Identification of promoter activity in gene-less cassettes from Vibrionaceae superintegrons

Abstract

Integrons are genetic platforms that acquire new genes encoded in integron cassettes (ICs), building arrays of adaptive functions. ICs generally encode promoterless genes, whose expression relies on the platform-associated P_c promoter, with the cassette array functioning as an operon-like structure regulated by the distance to the P_c. This is relevant in large sedentary chromosomal integrons (SCIs) carrying hundreds of ICs, like those in Vibrio species. We selected 29 gene-less cassettes in four Vibrio SCIs, and explored whether their function could be related to the transcription regulation of adjacent ICs. We show that most gene-less cassettes have promoter activity on the sense strand, enhancing the expression of downstream cassettes. Additionally, we identified the transcription start sites of gene-less ICs through 5′-RACE. Accordingly, we found that most of the superintegron in Vibrio cholerae is not silent. These promoter cassettes can trigger the expression of a silent dfrB9 cassette downstream, increasing trimethoprim resistance >512-fold in V. cholerae and Escherichia coli. Furthermore, one cassette with an antisense promoter can reduce trimethoprim resistance when cloned downstream. Our findings highlight the regulatory role of gene-less cassettes in the expression of adjacent cassettes, emphasizing their significance in SCIs and their clinical importance if captured by mobile integrons.

Graphical Abstract

Graphical Abstract.

Introduction

Integrons are bacterial genetic elements that capture and rearrange genes embedded in structures known as integron cassettes (ICs) (reviewed in (1–3)). Integrons are known for their key role in the rise and spread of antibiotic resistance (AR) genes. Carried on conjugative plasmids, mobile integrons (MIs) circulate among the most important Gram-negative pathogens, disseminating >170 resistance determinants against most antibiotics (4). Yet, integrons are naturally found in almost 20% of bacterial genomes available in the databases. These are usually referred to as sedentary chromosomal integrons (SCIs) to distinguish them from MIs (5). Among SCIs, those in Vibrio species are archetypical, with very large arrays that reach 300 cassettes. Indeed, V. cholerae‘s superintegron was the first SCI discovered and is today the most studied and the paradigm in the field. Both types of integrons are not isolated form one another: mobile integrons are massively shed to the environment in the faeces of humans and animals (6). The conjugative transfer to and from environmental bacteria allows MIs to capture cassettes from SCIs (7) and to bring them back to clinical settings. There, they can easily spread if they provide an adaptive advantage, as is the case of AR genes.

All integrons are composed of a stable and a variable region. The stable region is a genetic platform that contains the integrase-coding gene (intI) with its associated promoter (P_int), the attI recombination site where integron cassettes are inserted, and the constitutive promoter P_C, that drives their transcription. The variable region of any integron is the cassette array, that can be very different between strains and represents a low-cost memory of valuable functions for the cell (Figure 1) (8).

Figure 1.

Typical structure of an integron. Integrons are composed of the integrase-coding gene (intI) and its promoter (P_int), an integron recombination site (attI), and a variable cassette array whose expression is driven by the P_C promoter. Genes are depicted as arrows. Cassette recombination sites (attC) are depicted as triangles. Garnet attCsites delimit a gene-less cassette. Integron cassettes can be integrated and excised from the platform through integrase-mediated recombination reactions between attIand attCsites. This can lead to the reshuffling of cassettes and the modulation of their expression.

Cassettes typically consist of a single promoterless gene and a recombination site, named attC (9,10). The integron integrase captures cassettes through site-specific recombination between the attI site in the platform and the attC in the cassette (11). Once acquired, cassettes are rendered functional by the activity of the P_C. Subsequent acquisition events at attI push existing cassettes away from the P_C decreasing their expression. Nevertheless, cassettes in the array can be excised through the recombination of contiguous attC sites (12) and reintegrated in first position, ultimately increasing their expression levels (13). The lack of promoters in cassettes and the reshuffling of the array, supports the role of the P_C as the exclusive promoter of the array. However, some exceptions to this rule have also been reported. Internal promoters have been identified in several ICs, such as toxin-antitoxin systems, the erythromycin resistance gene ere(A), the chloramphenicol resistance gene cmlA, or the quinolone resistance gene qnrVC1 (14–16). Yet, the impact that such promoters can have on downstream genes within arrays remains today largely unexplored.

