Transcriptional Effects of Melatonin on the Gut Commensal Bacterium Klebsiella aerogenes

Abstract

Klebsiella (nee Enterobacter) aerogenes is the first human gut commensal bacterium with a documented sensitivity to the pineal/gastrointestinal hormone melatonin. Exogenous melatonin specifically increases the size of macrocolonies on semisolid agar and synchronizes the circadian clock of K. aerogenes in a concentration dependent manner. However, the mechanisms driving these phenomena are unknown. In this study, we applied RNA sequencing to identify melatonin sensitive transcripts during culture maturation. This work demonstrates that the majority of melatonin sensitive genes are growth stage specific. Melatonin exposure induced differential gene expression of 81 transcripts during exponential growth and 30 during early stationary phase. This indole molecule affects genes related to biofilm formation, fimbria biogenesis, transcriptional regulators, carbohydrate transport and metabolism, phosphotransferase systems (PTS), stress response, metal ion binding and transport. Differential expression of biofilm and fimbria-related genes may be responsible for the observed differences in macrocolony area. These data suggest that melatonin enhances Klebsiella aerogenes host colonization.

1. Introduction

The gut commensal Klebsiella (née Enterobacter) aerogenes is a Gram-negative, flagellated bacterium that can colonize the human gastrointestinal tract at up to 10⁷ cfu/g (1). K. aerogenes is capable of causing community- or hospital-acquired bloodstream infections (BSI) and is associated with more severe clinical outcomes compared to BSIs caused by other Enterobacter species, including acute kidney injury in 43% of cases, septic shock in 30% of infected patients, and in hospital mortality of 28% of infected patients (2). Additionally, K. aerogenes is also a cause of secondary bacterial infections among patients who have contracted the SARS-CoV-2 coronavirus (COVID-19) (3–6).

Melatonin, a neurohormone primarily secreted by the pineal gland (7–9), is present at high concentrations in the gastrointestinal lumen (10). However, melatonin biosynthetic enzymes are also present in enterochromaffin cells (ECs) and the intestinal mucosa (11, 12). ECs within human and mouse gut tissue (13, 14) secrete melatonin into the lumen at concentrations exceeding serum levels (15). Melatonin is synthesized from the dietary essential amino acid tryptophan, and its concentration is increased after food consumption (16). Additionally, melatonin has an impact on circadian rhythmicity of the gut microbiota as a whole (17).

Conversely, several bacteria have been reported to modulate intestinal melatonin biosynthesis. For example, colonization of the gastric mucosa by H. pylori decreases the expression of enzymes involved in melatonin biosynthesis in the gastrointestinal track (18). In addition, endogenous spore forming gut bacteria have been shown to produce metabolites that stimulate secretion of the melatonin precursor serotonin by ECs (19). Furthermore, previous work by others has shown that the metabolome secreted by cultured human gut bacteria has the capacity to activate host melatonin receptors MTR1A and MTR1B (20). This evidence supports the hypothesis that gut melatonin plays an important role in a complicated relationship between commensal microbiota and the host. However, whether melatonin induces specific transcriptional programs in members of the gut microbiota has not been characterized to date.

K. aerogenes has a functioning circadian clock, which is synchronized by melatonin. As part of a metagenomic screen for gut microbiota that exhibit genomic signals that may confer sensitivity to the pineal and gut hormone melatonin, we discovered that K. aerogenes harbors sequences that are similar to the human melatonin receptor (7). Thus, we hypothesized that melatonin may serve as an exocrine signal for gastrointestinal microbiota (9). Further investigation revealed that the area of macrocolonies grown on semi-solid agar was greater when bacteria were cultured in the presence of 1 nM melatonin compared to colonies grown without melatonin (7). This effect was dose-dependent and specific to melatonin, as tryptophan, serotonin, and N-acetylserotonin had no effect at concentrations ranging from 1 pM to 1 nM. Melatonin also synchronized a circadian rhythm of gene expression in this bacterium, which was 1) temperature compensated and 2) capable of entraining to cycles of ambient temperature (T_A) similar to daily changes in human body temperature (8).

In order to identify melatonin-sensitive sequences in the K. aerogenes genome, we conducted an RNA sequencing screen on planktonic, aerated cultures maintained in the presence or absence of 1 nM melatonin and under conditions of exponential growth vs stationary phase. RNA sequencing (RNA-seq) is an unbiased high-throughput sequencing technique utilized to characterize the global transcriptional response of organisms, including bacteria, during varying conditions, such as different growth phases (21, 22). We chose 1 nM melatonin for this analysis based on a previous dose-response study in which maximal effects of melatonin (V_max) on K. aerogenes motility were observed at concentrations of 100 pM – 1 nM (7). In addition, 1 nM melatonin represents 232.3 pg/ml, which is approximately 4X peak serum melatonin titres in humans (27) and corresponds to physiologically-relevant concentrations found in the gut (27–30). The aims of this experiment were twofold: first, to characterize the transcriptome of K. aerogenes and to establish a differential gene expression profile as the culture matures, and second, to characterize transcriptomic changes induced by melatonin during the exponential and/or early stationary growth phases.

2. Methods

2.1. Effect of melatonin on K. aerogenes transcriptome evaluated using RNA High Throughput Sequencing

2.1.1. Media and culture method

The K. aerogenes strain (PmotA::luxCDABE) and melatonin (Sigma Aldrich, St. Louis, MO) were the same as in our previous study (7). Tryptic soy broth (TSB) (200 mL) was inoculated with an overnight culture of K. aerogenes in 1:100 ratio. Triplicate flasks were prepared with either 0nM or 1 nM of melatonin. All 6 flasks were incubated in a shaking incubator at 150 rpm and 37°C until the desired optical density (OD) was reached. Exponential growth samples were collected at OD₆₀₀=0.3 and stationary samples were collected at OD₆₀₀=3.0. Aliquots (3 mL) were collected, bacteria were mixed with a transcription stop solution (5% acidic phenol in ethanol (23)), centrifuged at 3000 rpm and 4°C, bacterial pellets were frozen in liquid nitrogen, and stored at −80°C until the extraction.

