Cellular Effects of 2′,3′-Cyclic Nucleotide Monophosphates in Gram-Negative Bacteria

ABSTRACT Organismal adaptations to environmental stimuli are governed by intracellular signaling molecules such as nucleotide second messengers. Recent studies have identified functional roles for the noncanonical 2′,3′-cyclic nucleotide monophosphates (2′,3′-cNMPs) in both eukaryotes and prokaryotes. In Escherichia coli, 2′,3′-cNMPs are produced by RNase I-catalyzed RNA degradation, and these cyclic nucleotides modulate biofilm formation through unknown mechanisms. The present work dissects cellular processes in E. coli and Salmonella enterica serovar Typhimurium that are modulated by 2′,3′-cNMPs through the development of cell-permeable 2′,3′-cNMP analogs and a 2′,3′-cyclic nucleotide phosphodiesterase. Utilization of these chemical and enzymatic tools, in conjunction with phenotypic and transcriptomic investigations, identified pathways regulated by 2′,3′-cNMPs, including flagellar motility and biofilm formation, and by oligoribonucleotides with 3′-terminal 2′,3′-cyclic phosphates, including responses to cellular stress. Furthermore, interrogation of metabolomic and organismal databases has identified 2′,3′-cNMPs in numerous organisms and homologs of the E. coli metabolic proteins that are involved in key eukaryotic pathways. Thus, the present work provides key insights into the roles of these understudied facets of nucleotide metabolism and signaling in prokaryotic physiology and suggest broad roles for 2′,3′-cNMPs among bacteria and eukaryotes. IMPORTANCE Bacteria adapt to environmental challenges by producing intracellular signaling molecules that control downstream pathways and alter cellular processes for survival. Nucleotide second messengers serve to transduce extracellular signals and regulate a wide array of intracellular pathways. Recently, 2′,3′-cyclic nucleotide monophosphates (2′,3′-cNMPs) were identified as contributing to the regulation of cellular pathways in eukaryotes and prokaryotes. In this study, we define previously unknown cell processes that are affected by fluctuating 2′,3′-cNMP levels or RNA oligomers with 2′,3′-cyclic phosphate termini in E. coli and Salmonella Typhimurium, providing a framework for studying novel signaling networks in prokaryotes. Furthermore, we utilize metabolomics databases to identify additional prokaryotic and eukaryotic species that generate 2′,3′-cNMPs as a resource for future studies.

as vasodilation and visual transduction in eukaryotes (1,2), while adenosine 39,59-cyclic monophosphate (39, governs numerous processes in diverse organisms, including steroidogenesis in mammals and carbon catabolism in bacteria (3,4). Cyclic dimeric-39:59-GMP (c-di-GMP) modulates biofilm formation in many bacteria, and other cyclic dimeric nucleotides control processes such as sporulation in prokaryotes and innate immunity in eukaryotes (5)(6)(7)(8). Studies also have demonstrated that acyclic nucleotides in bacteria, including guanosine 39-diphosphate, 59-(tri)diphosphate ([p]ppGpp), and P1,P4-di-adenosine 59-tetraphosphate (Ap4A), induce transcriptome and proteome remodeling (7) and regulate biofilm production via modulation of c-di-GMP levels (9,10), respectively. In addition to the nucleotide second messengers, primary nucleotide metabolism influences a host of physiological processes in bacteria. Perturbation of de novo nucleotide biosynthesis alters biofilm formation in Escherichia coli (11)(12)(13), exogenous addition of nucleosides and dNTPs promotes a positive chemotactic response in Vibrio fischeri (14), and pyrimidine nucleobases function as chemoattractants in E. coli (15). These findings illustrate the multifaceted nature of nucleotide signal transduction and suggest that other aspects of nucleotide synthesis and salvage regulate cellular processes, warranting further investigation into the biological effects of additional nucleotide pools.
An intriguing class of nucleotides is the 29,39-cyclic nucleotide monophosphates (29,. These were first observed in E. coli several decades ago (16), but exploration of their biological relevance has been limited. Recent studies have demonstrated that 29,39-cAMP mediates stress granule assembly in Arabidopsis thaliana (17). Earlier studies in mammals correlated the appearance of extracellular 29,39-cAMP with organ stress (18) and revealed that increased intracellular 29,39-cAMP levels stimulate Ca 21 efflux in rat liver cells and oligodendrocytes, resulting in depolarization of the mitochondrial membrane and concomitant apoptosis (19). In addition, our quantification of 29,39-cNMPs in rat organs showed organ-specific variable basal levels for 29,39-cAMP, -cGMP, -cCMP, and -cIMP (20). Together, these studies support a role for 29,39-cNMPs in regulating eukaryotic physiology; however, less work has been done to characterize the biological significance of these molecules in prokaryotes.
29,39-cNMPs have been detected in different bacterial species (29,39-cCMP and -cUMP in Pseudomonas fluorescens [21]; 29,39-cAMP in Staphylococcus aureus [22]; 29,39-cAMP, -cGMP, -cUMP, and -cCMP in E. coli [23]), with our group reporting the physiological quantification of these cyclic nucleotides and mechanism of cellular biosynthesis in E. coli (24). Within E. coli, RNase I, an RNase T2 family member (25), generates all detectable 29,39-cNMPs through hydrolysis of RNA, providing the first insight into the biosynthetic origin of these atypical nucleotides in any organism (24). Additional experiments identified physiological functions for 29,39-cNMPs and RNase I in biofilm production (24); separately, RNase I was also found to modulate E. coli motility, resistance to b-lactams and acid stress, and outer membrane structure (26). These preliminary findings suggest that there are additional 29,39-cNMP-and RNase I-dependent processes yet to be discovered.
In the present study, we investigate the effects of depleting intracellular 29,39-cNMP levels in two bacterial species, E. coli and Salmonella enterica serovar Typhimurium (S. Typhimurium), using transcriptome sequencing (RNA-seq) to inform a series of physiological assays. We analyzed the transcriptome of a wild-type (WT) strain expressing a heterologous mammalian 29,39-cyclic nucleotide phosphodiesterase (CNPase) that exhibits decreased 29,39-cNMP concentrations and compared it to an equivalent strain expressing a catalytically inactive CNPase variant (27) (CNPase H73L/H152L, termed "CNP-inact") that exhibits WT 29,39-cNMP levels. CNPase was chosen for its substrate specificity; unlike other cyclic nucleotide phosphodiesterases, it does not hydrolyze 39,59-cNMPs and acts primarily on 29,39 cyclic phosphates of 29,39-cyclic mononucleotides, as well as 29,39-cyclic phosphates on the 39 terminus of oligoribonucleotides (reviewed in reference 28). Both RNase I deletion and expression of CNPase reduce 29,39-cNMP levels; however, relative to our recently reported data comparing gene expression in an E. coli Drna mutant to that in the WT strain (26), a comparison of a WT strain expressing CNPase versus one expressing CNP-inact allows us to uncover processes that are modulated directly by 29,39-cyclic phosphate-containing molecules, while excluding any confounding effects of deleting RNase I. This is particularly important, as our previous work identified roles for RNase I that are not linked to cytoplasmic 29,39-cNMP levels. Our transcriptomic analysis revealed a large number of genes with diverse cellular functions that are differentially regulated, suggesting that 29,39-cNMPs and oligoribonucleotides with 39-terminal 29,39-cyclic phosphates affect a variety of physiological activities. Comparison of CNPase-dependent transcriptome changes in E. coli WT versus Drna strains demonstrated that most CNPase-dependent differential gene expression is likely to be due to altered levels of 29,39-cyclic mononucleotides versus those of oligoribonucleotides with 39-terminal 29,39-cyclic phosphates. These effects were further characterized in subsequent physiological assays, using the expression of CNPase to explore how levels of 29,39-cyclic phosphate-containing molecules affect cellular phenotypes. As an additional tool, cell-permeable 29,39-cNMP analogs, inspired by lipophilic derivatives of 39,59-cAMP and -GMP that are widely employed to study second messenger signaling in eukaryotes (3), were synthesized to further examine 29,39-cNMP-dependent phenotypes. Our studies link 29,39-cNMPs to the regulation of genes involved in several cellular processes in E. coli, including biofilm formation and motility, highlighting key pathways that are influenced by 29,39-cNMP pools and suggesting the presence of 29,39-cNMP-sensing proteins within the cell. In addition, our results support regulatory roles for both 29,39-cNMPs and RNA oligomers with 29,39-cyclic phosphate termini in E. coli responses to stress, including acid tolerance.
Finally, a major point of interest was whether 29,39-cNMPs have a common origin or play universal regulatory roles in diverse bacteria. In this study, we examine the topic by exploring similarities and differences in 29,39-cNMP generation and physiological impacts in E. coli and S. Typhimurium, a closely related but physiologically distinct enteric pathogen. As in E. coli, RNase I plays an essential role in the generation of 29,39-cNMPs in S. Typhimurium, suggesting a common biosynthetic mechanism in the two species. Analysis of transcriptome-wide changes in WT and Drna S. Typhimurium strains expressing CNPase or CNP-inact identified common and distinct cellular processes linked to 29,39-cyclic phosphate-containing molecules, compared to E. coli, some of which were further addressed with physiological assays. Investigation of existing metabolomic databases for 29,39-cNMPs has yielded further insights into their prevalence in other bacteria and eukaryotes, suggesting potential widespread importance of 29,39-cNMPs.

