growth-dependent expression pattern, being less expressed during logarithmic growth and being induced after entering stationary phase ( Figure 1B). Remarkably, CRP-cAMP contributes to the CsrC growth-dependent expression pattern. In crp, the expression of CsrC is induced when compared to WT at the logarithmic phase and to a lesser extent at the stationary phase, indicating that CRP-cAMP represses CsrC expression ( Figure 1B). By contrast, no effect on CsrB regulation by CRP-cAMP was observed. These results were unexpected since it was previously described that CRP-cAMP acts as an activator of both CsrB and CsrC expression when Salmonella is grown on LB agar media (Teplitski et al., 2006). To discern if the discrepancy could be consequence of differences in the genetic constructs used to monitor gene expression, a similar experiment as in Teplitski et al. (2006) was performed using our strains. The transcriptional expression was monitored after growth on LB agar media. Consistently with the previous report, the transcriptional expression of CsrB and CsrC was strongly diminished in a crp derivative strain compared to WT when Salmonella cells were grown on LB agar media as noted by the white colony phenotype of crp when compared to the blue colony phenotype of WT (Supplementary Figure S1; Teplitski et al., 2006). Altogether, CRP-cAMP seem to be required for activation of the expression of CsrB and CsrC in solid media, while it acts as a repressor of CsrC particularly at logarithmic growth phase. This indicates that CRP-cAMP-mediated regulation of CsrB and CsrC is dependent on the growth conditions.
The differential regulation of CsrC and CsrB by CRP-cAMP at the logarithmic growth phase was further corroborated by the direct RNA detection of CsrB and CsrC. In the WT, CsrC transcript was barely detected, indicating that CsrC expression is tightly silenced. In the crp strain, high levels of CsrC were detected indicating that CRP-cAMP is involved in the CsrC silencing during logarithmic growth ( Figure 1C). Contrasting the CRP-cAMP-dependent changes in CsrC levels, CsrB was not influenced by knockout of crp ( Figure 1C), thereby agreeing with the results of our transcriptional fusions ( Figure 1B). In addition, deletion of csrB cause a mild upregulation of CsrC, presumably by affecting the positive feed forward loop that free CsrA protein exerts on its repressors CsrB and CsrC (Romeo and Babitzke, 2018).
CRP becomes active upon binding to cAMP, which is produced by the adenylate cyclase Cya. Therefore, absence of crp or cya should display similar expression profiles. Accordingly, the expression of CsrC in logarithmic growth phase is induced in the cya derivative strain when compared to WT while no change was observed in the expression of CsrB ( Figure 1D). (C) Northern blot analyses of the csrB and csrC transcripts. Total RNA samples from cultures of WT (SV5015) and its derivatives crp, csrB, and csrC grown in LB up to an OD 600 nm of 0.4 were analyzed. Detection of the 5S transcript was used as a control. Transcriptional expression of csrB (D) and csrC (E) in the cya mutant strain and chemical complementation. Cultures of WT, crp and cya derivatives were grown in LB either in the absence or in the presence of cAMP (5 mM) at 37 • C up to logarithmic phase (OD 600 nm of 0.4) and samples were taken for β-galactosidase measurement. In B, C and D; β-galactosidase activity was determined for three independent cultures, average and standard deviation are presented. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant.
To further confirm the involvement of CRP-cAMP in CsrC regulation, chemical complementation of cya was carried out by ectopic addition of cAMP. The addition of cAMP repressed the expression of CsrC in the cya derivative strain while no effect was observed in CsrB expression ( Figure 1E). Our data indicate that CRP-cAMP is involved in the growth-dependent regulation of CsrC by repressing its expression.

CRP-cAMP-Mediated Repression of CsrC via a SirA-Independent Pathway
To further characterize the CRP-cAMP mediated regulation of CsrC, a chromosomal csrC-lacZ fusion was generated. In agreement with earlier results, chromosomal csrC-lacZ has a growth dependent expression pattern, where it is lowly expressed at the logarithmic growth phase and induced upon entry into early stationary phase (Figure 2A). Interestingly, we further observed an eightfold induction of the chromosomal csrC-lacZ fusion in the crp background when compared to the WT, which was only true during logarithmic growth (Figure 2A).