2.1.2. RNA extraction

RNA was isolated using the hot phenol method. Nucleic acid concentrations were measured using a Qubit^® RNA Assay Kit in a Qubit^® 2.0 Fluorometer (Life Technologies, CA, USA). RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Total RNA samples were submitted to Beijing Novogene Bioinformatics Technology Co., Ltd. (Beijing, China). A total of 3 μg RNA per sample was used as input material for the library preparation and sequencing. Ribosomal RNA depletion was achieved using a Ribo-Zero^™ Magnetic Kit (Illumina, Inc.).

2.1.3. Library preparation

A library for strand-specific transcriptome sequencing was made with NEBNext^® Ultra^™ Directional RNA Library Prep Kit for Illumina^® (NEB, USA) following the manufacturer’s recommendations, and index codes were added to each sample. Briefly, fragmentation was carried out using NEBNext First Strand Synthesis Reaction Buffer (5X). After cDNA synthesis, fragments of approximately 150~200 bp in length were purified with the AMPure XP system (Beckman Coulter, Beverly, USA). The PCR reaction was conducted using Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. Finally, products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system. RNA sequencing was performed at Novogene Co., Ltd. (Beijing, China) on an Illumina Hiseq platform.

The reference genome and gene annotation files of K. aerogenes KCTC 2190 (24) were obtained from the NCBI database. Clean reads were aligned to the reference genome using Bowtie2-2.2.3. (25). The number of reads that mapped to each gene was evaluated by HTSeq v0.6.1. Then the expected number of Fragments Per Kilobase of transcript sequence per Millions (FPKM) base pairs sequenced, of each transcript was calculated based on the length of the gene and depth of sequencing. FPKM considers the effect of sequencing depth and gene length for the reads count at the same time, and it is used for estimating gene expression levels (26).

2.1.4. Differential expression analysis

To quantify the gene expression levels, the FPKM of each gene was calculated. In general, an FPKM value of 1 was set as the threshold for determining whether the gene is expressed or not. K. aerogenes KCTC 2190 genome that we used as reference (24) contains 5,007 annotated genes (27). Principal component analysis (PCA) of the normalized expression patterns of each sample showed that there were culture conditions with specific trends among the specimens, with biological replicates isolated from the same growth conditions clustered together (Fig. 1A). The DESeq R package (1.18.0) (28) was implemented to identify the differentially expressed genes (DEGs). The obtained p-values were adjusted using the Benjamini and Hochberg’s FDR estimation method, and transcripts with an adjusted P-value <0.05 calculated by DESeq were identified as differentially expressed.

Figure 1.

Differential gene expression (DEG) of K. aerogenes cultured under four different conditions A. Principal Component Analysis reveals growth condition-specific patterns of gene expression. B. Cluster analysis of differentially expressed genes in four experimental groups. C. Venn diagram illustrating the number of shared and unique genes identified among the four growth conditions, FPKM>1 represents the expression threshold.

2.1.5. qPCR analysis

RNA was subjected to rigorous DNase treatment (Turbo DNase I, Life Technologies, USA), and 1 μg DNA-free RNA was reverse transcribed using iScript^™ reverse transcriptase (Bio-Rad, USA). cDNA of 13 genes (Primer sequences Table S5) was amplified by an Applied Biosystems Viia7 Real-Time PCR platform using TaqMan probe chemistry (IDT, USA). All qPCR reactions were performed in triplicate using 10-fold diluted cDNA. Relative fold changes in transcript abundance were calculated using the ΔΔC_T method, with target transcript levels normalized to multiplexed recA endogenous control probes in each sample from a total of three biological replicates per growth and treatment condition. ligA was selected as a melatonin non-responsive gene by R package GeNorm (29) using the expression patterns of sequenced samples.

2.1.6. sRNA analysis

Novel intergenic transcripts, discovered by Rockhopper (30), were aligned with sequences in NCBI NR database using Blastx. Novel transcripts without NR annotation and the length between 50 and 500 nt were designated sRNA candidates. RNAfold (31) was applied to predict the secondary structures of sRNAs.

2.2. Attachment to abiotic surfaces

To test the surface attachment of K. aerogenes we used modified methods from Myers-Morales and colleagues (32). Briefly, an overnight culture of K. aerogenes was diluted in 1:100 ratio with TSB containing 0 nM or 1 nM melatonin. The cell suspension (100 μl) was inoculated into wells of a UV-sterilized clear, flat bottom 96-well plate (Greiner Bio-One, Monroe, NC) and incubated statically for 24 hours at 37°C. Unbound cells were removed by rinsing with water, followed by inversion of the plate and tapping on absorbent paper. Adherent cells were subsequently stained using 0.1% crystal violet. The stain was removed by careful washing with water, and the wells were allowed to dry completely. Remaining stain was eluted using 33% acetic acid. Quantification was performed using a plate reader (iMark^™ Microplate Absorbance Reader, Bio-Rad, Hercules, CA, USA) by recording absorbance at 600 nm wavelength. Normality of data distribution was tested using JMP followed by one way ANOVA test (SAS Institute, version 14.0.0).

3. Results and Discussion

3.1. RNA-Sequencing of cultured Klebsiella aerogenes

The RNA sequencing experiment was performed to characterize the global transcriptome of cultured K. aerogenes from triplicate planktonic cultures of the wild type strain during the early exponential and early stationary phase. Additionally, samples from triplicate cultures treated with 1 nM melatonin were also analyzed, giving a total of four experimental conditions (Fig. 1B).

3.2. Transcriptional differences during culture maturation

Comparative analysis of exponential and stationary cells revealed clear transcriptional differences between the two growth stages (Fig. 1B). We detected expression of 4890 various transcripts, with 4395 (89.9%) mRNAs constitutively expressed under all experimental conditions (Fig. 1C). Additionally, 39 small RNAs (sRNAs) were identified, which to the best of our knowledge, is the first report of sRNA expression in this bacterium. Moreover, 21 sRNAs were significantly differentially expressed in either exponential or stationary growth (Table 9).