RESULTS
Transcriptomics reveals 29,39-cNMP-dependent processes in E. coli. Our group previously found that RNase I (encoded by the rna gene) generates all detectable 29,39-cNMPs in E. coli (26); however, since RNase I affects nucleotide metabolism in both the cytoplasm and the periplasm (24,26), we required methods to modulate 29,39-cNMP levels independently of RNase I expression to determine which cellular effects are directly attributable to fluctuations in 29,39-cNMP pools. To this end, we leveraged the catalytic domain of Rattus norvegicus cyclic nucleotide phosphodiesterase (29) (CNPase; UniProtKB P13233 for full-length protein) to hydrolyze 29,39-cNMPs in RNase I 1 (wild-type [WT]) cells. Previous work validated that E. coli expressing CNPase from plasmid pKT-CNP exhibits subquantifiable levels of 29,39-cAMP and -cGMP, and ;25fold and ;15-fold lower levels of 29,39-cCMP and -cUMP, respectively (24). Therefore, expression of CNPase enabled identification of 29,39-cNMP-linked processes through transcriptomic profiling of WT E. coli with reduced 29,39-cNMP levels, eliminating the confounding effects inherent to using the E. coli Drna strain as in our previous transcriptomic analysis (26). RNA-seq revealed 519 genes that were significantly differentially expressed (adjusted P [P adj ] , 0.05; .2-fold change in expression) in WT E. coli expressing CNPase compared to expression in equivalent cultures expressing a catalytically inactive CNPase variant (CNP-inact) (see Data Set S1 in the supplemental material). Further transcriptome analysis to address the effects of CNPase hydrolysis of substrates other than 29,39-cNMPs is presented below; this analysis supports that CNPase-mediated reduction of 29,39-cNMP levels affect gene expression as specified in the following results.
Approximately 13% (n = 67) of these identified genes were downregulated in exponential-phase samples expressing active CNPase, which exhibit decreased 29,39-cNMP levels (Data Set S1). Of these, only 10 genes were downregulated in both the E. coli WT CNPase strain and the previously analyzed Drna strain (Data Set S1) (26), suggesting that RNase I contributes to cellular processes in ways that are independent from its role in generating 29,39-cNMPs (26). The genes that were most downregulated by CNPase were napH (9.08-fold) and napG (7.50-fold); together, their gene products produce the NapGH ubiquinol reductase that participates in the Nap periplasmic nitrate reduction pathway during anaerobic growth ( Table 1). Expression of napD, which encodes a chaperone for NapA, the nitrate reductase catalytic subunit, was additionally downregulated (6.83-fold), suggesting an additional posttranslational level of control that may contribute to regulation of the nitrate reduction pathway (Table 1). Tryptophan biosynthesis was among the other pathways that were substantially downregulated, with RNA-seq identifying decreases in the transcript levels of trpB (7.12fold), trpA (6.84-fold), trpC (5.41-fold), and trpD (4.91-fold) ( Table 1). The trp operon is well known for its regulation, facilitated by both the Trp repressor and attenuation dependent on intracellular tryptophan levels; strikingly, these data indicate the possibility that 29,39-cNMPs additionally affect regulation of this biosynthetic pathway. The mechanisms by which 29,39-cNMPs regulate these and other metabolic pathways are a subject of ongoing investigation.
Approximately 87% (n = 452) of the differentially expressed genes were upregulated by the expression of CNPase, with 9 genes that were similarly upregulated in the Drna mutant (Data Set S1) (26). Several of the most highly upregulated genes are involved in acid resistance, discussed in greater detail below. Others are putatively involved in attachment, such as sfmA (5.40-fold) and sfmH (3.17-fold), which encode FimA (type 1 fimbrial protein) homologs, and several genes that are predicted to encode fimbria-like adhesin proteins (n = 15; yadN had the highest fold change at 4.56-fold) or are prophage genes of unknown or unconfirmed function (Table 1; see also Data Set S1).
Gene ontology (GO) analysis revealed that 29,39-cNMPs modulate genes encoding diverse protein classes; frequent GO terms for primary targets included transcription factor, transporter, hydrolase, transferase, lyase, and oxido-reductase (Fig. 1A). 29,39-cNMPs also could be linked to the regulation of genes that impact numerous cell functions, including biofilm formation, motility, and multiple stress responses. These results informed a series of physiological assays to explore phenotypic changes as a response to altering the cyclic nucleotide pools.
29,39-cNMPs influence flagellar motility in E. coli. Expression of CNPase led to a slight increase in the mRNA levels of the flagellar-associated genes fliR (2.39-fold), which encodes a flagellar export protein, and fliE (3.48-fold), which encodes part of the flagellar basal body (Table 1). In contrast to the Drna strain, which exhibited significant upregulation of genes related to flagellum biosynthesis and chemotaxis (24,26), decreasing the 29,39-cNMP concentration through expression of CNPase did not alter abundance of other flagellumor chemotaxis-related transcripts.
To compare the phenotypic consequences of the altered motility gene expression profiles, we assayed the effect of 29,39-cNMPs on flagellum-dependent swimming motility. In agreement with the increased expression of chemotaxis and motility genes in the Drna mutant relative to that in the WT, previous work had demonstrated that the RNase I-deficient mutant was hypermotile (30) ( Fig. 2A and B). However, hydrolysis of 29,39-cNMPs in WT cells expressing CNPase does not alter cell motility significantly compared to that of WT cells expressing CNP-inact, indicating that the increased expression of fliR and fliE is insufficient to confer hypermotility in the presence of RNase I (Fig. 2C). It is of note that the E. coli WT strain BW25113 is known to have poor motility, and quantitative Western blot analysis demonstrated that flagellin (FliC) levels, which reflect late-stage flagellar synthesis, are undetectable in this strain (26); the baseline low level of flagellar biosynthesis may account for the lack of detectable transcriptomic or phenotypic effects by CNPase (31). In contrast, the Drna strain exhibits enhanced FliC levels compared to those of the WT parent strain (26). While CNPase is a useful tool for examining phenotypes that arise from depleting 29,39-cNMP pools, we also developed methods that allow us to increase 29,39-cNMP levels and probe the effects in the E. coli Drna strain, which produces flagella and is motile. Cells were treated with 59-O-butyryl 29,39-cAMP (Bt-cAMP), 59-O-butyryl 29,39-cUMP (Bt-cUMP), or 59-Obutyryl 29,39-cGMP (Bt-cGMP), which were synthesized based on existing lipophilic 39,59-cAMP and -cGMP ester analogs used in eukaryotic systems to facilitate access of the cyclic nucleotides into the cytoplasm (described in the supplemental material) (32). The addition of the cell-permeable cyclic nucleotide derivative Bt-cGMP to the growth medium resulted in observable increased 29,39-cGMP levels (Fig. S1), which are comparable to the endogenous level in WT E. coli at mid-exponential growth (24); Bt-29,39-cGMP was chosen to quantify permeability and hydrolysis of the butyryl group due to the low limits of quantification and detection. Addition of Bt-29,39-cAMP resulted in attenuated motility of the E. coli Drna strain in a dose-dependent manner without affecting the motility of the WT control strain ( Fig. 2A and B), demonstrating that 29,39-cNMPs modulate motility in a hypermotile E. coli strain deficient for endogenous 29,39-cNMPs. Importantly, neither Bt-cAMP nor Bt-cUMP inhibited cell growth ( Fig. S2A and B), and sodium butyrate (NaBt), the product of 59-O-ester  To further probe the molecular basis for how increased 29,39-cNMP levels alter the motility of the Drna strain, the abundance of FliC was quantified by Western blot analysis in WT and Drna strains grown in the presence and absence of Bt-cAMP. The Drna strain exhibited a hyperflagellated phenotype, as previously described (26), but treatment with Bt-cAMP and Bt-cUMP did not result in decreased FliC levels (Fig. 2D), despite their inhibitory effects on motility ( Fig. 2A and B). These findings demonstrate that exogenously increasing 29,39-cNMP concentrations impacts motility without altering the flagellar biosynthetic pathway that leads to FliC expression, indicating that 29,39-cNMPs attenuate motility in the Drna mutant through alternative regulatory mechanisms such as modulation of chemotaxis signal transduction.
29,39-cNMPs modulate biofilm formation. Our previous work demonstrated that the Drna mutant strain displays a hyperbiofilm phenotype due to increased expression of the Curli fiber genes csgBAC (24); Curli is the major proteinaceous component of the Salmonella Typhimurium (C and D) expressing CNPase compared to those in strains expressing CNP-inact. Pie charts were generated using the Panther Gene Ontology Database (http://pantherdb.org/). Differentially regulated protein classes were determined based on genes that are upregulated by expression of CNPase (A and C) and genes that are downregulated by expression of CNPase (B and D). Gene expression data were obtained from analysis of 3 biological replicates that were significantly differentially expressed (adjusted P [P adj ] , 0.05; .2-fold change in expression). extracellular matrix generated by E. coli and other members of the Enterobacteriaceae family (33). Biofilm production was likewise increased in WT cells expressing CNPase (24). The RNA-seq results for transcriptomic changes in cells expressing CNPase indicates that depletion of 29,39-cNMPs causes upregulation of Curli assembly gene csgC (3.74-fold) and type IV pilin assembly gene hofC (2.06-fold), along with several Yad fimbrial adhesion genes (yadCKLMN) ( Table 1). These data provide insight into the transcriptional mechanisms underlying the increased biofilm production upon CNPasemediated 29,39-cNMP hydrolysis and demonstrate that 29,39-cNMP depletion is sufficient to increase Curli gene expression, even in the presence of RNase I.