The BarA-SirA two-component system was described to positively regulate the expression of CsrC (Teplitski et al., 2006). The possible involvement of BarA-SirA in the regulation of CsrC by CRP-cAMP was assessed. SirA deletion leads to a decrease of the overall transcriptional expression of the chromosomal csrC-lacZ fusion, 28.2 ± 0.8 Miller units in sirA sirA, and crp sirA genetic backgrounds. The transcriptional expression is shown in relative values. In each case, the reference (crp + , WT) was set as 1.0. Miller units of the crp + backgrounds were 308 ± 19 and 28 ± 0.8 for WT and sirA, respectively. β-galactosidase activity was determined for three independent cultures, average and standard deviation is presented. ****p < 0.0001; ns, not significant. compared to 308.8 ± 19.3 Miller units in WT. Interestingly, despite the overall lower expression levels of CsrC in the sirA background, the deletion of crp led to a sixfold increase in the transcriptional expression of csrC-lacZ compared to WT (Figure 2B), indicating that CRP-cAMP modulates csrC expression via a SirA-independent mechanism.
In E. coli, it was proposed that CRP-cAMP represses CsrC expression through binding to the upstream region of csrC and competing with UvrY, the SirA homolog (Pannuri et al., 2016). In Salmonella, SirA induces CsrC expression by binding ∼160-168 bp upstream of the csrC promoter ( Figure 3A; Martínez et al., 2014). Alignment of the upstream promoter regions of CsrC from E. coli and Salmonella showed that SirA binding site is conserved, whereas the CRP-cAMP binding site displays a lesser extent of conservation (Supplementary Figure S2). To characterize if CRP-cAMP binding to the csrC promoter is required for the described regulation, an additional csrC-lacZ fusion was generated where the putative binding sites of these transcriptional factors are not present. As shown in Figure 3A, the pQF50 cloned fragment csrC 347 -lacZ (−347, +60) maintains SirA/CRP binding sites while csrC 91 -lacZ (−91, +60) does not. Remarkably, in both csrC 347 -lacZ and csrC 91 -lacZ an induction of expression is observed in the crp strain when compared to WT (Figure 3B). These results suggest that CRP-cAMP represses the expression of CsrC during logarithmic growth phase via a mechanism that is independent of the competition between SirA and CRP for the csrC promoter.

The sRNA Spot 42 Positively Regulates CsrC
In Salmonella, CRP-cAMP represses the expression of the regulatory sRNA Spot 42 in a growth-dependent manner (El Mouali et al., 2018). The similarity of the expression profiles of Spot 42 and CsrC, let us to hypothesize that CRP-cAMP might regulate CsrC via modulation of Spot 42 levels. To this end, we determined CsrC expression during logarithmic growth upon ectopic expression of Spot 42. Remarkably, CsrC transcript levels were strongly induced upon Spot 42 overexpression (pBRSpot 42) when compared to the vector control strain (pBRVC) (Figure 4A). In contrast, no accumulation of CsrB was detected upon overexpression of Spot 42 ( Figure 4A). Further supporting our hypothesis, Spot 42 seemed to differentially regulate CsrB and CsrC expression (Figure 4A), similar to the described CRP-cAMP-mediated regulation ( Figure 1C). To verify this observation, we overexpressed Spot 42 in the chromosomal csrC-lacZ fusion background, resulting in a threefold induction of CsrC when compared to the vector control strain (Figure 4B). The Spot 42-mediated regulation of CsrC may account for the described repression of CsrC by CRP-cAMP. The contribution of Spot 42 on the induction of CsrC in crp was further assessed. The CsrC derepression in a crp mutant strain drops in absence of Spot 42 ( spf ) as compared to a Spot 42 proficient strain indicating that Spot 42 contributes to the CsrC derepression in a crp background ( Figure 4C).