Although ICs generally contain a single ORF that occupies most of the cassette, some studies have reported the presence of cassettes that do not contain such ORFs, and hence seem to have a biological role other than encoding a protein. It has been proposed that these gene-less cassettes may act as regulatory elements within an integron, including features like binding sites for regulatory proteins, small RNAs, or promoters, facilitating the evolution of genomes (17,18). The presence of internal promoters in gene-less cassettes has actually been experimentally demonstrated, but only in two cassettes from the atypical Treponema integrons (19).

In this study, we tested whether gene-less cassettes act as internal promoters within cassette arrays in SCIs in Vibrionaceae. To that end, we scanned the SCIs of four Vibrionaceae family species (V. cholerae, V. fischeri, V. parahaemolyticus and V. vulnificus) looking for gene-less cassettes. We identified and synthesized 29 gene-less cassettes and cloned all of them on a bidirectional reporter plasmid (pDProm), based on the expression of mCherry and gfp, and quantitated promoter activity on both DNA strands by fluorescence intensity. We found that promoter activity was common among cassettes in both strands, although with a clear bias towards the sense strand. Notably, certain gene-less cassettes led to a >100-fold increase in GFP fluorescence, suggesting that expression of cassettes can be strongly modulated by gene-less cassettes located upstream. We further corroborated these findings by confirming promoter activity through RT-qPCR. Additionally, we described the transcription start sites of a selection of ICs by performing 5′-RACE. Replacing the gfp gene for a trimethoprim resistance cassette (dfrB9), led us to find that at least two gene-less cassettes conferred a clinically relevant increase in trimethoprim resistance, both in V. cholerae and E. coli. Furthermore, we uncovered that the presence of transcription on the antisense strand within a cassette could effectively silence an upstream dfrB9 cassette, leading to a reduction in antibiotic resistance levels. We therefore reveal that gene-less cassettes from large SCIs play a regulatory role promoting the expression of distantly located cassettes, independent of their proximity to the P_C promoter. These ICs can have a strong impact in AR if integrated into mobile integrons.

Materials and methods

Bacterial strains, plasmids, and culture conditions

All bacterial strains and plasmids used in this study are listed in Table 1. All strains were cultured in both solid or liquid Luria-Bertani (LB) medium (BD) at 37°C. Kanamycin (Sigma Aldrich) was added when required at the following concentrations: 25 μg/ml for E. coli and 75 μg/ml for V. cholerae.

Table 1.