Intriguingly, CNPase expression induced transcription of the pgaABCD operon responsible for poly-N-acetyl-b-1,6-D-glucosamine (PNAG) biosynthesis in RNase I 1 E. coli (Table 1), whereas an Drna mutation dampened expression of the PNAG biosynthetic cluster (24). Transcription from the pga promoter is repressed by OmpR (34), and CNPase expression resulted in a modest decrease in ompR transcript abundance (1.8fold), while rna deletion did not affect the ompR mRNA level (26). These data suggest that the differential modulation of PNAG biosynthesis in WT cells expressing CNPase versus Drna cells could be regulated, at least in part, by OmpR.
To further evaluate the functional role of 29,39-cNMPs in biofilm formation, cultures of WT and Drna strains were treated with Bt-cAMP. Relative to untreated control cultures, the Drna mutant produced ;30% less biofilm in the presence of Bt-cAMP, while WT biofilm formation was unaffected (Fig. 2E), supporting the conclusion that 29,39-cNMPs regulate biofilm formation independently of RNase I. As mentioned above, Bt-cAMP did not affect the growth of either strain (Fig. S1). Moreover, sodium butyrate (NaBt) did not perturb biofilm formation (Fig. S5), demonstrating that the observed biofilm inhibition in Drna results from exogenously increasing the intracellular 29,39-cAMP concentration in E. coli lacking both endogenous 29,39-cNMPs and RNase I. 29,39-Cyclic phosphate-containing molecules regulate acid resistance. Expression of CNPase upregulates transcription of multiple genes involved in acid resistance systems (35,36), including the genes for glutamate decarboxylase gadA (3.91-fold) and gadB (7.44-fold), glutamate/gamma-aminobutyric acid antiporter gadC (4.35-fold), arginine/agmatine antiporter aidC (2.68-fold), lysine decarboxylase cadA (3.34-fold), cadC transcriptional activator for cadBA (3.28-fold), acid stress chaperones hdeB (6.00-fold) and hdeA (4.78-fold), and acid resistance membrane protein hdeD (4.28-fold) ( Table 1). The acid tolerance of WT cells expressing CNPase or CNP-inact was evaluated, and the data revealed a 10-fold increase in the survival rate at pH 2.5 upon CNPase expression compared to that of cells expressing CNP-inact ( Fig. 2F), suggesting that E. coli acid tolerance is enhanced by the regulatory impact of reducing the levels of 29,39-cNMPs and/or oligoribonucleotides with 39-terminal 29,39-cyclic phosphate (26). Our previous work demonstrated that rna gene deletion reduced transcript levels of gadX and gadY, which encode positive transcription regulators of the acid response genes gadAB and gadC (37), and concomitantly decreased survival in acidic medium relative to that of the WT strain (26). The differences in the acid resistance phenotype and regulation of genes involved in acid tolerance in the Drna strain versus the CNPase-expressing WT strain may reflect disparate effects of RNase I depletion versus 29,39-cNMP hydrolysis, or it may indicate that the complex regulatory network for acid stress response includes the other physiologically relevant CNPase substrate, oligoribonucleotides with 39-terminal 29,39-cyclic phosphates (38,39).
29,39-cNMP generation and physiological impacts in Salmonella Typhimurium. Like E. coli, S. Typhimurium is a member of the Gram-negative Enterobacteriaceae family. E. coli and S. Typhimurium diverged from a common ancestor approximately 120 million years ago (40); phenotypic differences between these species are not only due to genetic differences, including .800 E. coli genes that are absent in the S. Typhimurium genome and .1,100 S. Typhimurium genes that are absent in the E. coli genome, but also due to differential regulation of conserved genes (41). Due to these two organisms' evolutionary closeness yet significant phenotypic differences, S. Typhimurium was chosen as the first candidate for comparison in order to determine whether bacteria have universal mechanisms for generating 29,39-cNMPs and if there are physiological commonalities when 29,39-cNMP pools are altered.
To determine whether 29,39-cNMP levels fluctuate in S. Typhimurium in a growth phase-dependent manner, samples were collected at the mid-log, late log/early stationary, and late stationary phases during growth in M9 medium. 29,39-cAMP, -cGMP, -cCMP, and -cUMP were most abundant during logarithmic growth, as was seen in E. coli (24); however, levels of all of the nucleotides decreased over the growth curve in S. Typhimurium (Fig. 3A). The variation in 29,39-cNMP levels at different growth phases was at most ;5-fold, which is in stark contrast to the approximately 1,000-fold drop in the four 29,39-cNMP concentrations in E. coli during early stationary growth. Like E. coli, S. Typhimurium expresses an RNase I homolog encoded by the rna gene. RNase I is essential for the generation of 29,39-cNMPs in E. coli (24); to determine whether the same is true for S. Typhimurium, 29,39-cNMPs were quantified in an Drna strain under the same growth conditions as the WT strain. The cyclic nucleotides were below the level of detection at all stages of growth in the S. Typhimurium Drna strain, confirming that 29,39-cNMP generation in S. Typhimurium is likewise facilitated by RNase I (Fig. 3A).
The effects of 29,39-cNMPs on modulation of gene expression in S. Typhimurium were probed using RNA-seq analysis with S. Typhimurium 14028s WT strains expressing CNPase from pKT-CNP or CNP-inact from pKT-CNP-inact under growth conditions identical to those used for the E. coli RNA-seq analysis. Quantification of 29,39-cNMPs by liquid chromatography-tandem mass spectrometry (LC-MS/MS) for the three biological replicates of each strain used in the RNA-seq analysis confirmed that the 29,39-cNMP levels were significantly reduced when cells expressed active CNPase compared to those in the inactive variant (Fig. 3B). Comparative analysis of RNA-seq values for the CNPase versus CNP-inact strains indicated that 155 genes were .2-fold downregulated and 14 genes were .2-fold upregulated when active CNPase was expressed (correlating to a decrease in 29,39-cNMP levels) (Data Set S2). The overall number of dysregulated genes and the functions of many of those genes differed significantly between S. Typhimurium and E. coli expressing CNPase ( Fig. 1 and Table 1). Dysregulated genes in S. Typhimurium included multiple genes involved in extracellular polysaccharide production, flagellar assembly, heat shock response, peroxide stress response/SOS response, and RNA repair (Table 1); the physiological impacts of the dysregulation of these gene are addressed below. Approximately half of the 155 downregulated genes in S. Typhimurium expressing CNPase (n = 82) were genes carried by the prophages Gifsy-1, Gifsy-2, Gifsy-3, and ST64B, suggesting a potential role for 29,39-cNMPs in regulation of prophage gene expression that is activated by nucleic acid damage (42)(43)(44). This is consistent with the CNPase-dependent downregulation of genes involved in responses to nucleic acid damage, i.e., SOS response and RNA repair genes listed in Table 1 (45,46).
Only two specific genes showed consistent CNPase-dependent dysregulation in E. coli and S. Typhimurium, and both were related to the heat shock response. These were dnaJ (downregulated 2.15-fold in E. coli and 3.10-fold in S. Typhimurium), which encodes a member of the Hsp40 family of cochaperones that, with the chaperone DnaK, is essential for recovery from heat shock (47,48), and ibpB (downregulated 2.60fold in E. coli and 7.95-fold in S. Typhimurium), which encodes a subunit of the IbpAB small heat shock protein complex (49) ( Table 1). The following additional genes related to the heat shock response were downregulated in each species: dnaK (2.53-fold) and ybbN (2.05-fold) in E. coli and ibpA (7.89-fold) in S. Typhimurium (Table 1). Together, these data may suggest a role for 29,39-cNMPs in regulating the production of certain heat shock chaperones. To determine whether reduction of 29,39-cNMP levels influences the heat shock response in S. Typhimurium, recovery from heat shock stress was assayed for 14028s WT, CNPase, and CNP-inact expression strains. In addition, survival of peroxide-induced oxidative stress was assayed, as it was recently shown that DnaJ interacts with the RNA polymerase binding partner DksA in a redox-dependent manner during oxidative stress, and deletion of dnaJ increases susceptibility of S. Typhimurium to hydrogen peroxide-mediated killing (50). The results of the heat shock and oxidative stress survival assays indicate that the level of decreased expression of dnaJ and ibpAB in CNPase-expressing S. Typhimurium is not sufficient to significantly impact survival of heat shock or peroxide-induced oxidative stress ( Fig. 4A and B).
The genes most affected by CNPase activity in S. Typhimurium were genes of the RNA repair operon (rsr-rtcB-rtcA), which were significantly downregulated, as follows: rsr (12.75-fold), rtcB (13.94-fold), and rtcA (14.51-fold) ( Table 1). This operon is transcriptionally controlled by RtcR, a member of the bacterial enhancer-binding protein family of regulators that becomes activated in the presence of nucleic acid-damaging conditions, e.g., treatment with mitomycin C (MMC) or hydrogen peroxide (45). Expression of the RNA repair operon of E. coli (rtcB-rtcA) was not significantly altered by CNPase activity; this result is consistent with previous work demonstrating that the regulation of the RNA repair operons in E. coli and S. Typhimurium responds to different stress conditions (45). To further assess expression of the S. Typhimurium RNA repair operon when intracellular levels of 29,39-cNMPs are reduced by CNPase, promoter activity was measured using a previously developed reporter assay that utilizes a transcriptional fusion of the RNA repair operon promoter with xylE (14028s Drsr::xylE) (45). RtcR-dependent expression of the RNA repair operon was induced by treatment with MMC. The WT reporter strain (14028s DrsrI::xylE) showed a 6-fold increase in RNA repair operon expression compared to that in the untreated strain, consistent with previously published results (Fig. 4C) (45). Results were similar when strains expressed the CNPinact variant; conversely, when active CNPase was expressed, XylE activity was reduced to nearly the same level as that in untreated cells (Fig. 4C), confirming that the catalytic activity of CNPase significantly impacts expression of the RNA repair operon in S. Typhimurium. This was additionally confirmed for WT 14028s using reverse transcription-quantitative PCR (qRT-PCR) to quantify the levels of rtcB transcript; the same trends emerged, showing 12-fold and 10.5-fold increases in rtcB transcript levels in the MMC-treated WT and CNP-inact expression strains, respectively, that was abolished in the CNPase expression strain (Fig. S6). These results indicate that CNPase degrades a signal for activation of RtcR, thereby inhibiting initiation of transcription of the RNA repair operon; however, surprisingly, XylE expression in a Drna reporter strain (14028s Drna Drsr::xylE) that eliminates production of 29,39-cNMPs (Fig. 3A) was similar to WT levels (Fig. 4C).