Interestingly, it has been described that SirA can bind to the promoter region of spf (Zere et al., 2015), suggesting that SirA might be regulating Spot 42. However, transcriptomic data of sirA compared to WT in Salmonella indicates that Spot 42 it is not regulated by SirA (Colgan et al., 2016). The effect of SirA on Spot 42 expression was assessed, using a chromosomal spf -lacZ in the presence and absence of SirA in both WT and crp genetic backgrounds at logarithmic growth phase. As previously shown, Spot 42 expression is induced in the crp mutant compared to WT (El Mouali et al., 2018; Figure 5A). Remarkably, no regulation was observed in the absence of SirA and induction of Spot 42 expression in the crp mutant strain is still observed in absence of SirA ( Figure 5A). These results indicate that SirA it is not regulating Spot 42 expression under the studied conditions. Furthermore, SirA was not required for the positive Altogether, our data indicate that CRP-cAMP differentially regulates CsrB and CsrC at the logarithmic growth phase where it represses specifically CsrC but not CsrB. The CRP-cAMP regulated sRNA Spot 42 arise as new regulator of the Csr regulon in Salmonella.

DISCUSSION
CRP-cAMP is a global regulator initially described to regulate metabolic genes in response to nutrient stimuli. Adding to its more prominent role as transcriptional regulator of mRNAs, CRP-cAMP was also described to regulate the expression of sRNAs such as CyaR, Spot 42, and FnrS (Polayes et al., 1988;Papenfort et al., 2008;De Lay and Gottesman, 2009;Durand and Storz, 2010). CRP-cAMP also modulates the expression of the sRNAs CsrB and CsrC which regulate CsrA protein activity via titration. Previous studies reveal apparent discrepancies in the role of CRP-cAMP on CsrB and CsrC expression. In E. coli, CsrB and CsrC are both repressed by CRP-cAMP, being CsrC expression directly repressed by CRP-cAMP via binding competition with the transcriptional activator UvrY (Pannuri et al., 2016). CRP-cAMP activity is modulated, among others, by the phosphorylation state of the EIIA Glc , which is involved in glucose transport. When phosphorylated, it stimulates cAMP production, promoting the activity of CRP-cAMP. Surprisingly, dephosphorylated EIIA Glc interacts with CsrD and promotes the degradation of CsrB and CsrC by RNAseE (Leng et al., 2016). In other words, conditions that promote the transcriptional derepression of CsrC and CsrB via CRP-cAMP also promote the degradation of these transcripts (Leng et al., 2016;Pannuri et al., 2016). In contrast, in Salmonella CRP-cAMP has been described to play a positive role on CsrB and CsrC expression when it is grown on LB agar media (Teplitski et al., 2006). Although we corroborate these data, we also demonstrate that in LB liquid media, CRP-cAMP differentially regulates CsrB and CsrC. At the logarithmic phase, CsrC but not CsrB is repressed. The alterations in the cell physiology when growing planktonically and within a colony may account for the different regulation described for CsrB and CsrC in Salmonella. In addition of being differentially regulated by CRP-cAMP, CsrB, and CsrC depicted distinct expression profiles during the growth curve. CsrB expression seems to be constitutive, whereas, CsrC expression is silenced during logarithmic growth and induced in early stationary phase. The fact that production of CsrC but not CsrB is growth phase dependent indicates that each RNA responds differently to specific environmental inputs. CRP-cAMP is involved in the growth-dependent regulation of CsrC, suggesting that specific physiological signals that alter CRP-cAMP levels would cause alterations in the levels of CsrC and the concomitant alterations of free CsrA levels that would modulate gene expression to promote adaptation to the new conditions. The differential regulation of CsrB and CsrC, not only by CRP-cAMP, but potentially by additional regulators would provide sensitivity to the Csr system. Far from an ON/OFF state, the Csr regulon would display a scale of grays that would allow the bacteria to fine-tune gene expression in response to environmental stimuli. Interestingly, differential regulation of CsrB and CsrC by CRP-cAMP has been reported in Yersinia pseudotuberculosis, where CRP-cAMP activates the expression of CsrC and represses the expression of CsrB, highlighting complex species-specific regulation of the Csr system (Heroven et al., 2012).