Strains and plasmids used in this study

Strains
E. coli DH5α	V. cholerae N16961	Plasmid/cassette cloned in pDProm	Origin species	Genome
Lab strain	-	-	-	-
-	(60)	-	-	-
A369	-	pSU38 Plac::gfp (pA369)	-	-
A866	B118	pDProm empty vector	-	-
A371	B950	pSU38 PCW::gfp	-	-
A370	B949	pSU38 PCS::gfp	-	-
A887	B014	PlacattC gfp	-	-
A888	B012	PlacattC mCherry	-	-
B022	B103	C1 (VCA0294-VCA0297)	V. cholerae N16961	NZ_CP028828.1
B023	B104	C2 (VCA0325-VCA0328)	V. cholerae N16961	NZ_CP028828.1
A998	B078	C3 (VCA0382a-VCA0387)	V. cholerae N16961	NZ_CP028828.1
B024	B106	C4 (VCA0389-VCA0391)	V. cholerae N16961	NZ_CP028828.1
A999	B079	C5 (VCA0392-VCA0395)	V. cholerae N16961	NZ_CP028828.1
A983	B080	C6 (VCA0403-VCA0405)	V. cholerae N16961	NZ_CP028828.1
A984	B107	C7 (VCA0428-VCA0431)	V. cholerae N16961	NZ_CP028828.1
B025	B108	C8 (VCA0436-VCA0439)	V. cholerae N16961	NZ_CP028828.1
B066	B109	C9 (VCA0460-VCA0463)	V. cholerae N16961	NZ_CP028828.1
B007	B077	C10 (VCA0498-VCA0501)	V. cholerae N16961	NZ_CP028828.1
A977	B016	F1 (VF_A0658- VF_A0659)	V. fischeri ES114	NC_006841
A978	B017	F2 (VF_A0655- VF_A0656)	V. fischeri ES114	NC_006841
A893	B020	F3 (VF_A0636- VF_A0637)	V. fischeri ES114	NC_006841
B072	B110	P1 (D0853_RS06935-D0853_RS06955)	V. parahaemolyticus VPD14	NZ_CP031781
A953	B021	P2 (D0853_RS07000-D0853_RS07035)	V. parahaemolyticus VPD14	NZ_CP031781
B073	B111	P3 (D0853_RS07065-D0853_RS07090)	V. parahaemolyticus VPD14	NZ_CP031781
A997	B081	P4 (D0853_RS07095-D0853_RS07105)	V. parahaemolyticus VPD14	NZ_CP031781
B074	B112	V1 (FOB71_ RS17245- FOB71_ RS17270)	V. vulnificus FDAARGOS_663	NZ_CP044069
B076	B114	V2 (FOB71_ RS17150- FOB71_ RS17175)	V. vulnificus FDAARGOS_663	NZ_CP044069
B069	B115	V3 (FOB71_ RS17135- FOB71_ RS17150)	V. vulnificus FDAARGOS_663	NZ_CP044069
B070	B116	V4 (FOB71_ RS167700- FOB71_ RS16720)	V. vulnificus FDAARGOS_663	NZ_CP044069
A985	B082	V5 (FOB71_ RS167700- FOB71_ RS16720)	V. vulnificus FDAARGOS_663	NZ_CP044069
A986	B083	V6 (FOB71_ RS16665- FOB71_ RS16680)	V. vulnificus FDAARGOS_663	NZ_CP044069
A987	B084	V7 (FOB71_ RS16650- FOB71_RS16665)	V. vulnificus FDAARGOS_663	NZ_CP044069
A988	B085	V8 (FOB71_ RS16635- FOB71_RS16650)	V. vulnificus FDAARGOS_663	NZ_CP044069
A989	B086	V9 (FOB71_ RS16445- FOB71_RS16460)	V. vulnificus FDAARGOS_663	NZ_CP044069
A990	B087	V10 (FOB71_ RS16125- FOB71_RS1645)	V. vulnificus FDAARGOS_663	NZ_CP044069
A991	B088	V11 (FOB71_ RS16125- FOB71_RS1645)	V. vulnificus FDAARGOS_663	NZ_CP044069
A992	B089	V12 (FOB71_ RS15850- FOB71_RS15865)	V. vulnificus FDAARGOS_663	NZ_CP044069
B071	B117	V13 (FOB71_ RS15650- FOB71_RS15665)	V. vulnificus FDAARGOS_663	NZ_CP044069
C435	C438	450-bp lacZ fragment	V. cholerae N16961	NZ_CP028827.1
C436	C439	550-bp lacZ fragment	V. cholerae N16961	NZ_CP028827.1
C437	C440	650-bp lacZ fragment	V. cholerae N16961	NZ_CP028827.1
B933	B909	dfrB9	Enterobacter cloacae p068-50543A	ID: KC675185.1
B895	B897	Plac-dfrB9	-	-
B936	B912	C7-dfrB9	-	-
B937	B913	C8-dfrB9	-	-
B935	B911	F3-dfrB9	-	-
B934	B910	P4-dfrB9	-	-
B938	B914	V13-dfrB9	-	-
C266	C268	C8-dfrB9-F1	-	-
C267	C269	C8-dfrB9-C7	-	-