To determine whether there is potentially another source of 29,39-cNMPs during this treatment in S. Typhimurium, 29,39-cNMP levels were quantified in MMC-treated and untreated cells. The concentration of each of the 29,39-cNMPs in the Drna strain remained near the limit of detection for both treated and untreated cultures (Fig. 4D). Additionally, there was no significant difference in 29,39-cNMP levels when comparing MMC-treated versus untreated cultures in the WT, WT pBAD33-CNP, or pBAD33-CNPinact strains (Fig. 4D). Unlike the dramatic reduction in 29,39-cNMP levels seen with expression of CNPase from pKT-CNP (Fig. 3B), expression of CNPase from the pBAD33-CNP plasmid (induced by 0.02% arabinose) did not significantly decrease 29,39-cNMP levels (Fig. 4D); however, this same level of CNPase expression from pBAD33-CNP significantly reduced RtcR-dependent expression of the RNA repair operon (Fig. S6). These results indicate that activation of the RNA repair operon in S. Typhimurium cannot be clearly linked to fluctuating 29,39-cNMP levels. We attempted to further explore whether increasing the 29,39-cNMP levels would activate expression of the RNA repair operon in the absence of nucleic acid damage; to this end, the reporter strain (14028s Drsr::xylE-kan) was treated with cell-permeable Bt-cAMP, and expression of the operon was measured at multiple time points over a 25-h period to allow time for intracellular hydrolysis of the butyryl group from Bt-cAMP to release 29,39-cAMP (Fig. S7). At the 1.5-, 3-, 6-, and 10-h time points, expression of the operon was unchanged for cultures treated with Bt-cAMP, 29,39-cAMP, sodium butyrate, or water; at the 15-and 25-h time points, expression increased nearly 10-fold, but this increase was independent of the treatment and is likely due to cellular stress in the late stationary phase (.12 h after reaching the stationary phase) (Fig. S7). Since Bt-cAMP uptake and hydrolysis to 29,39-cAMP in S. Typhimurium is likely to be similar to that shown for Bt-cGMP in E. coli (Fig.  S1), these results support the conclusions that altered 29,39-cNMP levels are not the primary signal for RtcR activation in S. Typhimurium and the CNPase substrate that effects RtcR activation and expression of the RNA repair operon is an oligoribonucleotide with a 39-terminal 29,39-cyclic phosphate, which is further addressed below.
Eight genes of the wca colanic acid biosynthesis operon were downregulated in S. Typhimurium expressing active CNPase (ranging from 2.77-fold to 4.39-fold) ( Table 1). Colanic acid is an extracellular polysaccharide (EPS) whose function is not well understood, though research has suggested a role in biofilm or capsule formation and maintenance of the transmembrane electrical potential in the environment (51). Two additional genes, yjbE and yjbF, which are involved in the synthesis of noncolanic acid EPS, were also downregulated (2.98-fold and 2.40-fold, respectively) ( Table 1). Both of these EPS pathways were recognized as part of the Rcs regulon (51), suggesting that reduction of 29,39-cNMPs may impact Rcs-mediated EPS biosynthesis. A Congo red assay was performed to determine whether EPS levels are impacted during expression of CNPase in S. Typhimurium. The strains expressing CNPase or CNP-inact showed reduced levels of EPS compared to those in the WT strain, with the inactive variant exhibiting a significant decrease of 2.4-fold; however, the EPS levels were not significantly different between the strains expressing the active and inactive CNPase variants, which suggests that the 29,39-cNMP pools did not substantially alter EPS biosynthesis (Fig. 4E). A variety of single-gene deletion mutant controls (Drna, DwcaI, DyjbE, and DcsgD, a regulator of biofilm formation) additionally showed no significant change in EPS levels compared to those of the WT (Fig. 4E); this indicates that, under the growth conditions used, the quantifiable levels of EPS do not significantly change when expression of a single biosynthetic pathway is altered, although the composition of the EPS could potentially be affected.
Of the 14 genes that were upregulated by CNPase, the following four are related to flagellar biosynthesis: flgBCD, which are components of the flagellar basal body (5.09-, 5.04-, and 4.40-fold, respectively), and flgJ, which encodes a flagellum-specific peptidoglycan hydrolase (3.04-fold) ( Table 1). As noted above, CNPase likewise affected the expression of genes related to the flagellar basal body in E. coli. To determine if a motility phenotype could be observed in S. Typhimurium, the same swimming motility assay was performed, although, unlike the E. coli parent strain, 14028s is highly motile. The Drna strain was included to determine if complete abolition of 29,39-cNMPs results in a hypermotile phenotype, as observed in E. coli. In these assays, at the 10 h time point, the WT and Drna strains exhibited equivalent motility, but the WT strains expressing either CNPase or CNP-inact showed significantly reduced motilities relative to that of the WT strain without the plasmids (P , 0.001) (Fig. 4F). At the 24-h time point, motility halos could not be measured for the WT and Drna strains that did not contain the CNPase/CNP-inact plasmids because the halos had already reached the edges of the motility plate; the WT strain expressing CNP-inact exhibited reduced mo- the pKT vectors significantly reduced WT motility relative to that of the WT without the plasmid; the Drna strain did not exhibit significantly altered motility compared to that of the WT. Except where otherwise stated, all data are representative of at least 3 biological replicates; error bars represent 61 standard deviation. Statistical significance was determined by a one-way ANOVA with Bonferroni and Holm used as post hoc tests to determine significance between samples. A P value of ,0.001 (***) or ,0.01 (**) was considered statistically significant. tility relative to that of the strain expressing the active CNPase (P , 0.01) (Fig. 4F). Overall, these results indicate that reduction in 29,39-cNMP levels by deletion of RNase I does not impact flagellar motility in S. Typhimurium and that CNPase hydrolysis of its cellular substrates also does not directly affect motility, since expression of either active CNPase or CNP-inact reduced motility. The effects of the pKT-CNP and pKT-CNPinact expression plasmids on cell growth are addressed below. Based on RNA-seq data, no S. Typhimurium genes related specifically to chemotaxis, rather than flagellar assembly, were altered in their expression by CNPase activity (Data Set S2), which could explain why motility is affected differently in E. coli versus S. Typhimurium.
Taken together, these data demonstrate that 29,39-cNMPs and other RNA oligomers with terminal 29,39-cyclic phosphates differentially affect both transcriptional regulation and physiology of even closely related species. This suggests that different organisms may monitor their intracellular levels of 29,39-cNMPs and other 29,39-cyclic phosphatecontaining molecules and modulate different behaviors in response to this input.
Effects of hydrolysis of CNPase substrates other than 29,39-cNMPs. It has previously been demonstrated that CNPase cleaves the terminal cyclic phosphate on oligoribonucleotides in addition to targeting cyclic nucleotide monomers (38,52); both of these CNPase substrates are physiologically relevant in bacteria (39). To determine whether the effects of CNPase on the E. coli and Salmonella transcriptomes are due to 29,39-cNMPs vrsus 29,39-cyclic phosphates on oligoribonucleotides, we performed RNAseq of E. coli Drna and S. Typhimurium Drna strains expressing either CNPase or CNPinact and compared the results to those from WT-CNPase versus WT-CNP-inact.
Analysis of the transcriptomes for E. coli Drna expressing CNPase and CNP-inact revealed 178 differentially expressed genes, of which 38 are also dysregulated by CNPase expression in the WT strain (Data Set S3), with 33 genes exhibiting similar dysregulation; 13 of these 33 genes are involved in multiple acid stress response systems (Table 1). This result suggests that CNPase-mediated hydrolysis of RNA oligomers with 29,39-cyclic phosphate termini rather than 29,39-cNMPs is responsible for increased expression of acid stress response genes (Table 1) and a higher level of acid resistance (Fig. 2F) in the WT strain expressing CNPase. It should be noted that this result is consistent with our previously reported analysis of the E. coli Drna strain, which showed decreased expression of gadX and gadY of the glutamate-dependent acid resistance system and decreased acid resistance (26); there are a number of regulatory circuits for the various acid tolerance systems (53) that could be differentially responsive to 29,39cyclic phosphate-containing molecules. Overall, this analysis of the transcriptomes for E. coli Drna strains expressing CNPase and CNP-inact demonstrates that only 38 of the 519 genes identified as dysregulated by CNPase expression in the WT strain are dysregulated by CNPase in the absence of 29,39-cNMPs (Data Set S3).