SirA is an activator of CsrB and CsrC (Teplitski et al., 2006;Martínez et al., 2011Martínez et al., , 2014. While we observe that SirA is required for full activation of CsrC, our data let us conclude that SirA is not involved in the deregulation of Transcriptional expression of the chromosomal csrC-lacZ fusion was monitored in WT, spf, crp, and spf crp genetic backgrounds. In all cases cultures were grown in LB at 37 • C up to an OD 600 nm of 0.4. β-galactosidase activity was determined for three independent cultures, average and standard deviation is presented. ****p < 0.0001, ***p < 0.001. CsrC when CRP-cAMP is depleted. The CsrC deregulation was detected both in absence of SirA and when the SirA binding site was removed from the csrC promoter. Our data indicate that the sRNA Spot 42, which is transcriptionally repressed by CRP-cAMP at the logarithmic growth phase, positively regulates CsrC expression but it does not affect CsrB (Figure 6). This suggests a model where CRP-cAMP differentially regulates CsrB and CsrC via derepression of the trans-encoded sRNA Spot 42. Of note, Spot 42 does not seem to be solely responsible for the derepression of CsrC in absence of crp, as a partial derepression of CsrC is still observed in absence of crp and spf. Suggesting that, CRP-cAMP additionally represses CsrC expression through a Spot 42 independent mechanism (Figure 6).
The role of Spot 42 in the regulation of CsrC seem to be restricted to logarithmic growth phase and it is not involved in the stationary phase dependent induction. Expression of csrC-lacZ is induced at stationary phase compared to logarithmic growth phase in both the WT and the spf backgrounds (Supplementary Figure S3). Consistently, transcriptomic data indicates that the expression level of Spot 42 is downregulated 100-fold upon entry into stationary phase when compared to logarithmic growth (Kröger et al., 2013). In agreement, Spot 42 might be responsible to fine-tune the expression of CsrC respectively. In all cases, cultures were grown in LB at 37 • C up to an OD 600 nm of 0.4. β-galactosidase activity was determined for three independent cultures, average and standard deviation are presented. *p < 0.05; ***p < 0.001; ****p < 0.0001; ns, not significant. in conditions where the CRP-cAMP activity is downregulated similarly as it occurs for the Spot 42-regulated hilD mRNA (El Mouali et al., 2018).
In this report, we provide new insights in the link between the CRP-cAMP and Csr regulons in Salmonella. We demonstrate that CRP-cAMP differentially regulates CsrB and CsrC. These two sRNAs are considered functionally redundant but our results suggest that they respond distinctly under specific environmental and physiological conditions in Salmonella. The presence of CsrB alone would affect the free pool of CsrA to a lesser extent than when both CsrB and CsrC sRNAs are present. This finetuning of the free CsrA levels would affect the Csr regulon where high affinity mRNA targets will require reduced amounts of free CsrA whereas low affinity targets will require full derepression of CsrA. We show that CRP-cAMP and its target sRNA Spot 42 contribute to the overall levels of CsrB and CsrC. Its contribution allows Salmonella to tightly control the levels of free CsrA in response to environmental stimuli. These features contribute to the versatility of pathogenic bacteria such as the model organism Salmonella enterica.

Bacterial Strains, Plasmids, and Growth Conditions
The bacterial strains, derivatives of Salmonella enterica serovar Typhimurium SL1344, were cultivated in Lysogeny broth (LB; tryptone 10 g/l, yeast extract 5 g/l and sodium chloride 10 g/l). When required, media was supplemented with the antibiotics indicated, ampicillin (Amp) 100 µg/ml, chloramphenicol (Cm) 15 µg/ml or kanamycin (Km) 50 µg/ml. Bacterial cultures were inoculated at an OD 600 nm of 0.001, cultures reaching OD 600 nm 0.4 were considered logarithmic growth and reaching OD 600 nm 2.0 was considered early stationary phase. The bacterial strains and plasmids used in this study are listed in Supplementary Tables S1, S2, respectively.