Table 9.

sRNA expression pattern

Gene_id	readcount_expo_con	readcount_stat_con	log2FoldChange	pval	padj
sRNA00001	1661.38506	28.6096106	5.8597	1.02E-12	2.89E-12
sRNA00009	499.442458	71.2882052	2.8086	8.58E-60	9.26E-59
sRNA00013	659.34819	95.8834936	2.7817	9.90E-26	4.80E-25
sRNA00026	4.00014301	42.5159031	−3.4099	1.38E-11	3.73E-11
sRNA00027	356.571513	84.678909	2.0741	0.0011199	0.0018667
sRNA00035	620.289457	189.709148	1.7092	1.56E-14	4.79E-14
sRNA00042	181.075443	634.487767	−1.809	1.78E-10	4.57E-10
sRNA00043	14.9054966	1438.24867	−6.5923	2.79E-257	2.79E-255
sRNA00053	31.9621992	4.77211771	2.7437	1.90E-06	3.92E-06
sRNA00061	658.265716	255.432909	1.3657	0.011227	0.01665
sRNA00066	3803.19258	1102.70974	1.7862	0.0001737	0.00031008
sRNA00076	9.07422177	165.897356	−4.1924	1.58E-11	4.26E-11
sRNA00084	8.82746273	828.212619	−6.5519	1.48E-56	1.51E-55
sRNA00111	864.965138	291.309708	1.5701	5.00E-07	1.07E-06
sRNA00116	158.407657	39.0178968	2.0214	2.89E-10	7.36E-10
sRNA00126	44.5816367	5.93101536	2.9101	0.0025687	0.0041066
sRNA00129	12.9948782	1261.6247	−6.6012	4.10E-242	3.59E-240
sRNA00139	604.077775	3188.35381	−2.4	1.41E-82	2.24E-81
sRNA00143	12.7835526	150.111322	−3.5537	4.22E-36	2.79E-35
sRNA00145	401.023505	597.039793	−0.57414	0.004809	0.0074597
sRNA00146	277.30279	555.075034	−1.0012	6.40E-12	1.76E-11

During the exponential phase, bacteria exhibit a constant, rapid growth rate and uniform metabolic activity. As the number of microorganisms increases, and the nutrient availability is limiting, bacteria enter the stationary phase, where division rate and metabolism slows (33). Most microbes have developed complex mechanisms for surviving in nutrient-limited conditions; cells can enter and persist in this long-term, limited growth stage until eventual death after completely exhausting the available nutrients. Subsequently, this non-exponential growth state also leads to diverse changes in bacterial physiology, metabolism and energetics, ranging from carbon storage and cell wall modifications to macromolecule synthesis and stability (34).

We found that 6.7% of genes were differentially expressed as cultures mature. 4395 transcripts were detected in all samples (Fig. 1C). Comparison of control samples from the two growth stages using Venn diagram, with the FPKM >1 as the expression threshold, showed 126 transcripts were unique to the exponential growth, and 198 transcripts were exclusively expressed in stationary growth (Fig. 2B). When DEGs from the same samples were plotted, 1827 genes were upregulated and 1828 genes were downregulated (Fig. 2A, Table S1).

Figure 2.

Comparison of differential gene expression (DEG) in K. aerogenes during exponential and stationary growth. A. Volcano plot depicting up- and downregulated genes. Each dot in the volcano plot represents one gene. B. Venn diagram depicting the number of genes expressed under two growth stages. FPKM>1 represents the expression threshold. C. Scatterplot depicting 20 the most significantly upregulated KEGG pathways in exponential cells as compared to stationary cells. The x-axis represents the pathway enrichment P value, and the y-axis indicates the KEGG pathway. D. Downregulated KEGG pathways enriched in exponential cells as compared to stationary cells.

To further characterize these differences, we examined the functions of DEGs using the Kyoto Encyclopedia of Genes and Genomics (KEGG) pathway enrichment analysis. To investigate changes occurring as a culture matures, the percentage of genes in each KEGG category was analyzed. Twenty of the most significantly enriched pathways (q < 0.05) among the differentially expressed genes are illustrated in Fig. 2C (upregulated pathways) and Fig. 2D (downregulated pathways). Analysis of these pathways revealed differences in many cellular processes, including metabolic pathways, flagellum assembly, protein export, ribosomal pathways, amino acid degradation, and biosynthesis of secondary metabolites.

3.3. Melatonin-induced transcriptional changes

Our RNA-seq experiment revealed that melatonin affected different genes depending on the growth stage: 81 genes were differentially expressed in exponential growth, from which 34 were upregulated and 47 were downregulated (Fig. 3A, Table S2). During stationary growth, 30 genes were differentially expressed (Fig. 3B, Table S3); we observed 11 upregulated and 19 downregulated mRNAs. To confirm RNA-seq results, we performed qPCR analysis of six differentially expressed transcripts for each growth condition, as well as one transcript that was produced at similar levels during exponential and static growth (ligA), and was non-responsive to melatonin (Fig. 3B and 3D). The qPCR results corroborated the RNA-Seq analyses. The presence of melatonin in the media affected genes within many KEGG pathways, but no pathways were significantly enriched as a whole. Nevertheless, specific genes within these pathways were affected with statistical significance (Fig. S1).

Figure 3.

Comparison of melatonin-induced differential gene expression (DEG) in K. aerogenes. A. Volcano plot depicting numbers of upregulated and downregulated genes produced during exponential growth by cells cultured with 1 nM or 0 nM melatonin. Each dot in the volcano plot represents one gene. Dashed line defines the adjusted P value threshold for significance. Genes which do not show significant differential expression are plotted in black. B. qPCR analysis of selected melatonin-responsive genes that are upregulated (pmfE, fimC, and lamB) or downregulated (raiA, bssR, and ykgJ) in exponential culture. C. Volcano plot depicting numbers of upregulated and downregulated genes produced during stationary growth by cells cultured with 1 nM or 0nM melatonin. D. qPCR analysis of selected melatonin-responsive genes that are upregulated (fimA, pmfA, and pgbA) or downregulated (hypB, malM, and tnpA) in stationary culture. In B and D, each dot represents one biological replicate, and error bars represent the standard error of the mean.