Gene-less cassettes fragments design

An in silico analysis using Geneious Prime software (version 2.2) was carried out for the identification of ICs of several Vibrionaceae species that do not contain annotations nor potential genes (i.e.: ORFs smaller than 210 bp (70 amino acids)) beginning with ATG, TTG or GTG (20): nine gene-less cassettes were identified in V. cholerae N16961 (NZ_CP028828.1); three in V. fischeri ES114 (NC_006841); four in V. parahaemolyticus VPD14 (NZ_CP031781); and thirteen in V. vulnificus FDAARGOS_663 (NZ_CP044069). DNA fragments were designed for synthesis by adding to each gene-less cassette fragment its own attC site and approximately 20 bp homology upstream and downstream of the pDProm plasmid backbone (Supplementary Figure S1). An ORF-containing cassette (78 amino acids) with no gene annotation, was also selected from the superintegron of V. cholerae N16961 (cassette C7) as an ORF-carrying cassette control. It is of note that attC sites are composed of R’’, L’’, L’ and R’ boxes (Supplementary Figure S1). The consensus sequence of the R’ box is 5′-G^↓TTRRRY-3′, and it is where the crossover takes place (arrow). Hence, when a cassette is incorporated to the array, the R’ box is split in two parts: the RRRY nucleotides are at the 5′end of the cassette, fused to the attC of the previous cassette. This means that, when looking at cassette arrays, the exact bases in R’ box are not conserved, but rather contingent on the identity of the cassette downstream. When designing the cloning strategy of cassettes, we could not rule out this fact could have an influence in our results. We hence sought to add a consensus RRRY sequence to the R' box of the attC sites of our cassettes. To do so, we extracted and aligned the RRRY bases at the 3′ of all selected cassettes. The nucleotides ATGT were identified as the most frequent in this subset of attCs (Supplementary Figure S1, inset). Despite not being exactly RRRY (R is G or A but not T), this sequence was fused in all cassettes to the conserved GTT triplet, conforming a consensus R’ box that is biologically representative. We then added approximately 20 bp homology with the plasmid backbone upstream and downstream the synthetic fragments and cloned them into pDProm using Gibson assembly. Additionally, two control fragments were designed containing either gfp or mCherry under the control of a P_lac promoter upstream of a V. cholerae 128 bp-attC site (VCA0398) to detect possible effects that these recombination regions may have on transcription. The control fragments also contain approximately 20 bp homology with the plasmid backbone upstream and downstream for further cloning. The designed DNA fragments containing the gene-less cassettes and the control cassettes were synthesized by Integrated DNA Technologies (IDT). Since gene-less cassettes do not contain any annotations, they are named in this work with a letter belonging to the species they belong to, followed by correlative numbering. In Table 1, the cassettes upstream and downstream the gene-less cassettes are specified. Gene-less cassettes sequences and control fragments are listed in Supplementary Table S1.

Construction of the plasmid pDProm and derivatives

Primers used in this study are described in Supplementary Table S2. The mCherry-coding gene was synthesised by IDT with 20 bp homology, upstream and downstream, with the pA369 plasmid, a pSU38 that contains the gfp gene (Lab collection). The pA369 backbone was amplified by PCR using primers bb_pSU38_long F and bb_pSU38_long R. The mCherry-containing fragment was cloned into the pA369 amplified backbone using Gibson assembly (21), giving rise to the plasmid pDProm. Briefly, a 4 μl reaction is prepared by mixing 1 μl of insert, 1 μl of linearized plasmid and 2 μl of the 2× Gibson assembly buffer (5× ISOBuffer, 10 000 u/ml T5 exonuclease, 2000 u/ml Phusion polymerase, 40 000 u/ml Taq ligase, dH2O). The mix is incubated for 30 min at 50°C. The Gibson assembly product was then transformed into E. coli DH5α competent cells. Then, pDProm was extracted with the GeneJET Plasmid Miniprep (Thermo Scientific), according to the manufacturer's instructions.

To evaluate the promoter activity of the selected cassettes, the pDProm backbone was amplified by PCR using primers bb_pSU38_F and bb_pSU38_mCherry_R. Subsequently, all synthesized cassettes-containing fragments were cloned between the two reporter genes following the previously described Gibson assembly protocol. As an additional control, three fragments derived from the lacZ gene (VC00556) of varying sizes (450-, 550- and 650 bp) were cloned into pDProm. The amplification of the three fragments was performed with primers pDProm_lacZ_F, and pDProm_lacZ450_R, pDProm_lacZ550_R or pDProm_lacZ650_R, respectively. Gibson assembly was carried out for the cloning of the three lacZ fragments into pDProm. The resulting pDProm derivatives were then transformed into E. coli DH5α competent cells. The presence of the inserts was confirmed by PCR using primers MRVII and MFD and verified by DNA sequencing.

Transformation of pDProm plasmids in V. cholerae

All pDProm plasmids, containing the gene-less cassettes or lacZ fragments as inserts, were extracted from E. coli DH5α cultures using the GeneJET Plasmid Miniprep (Thermo Scientific), following the manufacturer's instructions. For the preparation of V. cholerae electrocompetent cells, a single colony of V. cholerae N16961 was initially inoculated in 3 ml of LB medium and cultured for 20 h at 37°C. Then, a 1:100 dilution was performed in 10 ml of LB and cultures were grown until reaching an OD₆₀₀ of 0,5. The cultures were washed three times in G buffer (137 mM sucrose, 1 mM HEPES, pH 8.0) by centrifuging at 4000 rpm and 4°C for 8 min. The final pellet was resuspended in 80 μl of G buffer.