In the S. Typhimurium Drna strain, 669 genes exhibit CNPase-dependent differential gene expression, but only 27 of these genes are similarly dysregulated by CNPase expression in the WT strain, which include the flagellar assembly genes flgB, flgC, flgD; RNA repair genes STM14_4239, rtcB, and rtcA; heat shock response genes dnaJ, ibpB and ibpA; and SOS response genes recN and yebG (Table 1; see also Data Set S4 in the supplemental material). The CNPase-dependent differential expression of these 27 genes in both the presence (WT strains) and absence (Drna strains) of 29,39-cNMPs suggests that oligoribonucleotides with 39-terminal 29,39-cyclic phosphates are involved in their regulation. Recent work by Hughes et al. (54) shows that the regulatory ligand that activates RtcR, thereby stimulating transcription of STM14_4239, rtcB, and rtcA from the RNA repair operon promoter, is a tRNA fragment ending with a 29,39-cyclic phosphate. The CNPase-dependent dysregulation of the RNA repair operon in both the WT and Drna strains, as well as the results of our RtcR activation assays ( Fig. 4C and Fig. S6 and S7), are consistent with CNPase-mediated cleavage of the 29,39-cyclic phosphate moiety of these cleaved tRNAs preventing activation of RtcR. An additional 13 genes exhibited dysregulation in both Drna-CNP/inact and WT-CNP/inact strains, but with opposite regulatory effects; these included mostly prophage genes (Data Set S4).
RNA-seq analysis of S. Typhimurium WT and Drna strains without the CNPase/CNPinact expression plasmids revealed 149 genes that are differentially expressed in the absence of RNase I (Data Set S5); none of these 149 genes overlap the gene set that is dysregulated by CNPase expression in both the WT and Drna strains. This result supports that the CNPase substrate responsible for the altered expression of the RNA repair genes, as well as the other similarly dysregulated genes between the WT and Drna CNPase/CNP-inact strains (Table 1 and Data Set S4), is an oligoribonucleotide with a 39-terminal 29,39-cyclic phosphate.
To further assess physiological effects of CNPase or CNP-inact expression from the pKT vectors in E. coli and S. Typhimurium, growth curves were performed with both WT and Drna strains (Fig. S8); the Fig. S8 legend includes the calculated doubling times and statistical analyses. For E. coli and S. Typhimurium WT and Drna strains, there was a small, but significant, increase in the doubling times for all but one of the strains expressing CNPase or CNP-inact relative to those for the strains without the pKT vectors. However, there was no significant difference between the doubling times for strains expressing CNPase and CNP-inact. These results indicate that the effects of the pKT-CNP or pKT-CNP-inact plasmids on E. coli and S. Typhimurium growth rates are independent of the catalytic activity of CNPase. In addition, the approximately 2-fold reduced growth rate for WT S. Typhimurium expressing CNPase or CNP-inact relative to that of the WT strain is consistent with the reduced flagellar motility for the WT strain expressing either CNPase or CNP-inact (Fig. 4F); since both the active and inactive CNPase affect S. Typhimurium growth and motility, hydrolysis of 29,39-cyclic phosphates does not cause the phenotype. The effect of CNPase/CNP-inact on E. coli and S. Typhimurium growth may be due to binding of substrate molecules, thereby preventing their regulatory activities or turnover. Consistent with this hypothesis, CNPase and CNPinact exhibit the same K m values but very different k cat values (27); in addition, CNPase has been shown to bind eukaryotic poly(A) 1 mRNA independent of its catalytic activity (29).
29,39-cNMPs in other organisms. Using literature review along with metabolomics data within a number of publicly available databases, 29,39-cNMPs were identified in data sets from Staphylococcus aureus, Pseudomonas fluorescens, Mus musculus, and Arabidopsis thaliana, as well as within human tissues ( Table 2). All four of the 29,39-cNMPs were not reported in each data set; however, the data were often from untargeted metabolomics runs where 29,39-cNMPs were not specifically under investigation (60,66,87). In other cases, only certain 29,39-cNMPs were targeted for investigation (17,21,64,65,85,88). All of the data sets utilized either tandem mass spectrometry (MS/ MS), in which the diagnostic fragments allow unambiguous identification of 29,39-cNMPs, or LC-MS (without additional fragmentation), which allows for the identification of 29,39-cNMPs due to retention time.

DISCUSSION
Nucleotide signaling mediates responses to various environmental stimuli in diverse bacterial taxa, orchestrating complex processes such as the motility-sessility transition, sporulation, and virulence factor production (7). The present work utilizes chemical and enzymatic perturbation of in vivo 29,39-cNMP levels in coordination with global transcriptome profiling and phenotypic analyses to characterize the function of 29,39-cNMPs in bacterial nucleotide signaling. These studies have identified multiple roles for 29,39-cNMPs, as well as for oligoribonucleotides with 29,39-cyclic phosphate termini, in bacterial physiology within distinct cellular processes.
While the mechanism through which 29,39-cNMPs govern specific cell processes is a current subject of investigation, it is possible that regulation occurs in part through direct interactions of the nucleotides with nucleotide-binding effectors. There is precedence for this hypothesis, as recent work identified the polyadenylate-binding protein Rbp47b in A. thaliana as the first known 29,39-cNMP-binding effector in any organism (17). Rbp47b serves as a scaffold for assembly of the multiprotein stress granule complex, and this assembly is facilitated by 29,39-cAMP binding to Rbp47b (17). Intriguingly, Rbp47b also binds mRNA (67), suggesting that stress granule formation is modulated by the extent of mRNA decay in the cell, as degradation of poly adenylated mRNA likely increases the 29,39-cAMP concentration. These findings allude to the existence of other effectors in diverse organisms that sense RNA-derived 29,39-cNMPs as reporters of cellular stress. Another possibility is that 29,39-cNMPs play an indirect role in gene regulation by affecting nucleotide metabolism, influencing the concentrations of various nucleotide pools that themselves regulate a subset of genes.
Our previous work demonstrated that, in E. coli, all detectable 29,39-cNMPs are generated by the exoribonuclease RNase I (encoded by rna). Interestingly, in this work, we found that expressing CNPase to directly target the cyclic nucleotide pools leads to the dysregulation of many transcripts that were not previously found to be perturbed by deletion of rna (26), indicating that RNase I elicits cellular effects that are separate from those caused specifically by 29,39-cNMP fluctuations. One striking example is the Trp biosynthetic cluster trpDCBA, which was ;5to 7-fold downregulated in RNase I 1 E. coli expressing CNPase but was not altered by rna gene deletion. A possible explanation for these CNPase-specific effects is altered flux through 29,39-cNMP catabolic pathways, as CNPase expression inevitably perturbs other nucleotide/side pools due to 29,39-cNMP hydrolysis. In addition, CNPase cleaves terminal phosphates of cleaved regulatory oligonucleotides that affect transcript levels. Conversely, catabolism of 29,39-cNMPs does not occur in the Drna strain, as this strain lacks detectable 29,39-cNMP concentrations (24), which likely alters nucleotide salvage pathways.
In addition to serving as key components of cellular metabolism, certain nucleotides are established modulators of bacterial chemotaxis and motility. Here, we identified 29,39-cNMPs as a novel component of nucleotide signaling in motility regulation. RNA-seq data link 29,39-cNMPs to transcriptional regulation of intracellular components of chemotaxis and flagellar assembly via a failure to generate 29,39-cNMPs (Drna) (Fig.  2F) or via depletion of existing 29,39-cNMP pools (CNPase). The effects of CNPase expression on motility-related genes were minor and were echoed in results of a soft agar motility assay, with the WT strain expressing CNPase showed no change in motility. However, the hypermotile Drna strain partially attenuated motility due to the addition of cell-permeable 29,39-cNMP analogs to the medium in a manner that does not affect FliC abundance ( Fig. 2A, B, and E). We previously determined that 29,39-cNMPs cannot traverse the E. coli cell envelope (24) and that 29,39-cAMP has no effect on swimming motility (see Fig. S5 in the supplemental material), indicating that 29, 39-cNMPs do not function as exogenous chemoattractants or -repellents, as these processes require transport across the outer membrane and subsequent binding to periplasmic methyl-accepting chemotaxis proteins (MCPs) (68). Together, these data suggest that 29,39-cNMP levels impact motility at an as-yet-unidentified stage of biosynthesis or chemotaxis, which may become clearer upon repeating these experiments in a more motile strain of E. coli.
Nucleotide metabolism and c-di-GMP signaling govern biofilm production, in addition to regulating flagellar motility. Previous phenotypic assays and global gene expression analyses with WT and Drna E. coli strains demonstrated that 29,39-cNMPs also function in biofilm regulation (24), thus identifying a novel nucleotide pool involved in biofilm formation. The present work provides further insight into the role of 29,39-cNMPs in biofilm production through the discovery that decreasing 29,39-cNMP levels upregulated expression of Curli assembly gene csgC (Fig. 3B), and treatment with Bt-cAMP attenuated the increased biofilm phenotype in RNase I-deficient E. coli (Fig. 3A). Perturbation of 29,39-cNMP levels or deletion of the RNase I gene additionally affected transcription of genes involved in nucleotide metabolism, establishing a link to prior studies that indicated that perturbation of de novo pyrimidine or purine nucleotide biosynthesis alters biofilm morphology in E. coli (11)(12)(13). Indirect modulation of c-di-GMP biosynthesis functions as one mechanistic link between primary nucleotide metabolism and biofilm formation, as antimetabolite-mediated disruption of de novo purine synthesis decreases the concentration of c-di-GMP and impairs biofilm production in E. coli (13). Therefore, the finding that disrupting 29,39-cNMP concentrations dysregulated the expression of purine and pyrimidine nucleotide metabolic genes suggests that 29,39-cNMPs modulate nucleotide pools and can perturb c-di-GMP signaling, which likely mediates the biofilm phenotype in WT cells expressing CNPase. The expression of several genes encoding diguanylate cyclases (DGCs) and c-di-GMP phosphodiesterases (PDEs) also was altered by CNPase expression, further suggesting that 29,39-cNMPs influence biofilm production through modulation of distinct c-di-GMP metabolic enzymes. Although previous investigations in our group determined that the total concentration of c-di-GMP is not perturbed in E. coli lacking RNase I (24), modulation of individual DGCs or c-di-GMP PDEs can impact E. coli biofilm production without altering the global abundance of c-di-GMP (69). The emerging regulatory functions for 29,39-cNMPs in biofilm formation could inform the discovery of novel adjuvants to remedy chronic, biofilm-associated microbial infections.