Two transcripts encoding periplasmic proteins – the maltose regulon periplasmic protein MalM and an unannotated transcript that encodes a putative periplasmic protein (Table 1) – were affected by melatonin in both exponential and stationary growth phases. The CDD database suggested that the unannotated putative protein belongs to the aldolase family, resembling the keto-3-deoxygluconate 6-phosphate (KDGP) aldolase that catalyzes the conversion of KDGP to produce pyruvate and glyceraldehyde-3-phosphate (35). In E.coli, KDGP aldolase is encoded by the eda gene (Entner-Doudoroff aldolase) which is under control of at least 3 transcriptional regulators GntR, KdgR, and PhoB (36). Subsequently, we observed differential expression of GntR gene in our dataset (see below, section 3.3.1.). Additionally, eda is necessary for E. coli colonization of the large intestine of streptomycin-treated mice (37). Interestingly, during exponential growth, both transcripts were upregulated whereas in the stationary phase both transcripts were downregulated. This could be due to differential binding sites for various transcriptional regulators in the promoter sequences, or regulation by numerous sRNAs.

Table 1.

Melatonin-sensitive DEGs affected in both growth stages

growth phase	Gene_id	log2Fold Change	pval	Product
exponential	EAE_08405	1.73	0.00010006	maltose regulon periplasmic protein
	EAE_09645	0.85	0.00032077	KDGP aldolase
stationary	EAE_08405	−0.99	7.61E-05	maltose regulon periplasmic protein
	EAE_09645	−0.59	1.62E-05	KDGP aldolase

3.3.1. Transcriptional regulators

Within the 81 DEGs from exponential cells, we identified five transcriptional regulators that were downregulated in response to melatonin: TetR family transcriptional regulator, Helix-Turn-Helix (HTH) DNA binding domain from the MerR superfamily, DNA-binding transcriptional regulator Crl, phosphonate utilization transcriptional regulator PhnR, and GntR family regulatory protein (Table 2). The transcription factor family GntR, first described in B. subtilis and named after the gluconate-operon repressor, is widely distributed among bacteria and regulates various biological processes (38). These regulatory proteins are comprised of a DNA-binding domain and a signaling domain. All representatives of this family share highly similar N-terminal HTH (helix-turn-helix) DNA-binding domains, but the C-terminal domains are often dissimilar and bind specific coactivator molecules that change their conformation and allow DNA binding. In many bacteria, GntR transcription factors are involved in regulation of carbohydrate transport and metabolism, and in at least in S. mutans, biofilm formation (39).

Table 2.

Relative expression of transcriptional regulators in K. aerogenes grown with 1 nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_03430	GntR	GntR family regulatory protein	exponential	−1.677
EAE_17995	MerR	MerR superfamily transcription regulator	exponential	−1.729
EAE_10115	TetR	TetR family transcriptional regulator	exponential	−1.730
EAE_11960	Crl	DNA-binding transcriptional regulator Crl	exponential	−2.055
EAE_07360	PhnR	phosphonate utilization transcriptional regulator PhnR	exponential	−2.556

MerR superfamily regulators also possess a HTH DNA-binding domain and regulate expression of metal ion transporters and oxidative stress regulons (40). Their mechanism of action is based on reconfiguration of spacer sequences located in the promoter region between the −35 and −10 promoter element. This change in MerR transcriptional regulator expression could be the reason for lower abundance in 7 genes related to the stress response (Table 7). Melatonin also affected many transcripts encoding metal ion transporters and proteins that bind metal ions (Table 8), which are discussed in detail in a separate section below (section 3.3.7.).

Table 7.

Relative expression of stress response genes in K. aerogenes grown with 1 nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_05770	UspB	Universal stress protein B	exponential	−1.98
EAE_05775	UspA	Universal stress protein A	exponential	−1.57
EAE_16700	YchH	stress response protein	exponential	−2.90
EAE_18855	lipoprotein	response to oxidative stress	exponential	−2.07
EAE_01045	RaiA	translation inhibitor protein	exponential	−4.33
Novel00008	RaiA	translation inhibitor protein	exponential	−5.64
EAE_10730	DnaK	molecular chaperone	stationary	−1.4
EAE_09130	GroES	co-chaperonin	stationary	−1.7
Novel00128/EAE_20600	UspF	universal stress protein F	stationary	1.39

Table 8.

Relative expression of ion transporters and ion binding proteins in K. aerogenes grown with 1 nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE 16280	BssS	biofilm formation regulatory protein	exponential	−1.675
EAE 22905	CutC	Copper homeostasis protein	exponential	−1.684
EAE 02500	RcnA	nickel/cobalt efflux transporter	exponential	2.172
EAE_09975	CusB	copper/silver efflux system membrane fusion	exponential	2.248
EAE 24250	YfaE	2Fe-2S ferredoxin YfaE	exponential	1.656
EAE 21475	YkgJ	ferredoxin	exponential	−3.131
EAE 16085	EfeB	iron-dependent peroxidase	exponential	−1.804
EAE 06085	Uup	divalent iron ion ABC transporter	stationary	−1.389
EAE_01610	HypF	zinc ion binding carbamoyltransferase	stationary	−1.504
EAE_01685	HypB	Hydrogenase isoenzymes nickel incorporation protein	stationary	−2.298
EAE_09640	DgaE	selenocysteine synthase	stationary	−1.778
EAE 01615	HydN	formate dehydrogenase-H ferredoxin subunit	stationary	−1.704
EAE 19675	AppB	cytochrome bd-II ubiquinol oxidase	stationary	−1.564
EAE 14725	YbiY	pyruvate formate-lyase 3-activating enzyme	stationary	−1.640

The sigma factor-binding protein Crl interacts with the sigma S (RpoS, σ^S), a stress-response sigma factor, which is activated when bacteria face unfavorable conditions, such as low temperatures, or during the stationary phase (41). RpoS-dependent gene activation results in general stress resistance (42), and Crl binding to the σ^S promotes expression of stress response genes and those required for production of curli fibers involved in biofilm formation (41). The reduced expression of Clr could be responsible for the decreased levels of stress response gene and biofilm-related gene transcripts in the presence of melatonin.