Genetic Manipulations
Deletion strains were generated by standard gene replacement as previously described (Datsenko and Wanner, 2000). For chromosomal csrC-lacZ transcriptional fusion. The csrC:Cm strain was cured from the antibiotic resistance as previously described (Cherepanov and Wackernagel, 1995). The resulting csrC:frt strain carrying the FRT scar was further used to obtain chromosomal csrC-lacZ by integration of pKG136 as previously described (Ellermeier et al., 2002). Chromosomal spf -lacZ was obtained as for CsrC (El Mouali et al., 2018). Oligonucleotides used for strains construction are listed in Supplementary Table S3.

β-Galactosidase Activity Assay
Strains of interest were grown to logarithmic growth phase (OD 600 nm 0.4) or early stationary phase (OD 600 nm 2.0). β-galactosidase activity was measured as described previously (Miller, 1992). Shortly, 100 µl of culture was added to 900 µl of buffer Z (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , and 50 mM 2-mercaptoethanol, pH 7) and cells were lysed by the addition of 10 µl of toluene. Reactions were incubated at 28 • C and β-galactosidase activity was measured upon addition of 200 µl of ONPG (4 mg/ml). Reactions were stopped by the addition of 500 µl of Na 2 CO 3 (1 M). The OD 420 nm and OD 550 nm was measured and Miller Units were calculated as previously described (Miller, 1992). β-galactosidase activity determination was performed in technical duplicates for each of three biological replicates.

Total RNA Isolation and Northern Blot
Strains of interest were grown to logarithmic growth phase (OD 600 nm 0.4). The biomass of 4 units of OD 600 nm was collected, and total RNA extracted by classic hot phenol method. Shortly, the cells were resuspended in 600 µl of TE (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) containing 0.5 mg/ml lysozyme. Then, 60 µl of 10% SDS (w/v) was added; mixed by inversion and incubated at 64 • C for 1-2 min. After the incubation, 66 µl of sodium acetate pH 5.2 (1 M) was added and mixed by inversion. For RNA extraction, 750 µl of Roti-Aqua phenol was added to the samples, mixed by inversion and incubated at 64 • C for 6 min. Upon centrifugation (15 min, 13,000 rpm, 4 • C), the top aqueous layer was transferred to a fresh 2 ml microcentrifuge tube, and 750 µl of chloroform was added. The samples were mixed and upon centrifugation (12 min, 13,000 rpm, 15 • C), the upper aqueous layer was transferred into a new tube and precipitated with 30:1 mix of ethanol and sodium acetate (1 M, pH 6.5). The samples were incubated for 2-3 h or overnight at −20 • C. The samples were then centrifuged and precipitated RNA was resuspended in water and concentration measured by NanoDrop. Samples of 10 µg of total RNA were subjected to electrophoretic separation in Tris-Borate-EDTA (TBE) 8% acrylamide gels containing 8.3 M urea. RNAs were transferred to Hybond N + (GE Healthcare) filters by semi-dry TBE based transference. Transcripts of interest were detected by hybridization with 5 radiolabeled oligos as probes. Images were obtained with the FLA-5100 imaging system (Fujifilm). Oligonucleotides used as probes are listed in Supplementary Table S3.

Statistical Analysis
Graph Pad 8.0 software was used for data analysis. Unpaired twotailed Student's t-test were carried out for two groups comparison and p < 0.05 were considered significant.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
YE, GE-M, and DG-P contributed to the investigation. YE and CB contributed to the conceptualization, investigation, formal analysis and writing the manuscript. All authors contributed to the article and approved the submitted version.