Electrocompetent cells (40 μl) were mixed with 1 μl of each purified plasmid in a prechilled Eppendorf tube and transferred to a 0.1-cm electroporation cuvette (MBP). Electroporation was performed at 1400 V for a 5 ms pulse using the Eporator (Eppendorf). Sequentially, cells were cultured with 1 ml of LB for 1 h at 37°C. Selection of V. cholerae clones containing the pDProm and derivative plasmids was accomplished by adding kanamycin (75 μg/ml). Plasmids presence was confirmed by PCR with primers MRVII and MFD and verified by DNA sequencing.

Flow cytometry analysis

To determine promoter activity, bacterial strains carrying the pDProm plasmids were cultured on LB supplemented with kanamycin (75 μg/ml) for 20 h at 37°C. Samples were diluted 1:400 on filtered NaCl (0,9 M) in 96-well plates (Nunc, Thermo Scientific) and analysed with a CytoFLEX-S flow cytometer (Beckman Coulter). The 488- and the 661-nm lasers were used for GFP and mCherry excitation, respectively. GFP and mCherry fluorescence intensity was detected through 525/40 nm (FITC) and 610/20 nm (ECD) band pass filters, respectively. At least three biological replicates and two technical replicates were used, and 30 000 events per sample were recorded. Flow cytometry data were analysed with the CytExpert software (version 2.4; Beckman Coulter).

RNA isolation and RNA-seq data analysis

Total RNA was extracted from V. cholerae N16961 cultures grown in LB at 37°C in both exponential (OD₆₀₀ 0,8) and stationary (OD₆₀₀ 2,8) growth phases using the RNeasy® Mini Kit (QIAGEN), following the manufacturer's protocol. To eliminate any residual DNA, RNA was treated using the TURBO DNA-free™ Kit (Invitrogen). RNA concentration was measured using a BioSpectrometer (Eppendorf), and RNA integrity was determined using a Qubit™ 4 fluorometer (Invitrogen) with the RNA IQ Assay Kit (Invitrogen). Ribodepletion RNA library and sequencing was performed at the Oxford Genomics Centre using a NovaSeq6000 sequencing system (Illumina). Three biological samples per condition were sequenced. Additionally, RNA-Seq data of V. cholerae N16961 from an independent study (BioProject PRJNA420494; SRA identifiers: SRR6334020, SRR6334021 and SRR6334022) (22) was also analysed.

Raw reads from both studies were trimmed using Trim Galore v0.6.6 (https://github.com/FelixKrueger/TrimGalore) with a quality threshold of 20 and removing adapters and reads shorter than 50 bp. Trimmed paired reads were mapped to the V. cholerae N16961 reference genome (Accession number: chromosome 1 CP028827.1; chromosome 2 CP028828.1) using BWA-MEM v0.7.17 (23). Read count data was obtained with featureCounts from the Rsubread v2.10.2 package (24). The expression levels, given in TPM values (transcripts per kilobase million) for all protein-coding genes, were calculated with the Geneious Prime ‘Calculate Expression Level’ function and in R v4.2.2 (www.R-project.org) from FPKM values obtained with DESeq2 v1.36.0 (25). RNA-seq data are accessible through GEO Series accession number GSE229776 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE229776).

Quantitative real-time PCR

Total RNA was extracted, as previously described, from three biological replicates of strains B118, B012, B014, B079, B016, B017, B084, and B088, grown in LB at 37°C for 20 h. To eliminate any residual DNA, the extracted RNA was treated using the TURBO DNA-free™ Kit (Invitrogen). RNA concentration was measured using a BioSpectrometer (Eppendorf). Prior to cDNA synthesis, a DNA wipe-out step was conducted using the QuantiTect® Reverse Transcription Kit (QIAGEN), following the manufacturer's instructions. Subsequently, cDNA synthesis was carried out using the same kit with the following temperature steps: 42°C for 15 min, followed by 95 °C for 3 min. The obtained cDNA was diluted 10-fold and served as template in the qPCR.