Members of the TetR family are involved in control of multidrug efflux pumps, biosynthesis of antibiotics, response to osmotic stress and toxic chemicals, control of catabolic pathways, differentiation processes, and pathogenicity (43). TetR also controls the expression of tetracycline resistance efflux pump tetA, which removes antibiotic out of the cell before it can attach to the ribosomes and inhibit protein synthesis. The last transcript in this category, phosphonate utilization transcriptional regulator PhnR, is usually found nearby to or inside of operons for the degradation of 2-aminoethylphosphonate (AEP) in Salmonella (44).

3.3.2. Fimbria

Fimbria (also known as pili) appendages are used by bacteria for attachment to other microbes and biotic or abiotic surfaces; thus, pili are recognized virulence factors (45). Adhesive pili-mediated interactions can promote biofilm formation and are often essential to the successful colonization of the host (46). To our knowledge, the role of K. aerogenes pili in gut colonization has not been investigated in detail. Previous reports indicated that this microorganism possesses at least 2 kinds of fimbria: type 1 fimbriae, and hemagglutinating thin non-channeled fimbriae (47, 48). In the current study, we observed melatonin-induced differential expression of 5 pilus genes (Table 3).

Table 3.

Relative expression of fimbrial genes in K. aerogenes grown with 1nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_02585	PmfE	minor fimbrial subunit	exponential	4.129
EAE_02420	FimD	type 1 pilus assembly chaperone	exponential	1.975
EAE_02425	FimC	type 1 pilus outer membrane usher protein	exponential	1.881
EAE_02410	FimA	type 1 pilus pilin	stationary	1.295
EAE_02610	PmfA	major fimbrial subunit	stationary	1.409

The most highly upregulated melatonin-responsive gene in exponential cells (fold change = 4.129) was a minor fimbrial subunit pmfE (Proteus mirabilis fimbria). PmfE is a tip adhesin, facilitating attachment to the surrounding surface. Proteomic analysis performed on P. mirabilis identified PmfE among the most abundant extracellular proteins (49). Melatonin-induced upregulation of pmfE was confirmed by qPCR (1.77 fold increase by qPCR vs. 4.13 fold increase by RNA-seq analysis, Fig. 3B). Additionally, we observed upregulation of 2 other genes involved in pilus biogenesis fimD and fimC. FimC, the chaperone-usher pathway (CUP) usher is an outer membrane protein that drives pilus assembly. qPCR analysis confirmed the observed melatonin-induced increase in fimC transcript levels (1.77 fold increase by qPCR vs. 1.98 fold increase by RNA-seq analysis, Fig. 3B). FimD, the periplasmic chaperone, prevents pilin monomers from assembling in the periplasmic space (50).

Melatonin also upregulated fimbrial genes in stationary cells. These included genes encoding the type 1 major fimbrial protein FimA and the major fimbrial subunit PmfA, which is involved in the regulation of pilus length and adhesion (51). Increased levels of fimA transcript were confirmed by qPCR (1.17 fold increase by qPCR vs. 1.29 fold increase by RNA-seq analysis, Fig. 3D) in melatonin-treated stationary cells. Stationary growth is characterized by increased fimbriation in general (49); however, our results suggest that melatonin may enhance piliation.

3.3.3. Biofilm

A biofilm is a surface-associated community of sessile bacteria that exhibits multi-cellular behavior. An estimated 65–80% of human bacterial infections involve biofilms (52), and biofilms formed on urinary catheters frequently involve K. aerogenes (53). When melatonin was present in the media during the exponential growth, we observed lower expression of 4 genes involved in biofilm formation (Table 4), likely due to the downregulation of the GntR transcription factor. Biofilm-associated genes enriched by melatonin include the two-component-system connector protein YcgZ, the regulator of acid resistance influenced by indole AriR, the biofilm formation regulatory protein BssS, and the regulator of biofilm through signal secretion BssR. We confirmed the melatonin-induced downregulation of bssR by qPCR (−1.81 fold decreased transcript levels by qPCR vs. −1.82 fold decreased levels by RNA-seq analysis, Fig. 3B). YcgZ and ariR were both downregulated with an approximately 3-fold change. AriR was reported to play a role in biofilm development, repression of cellular motility, and acid-resistance (54). BssS and BssR regulatory proteins are involved in repression of cellular motility within biofilms (55). BssR (previously known as YliH) and BssS (YceP) are involved in repression of motility in the biofilms, and are activated in E. coli biofilms (56). As reported by Domka and colleagues, deletion of bssS and bbsR in E. coli increases biofilm formation, reduces the level of intracellular indole, and increases motility. These mutants also exhibited altered expression of stress response genes and genes encoding phosphotransferase systems (PTS).

Table 4.

Relative expression of biofilm genes in K. aerogenes grown with 1nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_14355	YcgZ	two-component-system connector protein	exponential	−3.148
EAE_14345	AriR	regulator of acid resistance influenced by indole	exponential	−3.088
EAE_16280	BssS	biofilm formation regulatory protein	exponential	−1.675
EAE_14770	BssR	regulator of biofilm through signal secretion	exponential	−1.819

The lower abundance of BssS and BssR transcripts in our data could result in the observed downregulation of stress response regulated genes and PTS genes. In our experiments, K. aerogenes were grown in planktonic culture under aerated conditions under which surface adhesion and biofilm formation was not observed. Therefore, these genes may have additional undiscovered functions. However, our lab has shown that K. aerogenes macrocolony size drastically increases when melatonin is present (7).

3.3.4. Phosphotransferase system (PTS)

The phosphotransferase system is a carbohydrate uptake mechanism that depends on energy generated from phosphoenolpyruvate (PEP). PTS catalyzes the transport and phosphorylation of mono and disaccharides across the cell membrane. This system consists of three components: 1) the cytoplasmic Enzyme I, which is a PEP-dependent protein phosphorylating kinase; 2) a heat-stable phosphoryl carrier protein (HPr); and 3) a sugar specific Enzyme II complex consisting of IIA, IIB, IIC (57). Enzyme I and HPr are general energy coupling enzymes, common to all Enzyme II complexes, and catalyzing transfer of phosphoryl group from PEP. Enzyme II is substrate specific, and its responsible for sugar translocation through the membrane using IIC permease and subsequent phosphorylation of this sugar particle. After being released from EII into the cytoplasm phosphorylated sugars enter glycolysis and generate PEP and ATP. Additionally, PTSs possess regulatory functions related to carbon, nitrogen, and phosphate metabolism; chemotaxis; potassium transport; and virulence in some bacteria (58).