Characterization of the generation of 29,39-cNMPs in S. Typhimurium revealed basic similarities with E. coli, including the essential role of RNase I in production of 29,39-cNMPs and the overall highest level of 29,39-cNMP expression during mid-logarithmic growth. However, there was little overlap in the cellular responses to altered levels of 29,39-cNMPs between these two highly related species; therefore, it is likely that investigation of the roles of 29,39-cNMPs in other microbes will reveal diverse regulatory proteins and pathways that respond to 29,39-cNMPs. In addition, it is now apparent that oligoribonucleotides with terminal cyclic phosphates also have regulatory activities, as demonstrated for RtcR activation and expression of the RNA repair operon (54), thus expanding the repertoire of cyclic nucleotides that control stress responses of bacteria.
In addition to our identification of 29,39-cNMPs in E. coli and S. Typhimurium, a search of publicly available metabolomics databases yielded hits for 29,39-cNMPs in additional bacterial species, as well as in eukaryotes. While some of the species have an RNase I or RNase T2 homologue, such as C. elegans and humans, not all have a known RNase T2 homologue (M. genitalium), suggesting that other RNase families that use a similar two-step RNA hydrolysis mechanism may serve as the source for 29, 39-cNMPs. Given previous work demonstrating a link between 29,39-cAMP and organ damage (18,70) in mammals and between 29,39-cAMP/2¨,39-cGMP levels and wounding in plants (64), it seems likely that 29,39-cNMP levels are broadly used by organisms to modulate their environmental responses.
Here, we identify 29,39-cNMP pools as novel components of bacterial signal transduction with implications in pathogenesis, expanding the scope of nucleotide signaling beyond the paradigmatic 39,59-cNMPs and c-di-NMPs. Transcriptional profiling and phenotypic investigations, in tandem with controlled manipulation of 29,39-cNMP levels, have identified bacterial processes regulated by 29,39-cNMPs, providing novel links between nucleotide metabolism and virulence-associated phenotypes such as motility and acid tolerance. Additional gene expression analyses and bioanalytical experiments suggest that these processes are mediated in part by dysregulated nucleotide homeostasis upon perturbation of 29,39-cNMP levels. Future experiments aim to identify potential 29,39-cNMP-binding effectors and provide additional mechanistic insight into the functions of RNase I and 29,39-cNMP pools in prokaryotic physiology. In addition, the recombinant CNPase and cell-permeable 29,39-cNMP derivatives developed here will enable dissection of processes linked to 29,39-cNMPs and T2 family RNases across the kingdoms of life.

MATERIALS AND METHODS
Bacterial strains, plasmids, general culture conditions, commercial chemicals, and statistical analyses. All relevant strains and plasmids used in this study are listed in Table S1 in the supplemental material. The RNase I-deficient E. coli strain (rna::kan Drna) in the BW25113 genetic background (lacI q rrnB T14 DlacZ WJ16 hsdR514 DaraBAD AH33 DrhaBAD LD78 ) was obtained from the Keio collection, along with the BW25113 WT strain (71,72). Locus-specific PCR amplifications were used to confirm disruption of the rna gene with the kan cassette, as reported previously (Table S2, oligonucleotides 1 to 4) (24). The WT S. Typhimurium strain used in this study is ATCC 14028s. Strains JEK20 (14028s Drna) and JEK30 (14028s DrecA) were generated by transduction of the WT strain with HT P22 lysate grown on appropriate single-gene deletion mutants from the BEI collection (described in (73). The cat or kan cassette, respectively, for selection of transductants was subsequently excised by transforming cells with pCP20 (72), and the deletion was confirmed by PCR with gene-flanking primers (Table S2, oligonucleotides 5 and 6 and oligonucleotides 7 and 8, respectively). Construction of the reporter strains JEK12 (14028s Drsr::xylE-kan) and JEK17 (14028s Drsr::xylE) was described previously (45). JEK29 (14028s Drna Drsr::xylE) was constructed similarly to JEK20, except the recipient strain for P22 transduction was JEK17. Strains 14028s DyjbE::kan, 14028s DwcaI::kan, and 14028s DcsgD::kan were obtained from the BEI SGD collection, and deletion-insertion mutations were confirmed by PCR with gene-flanking primers (Table S2, oligonucleotides 9 and 10, 11 and 12, and 13 and 14, respectively).
Bacteria were routinely cultured in Luria-Bertani (LB) medium or M9 minimal medium supplemented with 0.4% glucose and 0.2% Casamino Acids at 37°C with orbital shaking at 200 to 225 rpm, unless otherwise noted. Antibiotics were used at the following working concentrations unless otherwise noted: kanamycin (25 mg/ml), carbenicillin (100 mg/ml), and chloramphenicol (30 mg/ml). Plasmids pKT-CNP and pKT-CNP-inact (H73L/H152L), which encode bla for b-lactam-mediated selection, were obtained by restriction enzyme-based subcloning as previously described (24). Plasmids pKT-CNP(cat) and pKT-CNPinact(cat), encoding cat for chloramphenicol resistance, were constructed using polymerase incomplete primer extension (PIPE) cloning, using either pKT-CNP or pKT-CNP-inact as the vector template and pBAD33 as the template for the cat insert (Table S2, oligonucleotides 19 to 22) (74). For S. Typhimurium, plasmids pKT-CNP and pKT-CNP-inact were freshly transformed by electroporation prior to each experiment, as plasmids exhibited some toxicity. For qRT-PCR experiments, pBAD33-CNP and pBAD33-CNPinact were utilized as a less toxic alternative; for these, the Rattus norvegicus CNPase (rCNPase) catalytic domain was codon optimized and synthesized by GenScript (Piscataway, NJ) and subcloned into pET-28a within the NdeI and XhoI restriction sites. rCNPase was subcloned into pBAD33 using NdeI and XhoI for expression to make pBAD33-CNP. pBAD33-CNP-inact (H68L/H147L) was generated using the QuikChange site-directed mutagenesis kit (Agilent, Santa Clara, CA) with the provided primer sequences (Table S2, oligonucleotides 15 to 18).
Experiments utilized at least three biological replicates (n $ 3), and a two-sample t test was employed to assess statistical significance between two data sets, where equal or unequal variance was evaluated using an F test. A P value of ,0.05 was considered statistically significant. For larger data sets, statistical analysis was assessed through analysis of variance (ANOVA) to determine whether there was statistical significance, followed by post hoc Bonferroni and Holm tests to determine which groups were statistically significant. A P value of ,0.01 was considered statistically significant.
Transcriptome profiling of WT and Drna strains of E. coli and S. Typhimurium containing pKT-CNP or pKT-CNP-inact and of WT/Drna for S. Typhimurium. Overnight cultures for three biological replicates each of WT or Drna strains of E. coli BW25113 or S. Typhimurium 14028s containing pKT-CNP or pKT-CNP-inact, or WT and Drna strains of S. Typhimurium 14028s without plasmids, in M9 medium (0.4% glucose and 0.2% Casitone) with appropriate antibiotics were subcultured 1:100 in the same medium and grown at 37°C with shaking. When cultures reached an optical density at 600 nm (OD 600 ) of 0.1, expression of CNPase or CNP-inact was induced with 25 ng/ml anhydrotetracycline. Cultures were grown to an OD 600 of 0.5 to 0.6, and samples were collected for RNA-seq (and 29,39-cNMP quantification for the WT strains). Cells were pelleted by centrifugation, flash frozen, and stored at 280°C.
For the WT strains with the pKT plasmids, the frozen cell pellets were sent to the Emory University Yerkes Nonhuman Primate Genomics Core for RNA isolation, rRNA depletion, library construction and sequencing. Briefly, total RNA (2 mg) was extracted and subjected to rRNA depletion using the Bacterial Ribo-Zero rRNA removal kit (EpiCentre, Madison, WI) following the manufacturer's instructions. Complementary DNA (cDNA) libraries were prepared with the ScriptSeq v2 RNA-seq library preparation kit (Epicentre, Madison, WI, USA). The rRNA-depleted total RNA sample was fragmented using an RNA fragmentation solution, and the fragmented RNA was reverse transcribed into cDNA using random hexamer primers containing a tagging sequence at the 59 end; 39 tagging was accomplished using a terminal-tagging oligonucleotide (TTO) consisting of a random hexamer flanked by a 59 tag sequence and a blocked 39 terminus. The di-tagged cDNA was purified using AMPure XP beads (Agencourt, Beckmann-Coulter, USA), and PCR amplified to add index and sequencing adapters. After amplification, the final library was purified using AMPure XP beads, and the final pooled libraries were sequenced on the Illumina HiSeq 3000 system in a single-end (SE) 150-cycle format. Each sample was sequenced to approximate depth of 8 to 12 million reads. For data processing and statistical analysis, RNA-seq reads were aligned to the GenBank E. coli BW25113 genomic reference sequence (accession number CP009273.1) or GenBank Salmonella enterica subsp. enterica serovar Typhimurium 14028s sequence (CP001363.1) using the STAR aligner v2.5.2b (75), and transcript abundance was estimated using htseqcount v0.6.1p1 (76). Differential expression analysis was performed with DESeq2 (77). DESeq2 uses the Benjamini-Hochberg adjustment of P values for false-discovery rate (P adj ); log 2 fold change values for differential gene expression with P adj values of ,0.05 were considered significant. The NCBI index (ftp://ftp .ncbi.nlm.nih.gov/genomes/all/GCA/000/750/555/GCA_000750555.1_ASM75055v1) was utilized for gene annotation.