RT-qPCR was performed in an Applied Biosystems QuantStudio 3 using the QuantiTect® Multiplex PCR Kit (QIAGEN). The primers and probes used are listed in Table S2. The qPCR protocol consisted of a HotStarTaq DNA Polymerase activation step at 95°C for 15 min, followed by 40 cycles of denaturation at 94°C for 60 s, and annealing/extension at 60°C for 60 s. Relative changes in the expression of the gfp and mCherry genes in all the tested strains, relative to the B118 strain (harboring the pDProm empty plasmid), were determined according to the threshold cycle method (2−ΔΔCT). The results were normalized against the expression of the housekeeping gene gyrA, as described in (26). Mean values were calculated from three independent biological replicates, each with three technical replicates.

5′ Rapid amplification of cDNA ends

Two micrograms of previously isolated RNA from strains B079, B016, B017, B084 and B088 were used to identify transcription start sites using the 5′ RACE System for Rapid Amplification of cDNA Ends, version 2.0 (Invitrogen), in accordance to the manufacturer's protocol. The sequences of the primers used are listed in Table S2. First-strand cDNA was synthesized from total RNA using a gene-specific primer (GSP1). The cDNA products were subsequently amplified by PCR using Dream Taq DNA polymerase (Thermo Scientific), a nested gene-specific primer (GSP2) design to anneal to a site within the cDNA molecule, and a novel kit-supplied deoxyinosine-containing anchor primer. A cDNA control PCR was performed using sense gene-specific primer (GSP3), in conjunction with the GSP2 primer. If required, a second nested gene-specific primer (GSP4) was used for PCR amplification. PCR products were then analysed by agarose gel electrophoresis and purified using the GeneJET PCR purification kit (Thermo Scientific). The purified PCR products were subsequently cloned using the TOPO TA Cloning® kit (Invitrogen) and transformed into DH5α subcloning efficiency competent cells (Thermo Scientific), following the manufacturer's instructions. At least six independent clones from each cassette were selected for DNA sequencing. The transcription start sites (+1 site) were determined by aligning the obtained sequences to the corresponding reference genome, being the +1 the nucleotide immediately following the poly(dC) tail.

Bicyclomycin assay

To inhibit the activity of the Rho termination factor, a bicyclomycin assay was performed. Strains B118, B012, B014, B017, B079, B088, B084 and B016 were cultured on LB supplemented with kanamycin (75 μg/ml) for 20 h at 37°C. Samples were diluted 1:100 and bicyclomycin was added to the cultures at a concentration of 50 μg/ml at early exponential phase. After 2 h-incubation at 37°C, cultures were diluted 1:400 on filtered NaCl (0.9 M) in 96-well plates (Nunc, Thermo Scientific) and analysed with a CytoFLEX-S flow cytometer (Beckman Coulter), as above described.

Construction of the pDProm-dfrB9 plasmid and derivatives

In order to test the inducibility of antibiotic resistance due to the presence of promoter-containing gene-less cassettes, the resistance cassette dfrB9, which confers resistance to trimethoprim, was selected from the IntegrALL database (27) and synthesized by IDT. To obtain the pDProm-dfrB9 plasmid, the pDProm backbone was amplified excluding the gfp gene, while the dfrB9 was amplified with 20 bp homology with the backbone. The dfrB9 gene was then cloned into the backbone by Gibson assembly. The same procedure was performed with pDProm-P_lac-attC-gfp (the attC region is not included in this plasmid version), and with pDProm-C7, -C8, -F3, -P4 and -V13 plasmids. The backbones were amplified excluding the gfp gene, and the dfrB9 gene cassette was amplified with 20 bp homology with the corresponding backbone for further cloning downstream the gene-less fragments using Gibson assembly. The pDProm-dfrB9 plasmid and all the derivatives were transformed in V. cholerae N16961 and E. coli DH5α following the above-described protocols.

For the cloning of cassettes F1 and C7 downstream the dfrB9 in pDProm-C8-dfrB9, both cassettes F1 and C7 were amplified with 20 bp homology with the plasmid backbone. The plasmid pDProm-C8-dfrB9 was amplified and cassettes F1 or C7 were then cloned into the backbone by Gibson assembly. The obtained plasmids pDProm-C8-dfrB9-F1 and pDProm-C8-dfrB9-C7 were transformed in V. cholerae N16961. Plasmids presence was confirmed by PCR with primers MRVII and MFD and verified by DNA sequencing. Primers used for pDProm-dfrB9 plasmids construction are listed in Supplementary Table S2.