Examining transcriptional differences in exponential cells, we identified 2 PTS system genes that were downregulated in response to melatonin – the mannitol transporter MtlA and maltose transporter MalX (Table 5). We also observed downregulation of 2 genes encoding PTS-dependent dihydroxyacetone kinase subunits, DhaK and DhaL. Stationary cells were characterized by upregulation of the lactose/cellobiose-specific PTS family enzyme IIC component. This enzyme catalyzes the transfer of a phosphoryl group from IIB to sugar substrates associated with their translocation across the cell membrane (59). Concurrently, we observed downregulation of the PTS maltose/trehalose-specific EIIBC component and transcriptional regulator encoding component EIIA, which is critical for gene regulation.

Table 5.

Relative expression of PTS genes in K. aerogenes grown with 1 nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_06200	MtlA	mannitol transporter IIBC component	exponential	−1.835
EAE_18025	MalX	maltose transporter	exponential	−1.788
EAE_03850	DhaK	dihydroxyacetone kinase subunit	exponential	−2.633
EAE_03855	DhaL	dihydroxyacetone kinase subunit	exponential	−2.292
EAE_10515	CelB	lactose/cellobiose-specific enzyme IIC	stationary	1.330
EAE_09655	MtlA2	transcriptional regulator/PTS system, IIA component	stationary	−1.413
EAE_09695	TreB	trehalose/maltose-specific transporter subunits IIBC	stationary	−1.730

3.3.5. Carbohydrate transport and metabolism

Among the most upregulated transcripts observed during exponential growth were melatonin-regulated transcripts arising from genes involved in maltose transport. Of interest, we observed melatonin-dependent transcriptional responses in malM (maltose regulon periplasmic protein), lamB (maltoporin), and malG (maltose permease). In E. coli, malM and lamB are encoded within the malK-lamB-malM operon and are consequently part of the maltose regulon (60). qPCR analysis confirmed the observed increase in lamB transcript levels induced by melatonin in exponential cultures (2.65 fold increase by qPCR vs. 2.87 fold increase by RNA-seq analysis, Fig. 3B). A slightly smaller fold change (upregulation of 2.66-fold) was observed for the pyruvate formate lyase pflB. Interestingly, we also observed downregulation of the pyruvate formate-lyase 3-activating enzyme YbiY in the stationary cells.

Additionally, the presence of 1 nM melatonin in stationary cell culture induced the upregulation of 2 genes involved in O-glycosyl compounds hydrolyzation (bglC and bglA) that encode 6-phospho-beta-glucosidase. Simultaneously, we observed downregulation of maltose and trehalose transporters. Maltose and trehalose are disaccharides in which sugar groups are connected via an O-glycosidic bond. Glucose is the preferred energy source in many bacteria; when this simple sugar is depleted, alternative sugars are acquired from the media. Disaccharides need to be broken down by enzymes such as 6-phospho-beta-glucosidases to increase intercellular levels of glucose, which will become phosphorylated by EIIA component of the PTS. Dephosphorylated EIIA prevents adenylate cyclase from producing cAMP and stops the expression of alternative sugar transporters. The aforementioned chain of events could be responsible for lower expression of MalM maltose periplasmic protein and PTS system trehalose-specific EIIBC component in the stationary cells. Melatonin-induced downregulation of malM in stationary cultures was confirmed by qPCR (−1.74 fold decreased transcript levels by qPCR vs.−1.98 fold decreased transcript levels by RNA-seq, Fig. 3D).

3.3.6. Stress response

The ability to monitor the environment for toxic molecules and other disturbances is essential for bacterial survival. Therefore, bacteria evolved diverse signaling mechanisms to adapt gene expression to unfavorable conditions. As microbial culture matures, nutrients are depleted and metabolites accumulate. Expression of stress-related genes is crucial, especially during the transition to the stationary phase, due to limited resources and oxidative stress (61). Previous studies have demonstrated that melatonin possesses potent antioxidant properties (62).

We observed a lower abundance of 5 genes related to stress response when analyzing samples collected from exponential growth in 1 nM melatonin (Table 7). The most downregulated gene under these conditions, the translation inhibitor protein RaiA (fold change −4.33), is induced during stationary phase or cold temperatures (63) and when experiencing nitrogen limitation stress (64). Melatonin-dependent downregulation of raiA was confirmed by qPCR (−5.93 fold decreased transcript levels by qPCR, Fig. 3B). Interestingly, in our dataset we also observed an alternative transcript overlaid with raiA, which we named Novel00008. This mRNA was 380 bp and expanded both upstream and downstream from the originally annotated raiA, which measures only 336 bp.

The remaining downregulated stress response genes were two universal stress proteins, uspA and uspB (Table 7). Levels of UspA in E. coli have been shown to increase in response to a variety of stress conditions, such as carbon, nitrogen, phosphate, sulfate, and amino acid starvation, and exposure to oxidants (65). UspB is transcribed under stress conditions including starvation, and is required for ethanol tolerance in stationary phase (66). Furthermore, we observed a reduction lipoprotein transcript levels, which are associated with overcoming reactive oxygen species-induced stress. Additionally mRNA encoding YchH, a protein involved in reducing stress related to H₂O₂, cadmium, and acid exposure (67), was also downregulated (fold change= −2.9). Interestingly, ychH mutants show increased biofilm formation (67).

During stationary growth, melatonin also induced decreased expression of two stress related proteins, DnaK and GroES (fold change −1.4 and −1.7, respectively), and increased expression of a gene that we designated Novel00128. This novel transcript was 263 bp long and overlayed the universal stress protein F (UspF), which in the original genome annotation is 426 bp long. UspF (former YnaF) is a nucleotide binding protein, which is predicted to dimerize and to have an ATP-binding site (68). Even though uspF encodes a universal stress protein F, according to Nachin et al., uspF plays a minor role in the oxidative stress defense. Instead, UspF promotes fimbria-mediated adhesion at the expense of motility, as a uspF mutant displayed enhanced motility without an increase in the number of flagella (65).