For the Drna E. coli strains with pKT plasmids, total RNA was isolated from the frozen cell pellets and treated with DNase I using the New England Biolabs (NEB) Monarch total RNA miniprep kit, following the manufacturer's instructions. For the WT (without pKT plasmids) and Drna (with and without pKT plasmids) strain S. Typhimurium frozen cell pellets, total RNA was isolated and treated with DNase I as previously described (45). Total RNA was sent to Microbial Genome Sequencing Center (MiGS); 300 ng of RNA was used for library preparation with Illumina Stranded Total RNA Prep with Ribo-Zero Plus, following the manufacturer's protocol. The libraries were run on an Illumina NextSeq 2000 instrument, providing .12 million paired-end reads (2 Â 50 bp). MiGS performed analysis as follows. Sequencing data were processed with Illumina bcl2fastq2 conversion software; read mapping and quantification were performed with HISAT2 (78) and the feature "Counts" in the Subread package (79), respectively, using the same reference genomes as described above for the WT strain RNA sequencing. Read counts were normalized using the edgeR (80) trimmed mean of M values (TMM) algorithm. Differential expression analysis was performed using the edgeR quasilinear F test (glfTest) functionality against treatment groups; genes with an absolute log 2 fold change (log 2 FC) of .1 and a P adj value (FDR) of ,0.05 were considered differentially expressed. Principal-component analysis (PCA) and differentially expressed gene heat maps indicated that one of the three biological replicates for Drna-CNPase (deltaCNP1) was an outlier in the RNA-seq analysis of E. coli Drna strains containing pKT-CNP or pKT-CNP-inact. The data for deltaCNP1 were removed from the analysis, and differential gene expression analysis was performed with the data sets for the other two biological replicates of Drna-CNPase and the three biological replicates for Drna-CNP-inact (deltai1 to deltai3).
Growth curves. Growth curves were performed for E. coli and S. Typhimurium WT and Drna strains with and without the pKT-CNP and pKT-CNP-inact plasmids. For E. coli and S. Typhimurium strains, single colonies were inoculated into 3 ml M9 medium (0.4% glucose and 0.2% Casamino Acids) with appropriate antibiotics and were grown at 37°C overnight with shaking. E. coli cultures were diluted 1:100 in 2 ml fresh M9 medium in 24-well culture plates for growth at 37°C with automated OD 600 readings at 15-min intervals in a Biotek Epoch 2 microplate reader. S. Typhimurium cultures were diluted in 25 ml fresh M9 medium to an OD 600 of ;0.05 in 250-ml flasks and incubated with aeration at 37°C; samples were taken at 30-to 60-min intervals for OD 600 readings on a spectrophotometer. At an OD 600 of ;0.1 to 0.2, cultures were induced with 25 ng/ml anhydrotetracycline. Cultures were then grown to the late stationary phase. At least 3 biological replicates were performed for each strain. Doubling time (T d ) for each strain was calculated in Excel (Microsoft) using the slope (a) from the linear region of the exponential phase in a semilogarithmic plot of OD 600 versus time, as follows: T d = ln(2)/a.
Motility assay. Bacterial swimming motility was assayed as previously published, with minor modifications (33). Autoclaved motility medium (1% tryptone, 0.5% NaCl, and 0.3% agar) was cooled in a 60°C water bath. The medium was supplemented with appropriate antibiotics, 25 ng/ml anhydrotetracycline to induce pKT-CNP/pKT-CNP-inact, and/or 1 mM cell-permeable 29,39-cNMP derivatives as required for each experiment. The medium was aliquoted into 60-mm Â 15-mm petri dishes (10 ml of medium per dish) or 100-mm Â 15-mm petri dishes (25 ml/dish). After allowing the plates to solidify overnight, 2.5 ml of bacterial culture (diluted 1:100 from an overnight culture in LB) were inoculated into the surface of each plate. For E. coli, the plates were incubated for 24 h at 30°C; S. Typhimurium plates were incubated at 30°C and results recorded at 10 h and 24 h. The area of the motility zone was quantified by measuring the zone across three diameters.
Quantitative Western blot analysis of flagellar protein expression. Cultures of WT and Drna strains (1.5 ml) were grown in 15-ml plastic culture tubes in the presence of Bt-cAMP (1 mM), Bt-cUMP (1 mM), or vehicle (water). Upon reaching an OD 600 of ;0.6, 1 ml of culture was harvested by centrifugation at 10,000 Â g for 5 min at room temperature (RT). The pellet was flash frozen in liquid nitrogen and stored at 280°C. For Western blot analysis, the insoluble protein fraction was isolated by lysing cells into 100 ml of BugBuster (EMD-Millipore), according to the manufacturer's protocol. Following resuspension of the protein pellet in 100 ml of sodium phosphate buffer (50 mM [pH 7.4]), the protein concentration was normalized across all samples using a Bradford assay (protein assay dye; Bio-Rad). Protein samples were diluted 1:1 with 2Â Laemmli buffer, denatured by heating at 95°C for 10 min, and separated by SDS-PAGE on Criterion TGX precast midi protein gels (4 to 20% acrylamide; Bio-Rad) at 4°C. The Trans-Blot Turbo transfer system (Bio-Rad) was used to transfer separated proteins to nitrocellulose membranes (0.2 mm; Bio-Rad) using the "mixed MW" mode, according to the manufacturer's instructions. The blots were processed essentially as described in the Opti-4CN substrate kit (Bio-Rad). Briefly, blots were blocked by incubation in 3% Blocker solution (Bio-Rad) for 2 h at room temperature, followed by 12 h of incubation at 4°C with rabbit polyclonal anti-flagellin primary antibody (15,000-fold dilution, catalog no. 93713; abcam). The blots then were incubated for 1 h at room temperature with goat anti-rabbit secondary antibody conjugated to horseradish peroxidase (12,000-fold dilution, catalog no. 205718; abcam), and bands were detected by incubation with the Opti-4CN substrate for 15 min at room temperature. Blots were imaged using an Epson Perfection V600 photo scanner operating with the "professional" setting to obtain 16-bit gray-scale TIF image files, and band densitometry was quantified using ImageJ (National Institutes of Health). The identity of the FliC band was confirmed by Western blot analyses using strain BW25113 fliC::kan r (DfliC) (obtained from the Keio collection) (71) as a negative control and strain K-12 W3110 overexpressing FliC from plasmid pCA24N-fliC (obtained from the ASKA collection) (81) as a positive control (26).
Congo red biofilm assay. The Congo red assay utilized for EPS/biofilm quantification is based on a published protocol (35).
(i) E. coli. LB (3 ml) was inoculated with a single colony of BW25113 (WT) or Drna cells from LB-agar plates and cultured overnight in 15-ml plastic culture tubes. Each overnight culture was used to inoculate (1:50) 10 ml YESCA (1% Casamino Acids and 0.12% yeast extract) containing 0.0025% Congo red in 50 ml Celltreat conical tubes (sterile polypropylene) with the lids left loose for gas exchange. After reaching an OD 600 of ;0.3 to 0.4, 1 ml of each culture was transferred to a 1.6-ml Eppendorf tube and either treated with vehicle (YESCA medium) or with 500 mM Bt-cAMP. The cultures were incubated for 24 h at room temperature without shaking (lids were left open and the tubes were loosely covered in plastic wrap and foil). For each culture, 200 ml of supernatant was transferred to a 96-well microplate (Corning Costar) following centrifugation at 12,000 Â g for 15 min. Biofilm formation was quantified by recording the absorbance at 500 nm (A 500 ). To normalize for cell density, each culture was disturbed by pipetting, and 200 ml was transferred to a 96-well microplate prior to recording the OD 600 using a microplate reader. The A 500 for the supernatant was subtracted from the A 500 of the Congo red medium that was used to grow the cultures and then normalized to OD 600 for the resuspended pellets.
(ii) S. Typhimurium. The Congo red assay was slightly modified based on optimized conditions for biofilm formation in S. Typhimurium (59). A 5-ml aliquot of LB was inoculated with single colonies of 14028s WT, WT (pKT-CNP), WT (pKT-CNP-inact), Drna, DyjbE::kan, DwcaI::kan, or DcsgD::kan strains in 15ml polypropylene conical tubes and grown overnight at 30°C with shaking. Each overnight culture was normalized to an OD 600 of 0.8 and subcultured (1:100) into 5 ml dilute tryptic soy broth (TSB) medium (1:20 TSB in deionized water [diH 2 O]) containing 0.0025% Congo red and 25 ng/ml anhydrotetracycline. From each of the inoculated cultures, 1-ml aliquots were transferred into three 1.6-ml microcentrifuge tubes for technical replicates. Tubes were incubated with lids left open (lightly covered in plastic wrap) for 24 h at 30°C with slow shaking (24 rpm). The amount of Congo red that was incorporated into EPS/ biofilm was calculated as described above.