Determination of the minimum inhibitory concentration (MIC) of trimethoprim

The MICs to trimethoprim of V. cholerae N16961 and E. coli DH5α containing pDProm-dfrB9 and derivatives were performed following CLSI broth microdilution recommendations. Briefly, 10⁵ UFCs from an overnight culture were inoculated in 96-well plates (SPL Life Sciences) containing 200 μl of fresh MH (Oxoid) with doubling concentrations of trimethoprim (Sigma Aldrich). Plates were incubated at 37°C for 20 h without shaking. The MIC value was defined as the minimal concentration at which bacterial growth was inhibited in two biological replicates with two technical replicates each.

Results

Identification of gene-less cassettes from Vibrionaceaegenomes

We explored the data bases and selected four different genomes belonging to four different Vibrio species: V. cholerae N16961 (NZ_CP028828.1), V. fischeri ES114 (NC_006841), V. parahaemolyticus VPD14 (NZ_CP031781) and V. vulnificus FDAARGOS_663 (NZ_CP044069) with the aim of finding gene-less cassettes. The selection of these genomes was driven by several key factors: (i) they contain SCIs; (ii) these SCIs are of considerable length, with most of the array presumed to be silent if only the P_C is active and (iii) experimental evidence exists for cassette recruitment by MIs (28).

We selected all cassettes lacking annotated genes, further focusing on those not encoding ORFs exceeding 210 bp (70 amino acid-long proteins) (see discussion section). According to these constrains, a total of 29 cassettes within the four selected species did not contain protein encoding genes. The distribution of these gene-less cassettes was as follows: nine in V. cholerae N16961; three in V. fischeri ES114; four in V. parahaemolyticus VPD14; and thirteen in V. vulnificus FDAARGOS_663. Additionally, we incorporated a cassette from V. cholerae N16961 that, although not containing an annotated gene, contained an ORF of a size slightly above the limit (78-amino acid). Altogether, our analysis yielded 30 cassettes.

Construction and validation of the reporter plasmid pDProm

We designed a bidirectional reporter plasmid for the Detection of Promoter activity (pDProm) to use it with the gene-less cassettes from SCIs of Vibriospecies. This plasmid contains two divergent promoterless genes encoding the fluorescent proteins GFP and mCherry, allowing for the simultaneous detection of promoter activity in both DNA strands of a given insert (Figure 2A). To validate the utility of pDProm, we wanted to mimic the structure of an integron cassette containing a promoter and verify its activity. To do so, we inserted a P_lac promoter upstream a superintegron attCsite -also known as Vibrio cholerae repeats (VCRs) (29)- (the 128 bp attCsite of cassette VCA0398 in V. cholerae N16961), in both orientations into pDProm (Figure 2B). We then introduced both plasmids in V. cholerae N16961 and we compared fluorescence intensity of GFP and mCherry in these strains with the values of the empty pDProm. As shown in Figure 2C, the P_lac promoter increased GFP fluorescence 295-fold in P_lac-attC-gfp, and mCherry intensity 60-fold in P_lac- attC-mCherry. The lower fluorescence of the mCherry reporter is likely due to maturation or stability differences between fluorescent proteins. Indeed, unlike GFP, mCherry fluorescence does not increase during bacterial log phase and varies likely in response to different growth rates (30). However, a 60-fold increase is significant (P-value < 0.0006) and provides a good dynamic range to test promoter activity. No significant differences were observed between the GFP and mCherry fluorescence raw values given by V. cholerae N16961 and the empty-pDProm strain (Supplementary Figure S2). We hence conclude that the use of pDProm is suitable for the detection of promoters in DNA fragments.

Figure 2.

Structure and validation of the pDProm reporter plasmid.(A) Structure of the plasmid pDProm containing the reporter genes mCherryand gfp, the kanamycin resistance gene and the p15A origin of replication. The intergenic region sequence between gfpand mCherryis depicted including the ribosome binding sites (RBS) for both proteins (grey boxes and font) and the start codons (magenta and green font). (B) Schematic representation of the control DNA fragments. The pDProm-P_lac-attC-gfpplasmid carries the gfp under the control of a P_lac promoter upstream an attCsite fromV. cholerae N16961. The pDProm-P_lac-attC-mCherryplasmid carries the mCherryunder the control of a P_lac promoter upstream an attCsite from V. cholerae N16961. (C) Fold change values for fluorescence intensity given by pDProm-attC-gfpand pDProm-attC-mCherryrelative to the the values obtained from the empty pDProm (dashed line). Error bars represent standard deviation of fluorescence measurements of three biological replicates with two technical replicates each. The P-values (*) were calculated by comparing each measure with that of the pDProm using paired t-test (**P < 0.01–0001; *** P < 0.0001).