DnaK is a chaperone protein that plays a role in both the heat shock response and a number of cytoplasmic cellular processes including the folding of newly synthesized polypeptide chains (69), rescue of misfolded proteins (70), and protein secretion (71). Overexpression of DnaK and other stress-related genes have been reported in E. coli biofilms (72). Co-chaperonin GroES synthesis increases in response to various cell stresses to accommodate protein folding (73). Previous studies in E. coli determined that when levels of GroE were reduced, the synthesis of KDPG aldolase increased (74). However, in the current study, we observed a reduction of both transcripts during stationary growth in 1 nM melatonin.

3.3.7. Metal ion transport and binding

Metal ions can be toxic to microbes in high concentrations. However, bacteria require certain metal ions for normal growth. In most cases, the requirement for these ions arises from their roles as cofactors. Examples of such metallic cofactors include iron, copper, magnesium, manganese, zinc, cobalt, and selenium (75). Melatonin has been reported to form metal complexes with aluminum, cadmium, copper, iron, lead, and zinc (76–78). Therefore, when melatonin is present in the environment, the pool of available metal ions is reduced, leading to alterations in the expression of enzymes requiring metal cofactors or metal ion transporters.

We observed that melatonin affected many genes involved in metal ion transport and binding during both growth stages (Table 8). The majority of regulated genes are involved iron binding; however, we also detected differences in genes related to copper, nickel, zinc, and selenium transport. Most metal transport genes were downregulated by melatonin, with the exception of yfaE, cusB, and rcnA. YfaE, a [2Fe-2S] cluster-containing ferredoxin present in many bacteria, including E. coli and K. aerogenes, was upregulated by melatonin during exponential growth.

Concurrently, we observed significant downregulation (3 fold) of the YkgJ ferredoxin, which was confirmed by qPCR analysis (−3.36 fold decreased transcript levels by qPCR vs. −3.13 fold decreased transcript levels by RNA-seq, Fig. 3B). Melatonin-upregulated genes observed during exponential growth included the copper efflux system (cusB) and the nickel/cobalt efflux transporter (rcnA). CusB belongs to the copper ion two component system, and is involved in the detoxification of copper and silver ions in E. coli as part of the CusCFBA operon (79). RcnA is an inner membrane ABC-type transporter that is important for Ni²⁺ and Co²⁺ homeostasis in E. coli K-12 (80). Two copper-binding proteins were also downregulated in response to melatonin – the biofilm formation regulatory protein BssS and the copper homeostasis protein CutC, which is involved in modifying intracellular copper content (81).

Two genes belonging to D-glucosaminate degradation pathway, DgaE selenocysteine synthase and DgaF, were also downregulated by melatonin. DgaE converts D-glucosaminate-6-phosphate to 2-keto-3-deoxygluconate 6-phosphate (KDGP), which is cleaved by the KDGP aldolase DgaF to form glyceraldehyde-3-phosphate and pyruvate (82). DdaE is located within a single operon that also encodes DgaF and the mannitol/fructose-specific PTS transcriptional regulator component EIIA, which were all downregulated by melatonin in stationary cells. Downregulated transcripts also included the EfeB iron-dependent peroxidase, a periplasmic protein involved in the recovery of exogenous heme iron (83).

The most downregulated gene among 30 DEGs regulated by melatonin in stationary cells was hydrogenase nickel incorporation protein HypB (fold change −2.3 by RNA-seq and −3.15 fold change by qPCR, Fig. 3D) which is involved in the coenzyme biosynthetic process. HypB is responsible for the maturation of all hydrogenase isoenzymes and is required for the incorporation of nickel into the hydrogenase large subunit in E. coli (84). During stationary growth, melatonin also downregulated uup (divalent iron ion ABC transporter), hypF (zinc ion binding carbamoyltransferase), ybiY (pyruvate formate-lyase 3-activating enzyme), hydN (formate dehydrogenase-H), and appB (Cytochrome bd-II ubiquinol oxidase subunit 2). YbiY interacts selectively and non-covalently with one [4Fe-4S] cluster. HydN also contains [4Fe-4S] cluster and facilitates electron transport from formate to hydrogen. AppB encodes a subunit of Cytochrome bd-II and is involved in generating proton-motive force (85) and uses heme as a cofactor. Collectively, these transcriptional differences may have an impact on changes in energetics in melatonin-treated cells.

3.4. Klebsiella aerogenes transcribes numerous small RNAs

RNA-seq has the ability to discover non-coding RNAs (ncRNAs). In prokaryotes, non-coding RNAs between 50 and 500 nucleotides, with a stable secondary structure, are defined as small RNAs (sRNA). sRNAs are recognized as key regulators of prokaryotic gene expression, and the majority of sRNAs regulate target genes by base pairing with target mRNAs post-transcriptionally to disrupt transcript stability and/or the translation rate (86). This process is often facilitated by the RNA chaperone Hfq which stabilizes sRNA to allow interaction with the target mRNA (87). Small RNAs are not only capable of modulating transcription, mRNA stability, and translation, but also play a role in DNA maintenance or silencing.

Our analysis identified 39 sRNA transcribed by the K. aerogenes, 14 that were encoded on the negative strand and 25 encoded on the positive strand (Table S4). sRNA expression steadily increased during culture maturation. Twenty-one sRNAs were significantly differentially expressed when comparing exponential vs stationary growth (Table 9). We also predicted the secondary structures of novel K. aerogenes sRNA (Fig. 4). While we did not further investigate the function of the identified sRNAs, we provide the results as a resource to other investigators to stimulate future experimental studies.

Figure 4.

Predicted secondary structures of selected sRNAs, A. sRNA00001, B. sRNA00053, C. sRNA00008, and D. sRNA00137, respectively.