Acid sensitivity assay. Acid resistance was assayed according to published methodology (36). Upon reaching an OD 600 of 0.2, cultures of WT, WT pKT-CNP, and WT pKT-CNP-inact (3 ml) were treated with anhydrotetracycline (25 ng/ml) to induce protein expression. Incubation was continued to an OD 600 of ;0.6, and the cultures then were inoculated 1:20 into 2 ml of fresh M9 medium (0.4% glucose and 0.2% Casamino Acids) at either pH 2.5 (pH adjusted using HCl) or pH 7 (control) in a 24-well microtiter plate (sterile untreated polystyrene; VWR International). The cultures were incubated for 2 h, and volume-normalized CFU (CFU per milliliter) were quantified by 6 by 6 drop plating (82) to determine the survival rate of each strain at pH 2.5 relative to pH 7.
Extraction and quantification of 29,39-cNMPs. For 29,39-cNMP extraction, cells were harvested by centrifugation and subjected to a previously published extraction protocol (24). Briefly, frozen cell pellets were suspended in 500 ml of ice-cold acetonitrile/methanol/water (2/2/1 [vol/vol/vol]), lysed by sonication on ice, and centrifuged at 4°C at 3,000 Â g for 10 min. The supernatant was then removed and dried to remove solvent using a vacuum centrifuge. The dry extracted material was resuspended in 250 ml of sodium phosphate buffer (50 mM [pH 7.4]) containing 0.5 mM 8-Br 39,59-cAMP as an internal standard (IS). The extracts were centrifuged at 12,000 Â g for 30 min at 4°C and transferred to an LC-MS autosampler vial.
29,39-cNMPs were quantified via an LC-MS/MS method using 8-Br 39,59-cAMP as the internal standard, as previously described (20,24). 29,39-cNMP concentrations were determined using a calibration curve for each analyte, and the resulting 29,39-cNMP levels were adjusted based on the recovery efficiency of each 29,39-cNMP, followed by normalization to cell number, CFU, or cell density (24). The concentration of IS was 0.5 mM in all samples for calibration, and the concentrations of 29,39-cNMP standards ranged from 0.02 to 20 mM. All nucleotide concentrations in stock solutions were determined via UV-visible (UV-Vis) spectrophotometry (Cary Series, Agilent Technology, Santa Clara, CA).
LC-MS/MS conditions. The LC-MS/MS methodology was performed as previously described (20). Briefly, a Thermo Electron linear trap quadropole-Fourier transform mass spectrometry (LTQ-FTMS) instrument was employed for sample analysis and chromatographic analysis was performed using a Shimadzu autosampler and a Dionex Ultimate 3000 dual gradient pump. LC-MS instrumentation was controlled by Xcalibur and DCMSlink software (Thermo Scientific). Samples were separated using a reversed-phase Leapsil C 18 column (2.7 mm Â 150 Â 2.1 mm; Dikma Technologies, Inc., Lake Forest, CA). The mobile phase consisted of water with 0.1% formic acid (A) and methanol with 0.1% formic acid (B). The flow rate was 0.3 ml/min, and the following chromatography program was employed: 0% B from 0 to 4 min, then a gradient from 0 to 1.5% B from 4 to 15 min, followed by a gradient from 1.5% to 8% B over 15 to 20 min, followed by holding at 8% B from 20 to 25 min, then a gradient from 8% to 15% B from 25 to 28 min, followed by holding at 15% B from 28 to 35 min, and finally a gradient back to 0% B from 35 to 35.1 min prior to the column being reequilibrated by holding at 0% B from 35.1 to 45 min. The column was washed after analysis of every 2 to 4 extracts using a gradient from 0% to 100% B from 0 to 2 min, followed by holding at 100% from 2 to 10 min, then a gradient from 0% to 100% C (acetonitrile) from 10 to 12 min, followed by holding at 100% C from 12 to 20 min, followed by a final gradient from 0% to 100% A over 20 to 25 min and reequilibration at 100% A from 25 to 40 min. Electrospray ionization was performed in positive ion mode in the LTQ-FTMS instrument using a capillary voltage of 35 V, a 5-kV needle voltage, a capillary temperature of 275°C, and a 110-V tube lens voltage. Samples were detected in the ion trap using a 1amu isolation window and a normalized collision energy of 35 eV. An activation Q of 0.250 was used, with an activation time of 30 ms. Nucleotides were detected based on the protonated parent ions and quantified using the protonated nucleobase fragment ions; peaks were integrated using Xcalibur software (Thermo Fisher).
Growth conditions for quantification of 29,39-cNMPs in S. Typhimurium. To quantify 29,39-cNMPs at different phases of growth, single colonies of the 14028s WT were inoculated into 5 ml M9-glucose medium plus 0.2% Casamino Acids and grown at 37°C overnight with shaking. Cells were subcultured to an OD 600 of 0.05 in 50 ml fresh M9-glucose plus 0.2% Casamino Acids in 250-ml culture flasks. Cultures were grown at 37°C with aeration; 2-ml samples were collected at the mid-exponential phase (OD 600 = 0.5), the late exponential/early stationary phase (OD 600 = 0.9), and the late stationary phase (24 h after subculture). Cells were pelleted by centrifugation at 8,000 Â g for 3 min. After decanting the supernatant, pellets were flash frozen in a dry ice and ethanol bath and stored at 280°C. An additional 1 ml of sample was collected, washed in 1Â phosphate-buffered saline (PBS), and serially diluted. Dilutions were plated in triplicate on LB plates. Plates were incubated at 37°C overnight and were used to enumerate the CFU per milliliter present in each sample.
For 29,39-cNMP quantification during expression of CNPase or CNPase-inact, samples from the RNAseq preparations (described below) were serially diluted, and CFU per milliliter were enumerated by plating dilutions on LB agar plates. Aliquots (5 ml) of the remaining samples were pelleted by centrifugation in a swinging bucket centrifuge at 3500 Â g for 20 min. Pellets were resuspended in 1 ml PBS, transferred to a microcentrifuge tube, and pelleted by centrifugation at 8,000 Â g for 3 min. Pellets were flash frozen and stored at 280°C prior to extraction and quantification of 29,39-cNMPs.
For 29,39-cNMP quantification upon treatment with mitomycin C (MMC), single colonies of Drna 14028s or WT 14028s, with and without pBAD33-CNP/pBAD33-CNP-inact, were inoculated into 5 ml LB medium and grown at 37°C overnight with shaking. Cells were subcultured (1:100) in 25 ml fresh LB in 250-ml culture flasks. Cultures were grown at 37°C with shaking to an OD 600 of 0.1, at which point plasmids were induced with the addition of 0.02% L-arabinose. Cultures were then grown to an OD 600 of 0.3 to 0.4 Each culture was split into two 10-ml cultures in 50-ml polypropylene conical tubes; one culture remained an untreated control, and the other was treated with 3 mM MMC (Sigma-Aldrich, St. Louis, MO). Tubes were returned the incubator for 90 min of treatment.
XylE reporter assay for RNA repair operon expression. To measure activation of expression from the S. Typhimurium RNA repair operon promoter, XylE reporter strains were utilized as previously described (45). JEK17 (14028s Drsr::xylE) was freshly transformed with pKT-CNP or pKT-CNP-inact prior to each assay. Transformants were recovered in 1 ml SOC medium at 30°C for 1 h; the entirety of each transformation was then subcultured in 10 ml LB plus carbenicillin and grown at 30°C overnight, alongside JEK17 and JEK29 (14028s Drna Drsr::xylE). Overnight growth was subcultured (1:50) into 25 ml LB (supplemented with carbenicillin as required to maintain plasmids) in a 250-ml culture flask. Cultures were grown at 37°C with shaking to an OD 600 of 0.1, at which point 25 ng/ml anhydrotetracycline was added. Cultures were returned to the incubator until they reached the mid-log phase (OD 600 = 0.3 to 0.4). Each culture was split into two 10-ml cultures in 50-ml polypropylene conical tubes; one of each pair was an untreated control, and the other was treated with 3 mM MMC. Cultures were returned to the incubator for 90 min.
At the conclusion of the treatment time, 2 ml of culture was harvested for XylE activity assays. Assays were performed as described by Kurasz et al. (45). Briefly, cells were washed and resuspended in 50 mM phosphate buffer (pH 7.4). Samples were diluted in phosphate buffer to an OD 600 of 1.0. The XylE reaction was initiated by adding 100 ml of sample to 900 ml of catechol solution (10 mM catechol [Sigma-Aldrich] and 50 mM phosphate buffer), and the conversion of catechol to 2-hydroxymuconic semialdehyde was monitored continuously at room temperature with readings taken every 10 s for 1.5 min using a Thermo Scientific Genesys 20 spectrophotometer at a wavelength (l) of 375 nm. Each individual culture was assayed for 2 or 3 technical replicates. XylE activity was calculated as nanomoles per minute per 10 8 cells (as previously determined by plating assays) during linear activity, based on the Beer-Lambert law and using an extinction coefficient of 44,000 L Á mol 21 Á cm 21 for 2-hydroxymuconic semialdehyde. The simplified equation is as follows: XylE activity = [1,000 Â (DOD 375 /min)]/(44 Â number of cells Â 10 8 )].
To assay RNA repair operon expression in the presence of the cell-permeable 29,39-cAMP analog, Bt-cAMP, JEK12 (14028s Drsr::xylE-kan) was utilized in XylE assays as described above with the following modifications. Cultures were diluted from overnight growth to an OD 600 of 0.05 in 50 ml LB in 500-ml

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.