3.5. Attachment to abiotic surfaces

Surface adhesion is the first step of bacterial colonization and biofilm formation. adhesion and attachment can be mediated through nonspecific factors such as membrane charge and surface hydrophobicity (88), or specialized adhesins located on the outer membrane, such as fimbria and other non-polymeric proteins (72). Most microorganisms express several cell surface adhesins, which enable surface recognition and attachment to various molecules, such as host tissues or abiotic surfaces. For example, E. coli K-12 implements type 1 fimbria (89, 90), flagella (89), antigen 43 (90), and curli (91) to facilitate surface attachment. Furthermore, bacteria undergo major changes at the transcriptional level during transition from planktonic to sessile conditions (72). After observing melatonin-induced transcriptional differences in expression of type 1 fimbria in our dataset, we hypothesized that melatonin increases fimbria-mediated attachment to the abiotic surfaces.

To test this hypothesis, we performed a crystal violet biofilm assay to investigate differences in attachment in static cultures incubated with 0 nM or 1 nM melatonin. Our results demonstrate that K. aerogenes grown with 0 nM melatonin accumulated on an abiotic surface at a lower rate compared to bacteria cultured in 1 nM melatonin (Fig. 5). K. aerogenes produced significantly more surface-attached biofilm in the presence of melatonin by 24 hours. Thus, increased biofilm production is likely related to melatonin-induced increased fimbrial gene expression and pilus biogenesis.

Figure 5.

Impact of 1 nM melatonin on fimbria-mediated attachment. Levels of K. aerogenes biofilm adhered to 96-well plates after 24 h static culture was quantified by crystal violet staining. Biofilm of K. aerogenes bacteria grown in TSB with 0 nM melatonin (control) or 1 nM melatonin. Error bars represent the standard error of the mean from two independent biological replicate experiments. Statistical differences were determined by two-tailed students t-test. ****, P <0.001.

4. Conclusions

Our studies demonstrate that the effects of melatonin on the K. aerogenes transcriptome are largely growth-stage specific, with the exception of two transcripts encoding the maltose regulon periplasmic protein MalM and KDGP aldolase (Figure 6). Additionally, the number of DEGs affected by melatonin is much lower than the number of DEGs identified when comparing exponential and stationary cells. Melatonin, however, regulated expression of transcripts encoding outer membrane, periplasmic, and cytoplasmic proteins. The cellular processes that were most altered by melatonin include pilus biosynthesis, biofilm formation, stress response, carbohydrate transport, and metal ion homeostasis.

Figure 6.

Graphical summary of processes that were either up or downregulated by melatonin in exponential and stationary growth phase (Created with BioRender.com).

Previous research in our laboratory reported that K. aerogenes was sensitive to melatonin which induced phenotypical differences in macrocolony size and appearance. We also detected circadian expression of a PmotA::luxCDABE reporter (7). The current RNA-seq experiment showed that motA (flagellar motor protein) expression is not directly sensitive to melatonin nor does its expression vary depending on growth stage. Therefore, the observed circadian pattern of the luciferase signal and its synchronization must be a result of another mechanism.

In the present study, we detected lower expression of several genes related to the biofilm formation during exponential growth. Additionally, we observed higher expression of various pilus genes in both exponential and stationary growth. Pilus-mediated interactions promote biofilm formation (46). Therefore, we conclude that phenotypical differences induced by melatonin were the consequence of differential expression of biofilm and pilus genes. Regarding melatonin-induced synchronization of cultures, we observed that this indole molecule downregulated the expression of 5 transcriptional regulators. Thus, these differences may be responsible for synchronization of luciferase signal between separate plates. Based on the results of this study and the observed changes in biofilm, adhesive pilus biogenesis, and stress response gene expression, we conclude that melatonin enhances K. aerogenes colonization of the host and attachment to the gut epithelium.

Supplementary Material

1

Figure S1 KEGG Scatterplots depicting KEGG pathways affected by 1 nM melatonin; the size of the circle indicates the number of genes and the color represents Q values A. Pathways downregulated during exponential growth. B. Pathways upregulated during exponential growth. C. Pathways downregulated during stationary growth. D. Pathways upregulated during stationary growth.

2

Table S1 K. aerogenes transcripts differentially expressed in exponential growth as compared to stationary growth

3

Table S2 K. aerogenes transcripts differentially expressed in exponential cells inoculated with 1 nM melatonin as compared to control cultures

4

Table S3 K. aerogenes transcripts differentially expressed in stationary cells inoculated with 1 nM melatonin as compared to control cultures

5

Table S4 Identified small RNAs. Small RNAs were identified as described in the materials and methods. Included are the genetic element, start position, end position, strand, length, associated genes, and gene functions. Orders of annotations are given according to order on the reference nucleoid.

6

Table S5 Primer sequences for qPCR validation of RNA-Seq.

Table 6.

Relative expression of carbohydrate transport and metabolism in K. aerogenes grown with 1 nM melatonin as compared to control cells

Gene ID	Product	Description	Growth stage	fold change
EAE_08405	MalM	maltose regulon periplasmic protein	exponential	3.324
EAE_08400	LamB	maltoporin lamB	exponential	2.870
EAE_08380	MalG	maltose permease	exponential	2.491
EAE 22035	PflB	pyruvate formate lyase 1	exponential	2.663
EAE_08405	MalM	maltose regulon periplasmic protein	stationary	−1.986
EAE_00580	BglA	6-phospho-beta-glucosidase	stationary	1.348
EAE_17965	BglC	Aryl-phospho-beta-D-glucosidase	stationary	1.789

Highlights.

• The pineal and gut hormone melatonin is present in the lumen of the vertebrate gut.

• This hormone affects the growth properties of at least one gut commensal bacterium, Klebsiella aerogenes, by increasing the size of microcolonies and synchronizing circadian patterns of gen expression.

• In this study, RNA-Seq was applied to determine transcriptional effects of the hormone on K. aerogenes under different growth conditions.

• Most melatonin sensitive genes are growth stage specific.

• Melatonin affects genes related to biofilm formation, fimbria biogenesis, transcriptional regulators, carbohydrate transport and metabolism, phosphotransferase systems, metal ion binding, and transport.

• The data suggest that melatonin enhances host colonization.