We synthesized the 30 gene-less cassettes identified previously with their respective attCsites and cloned them into pDProm. We introduced all plasmids in V. cholerae N16961 and verified the sequence of all constructs. For the sake of simplicity, cassette names are dubbed with the first letter of the species they belong to, and a correlative number.

Determination of promoter activity

Using pDProm, cassettes containing a promoter on the sense strand will show an increase in GFP intensity, while those with a promoter on the antisense strand will present an increment in the mCherry fluorescence intensity. We have defined as sense strand the strand that would be under the influence of the P_C promoter. From the ten cassettes selected from V. cholerae N16961, eight showed significant promoter activity on the sense strand, with 9 to 66-fold increases in GFP fluorescence. Among those with promoter activity, cassettes C5, C8 and C10 also showed significant promoter activity on the antisense strand, ranging from 2.8- to 3.5-fold (Figure 3A). Regarding the four gene-less cassettes from V. parahaemolyticus VPD14, three of them (P1, P2 and P4) showed 8 to 28-fold increases in GFP fluorescence. The P1 cassette also gave a 2.5-fold increase in the fluorescence intensity by the antisense strand (Figure 3B). After the screening of the thirteen gene-less cassettes from V. vulnificus FDAARGOS_663, we identified promoter activity on the sense strand in eleven of them (all but cassettes V7 and V12), with increment values varying from 5- to 20-fold in comparison with the empty pDProm. Besides, we also found significant promoter activity on the antisense strand in cassettes V2, V3, V11 and V12, all of them with around 2.5-fold increase in mCherry fluorescence intensity (Figure 3C). Finally, we analyzed the gene-less cassettes from the superintegron of V. fischeri ES114 and detected significant promoter activity in all of them on the sense strand with fold changes ranging from 27- to 101-fold. We also detected the highest promoter activity on the antisense strand in cassette F1 with a 14-fold increase in mCherry fluorescence (Figure 3D).

Figure 3.

Determination of promoter activity of Vibrionaceaegene-less cassettes using pDProm. GFP (sense) and mCherry (antisense) fluorescence intensity fold change given by the selected cassettes from the superintegrons of (A) V. cholerae N16961, (B) V. parahaemolyticus VPD14, (C) V. vulnificus FDAARGOS_663 and (D) V. fischeri ES114, relative to the values given by the empty pDProm (dashed lines). (E) Fluorescence histograms and (F) fold change conferred by the P_CW and P_CS promoters in V. cholerae N16961 relative to the values given by the empty pDProm (dashed line). Error bars represent standard deviation of fluorescence measurements of three biological replicates with two technical replicates each. Histograms are the representation of one biological replicate. The P-values (*) were calculated by comparing each strain with the negative control strain B118 (empty pDProm) using one-way ANOVA with the Dunnett's multiple comparisons test (* P < 0.01–0.001; ** P < 0.001–0.0001; *** P < 0.0001).

To validate our findings, we introduced three control DNA fragments from the lacZ gene into pDProm, with sizes corresponding to those of the tested gene-less cassettes (450-, 550- and 650-bp), and measured the fluorescence intensity by flow cytometry. As expected, we observed no increase in the GFP or the mCherry expression levels when compared to the wildtype strain or the levels given by the empty pDProm vector (Supplementary Figure S3). According to these results, promoter activity seems to be prevalent among gene-less cassettes.

pDProm has a p15A origin of replication with 9–17 copies of the plasmid per cell (31–33). Hence, one could argue that the increases in fluorescence are due, at least in part, to the effect of copy number. So, in order to contextualize our results, we used a similar setup to measure the fluorescence levels of two variants of the P_C promoter commonly found in Class 1 integrons: the P_CS, which is considered a ‘strong’ promoter, and the P_CW, a ‘weak’ promoter. Previous reports show that P_CS is 30-fold stronger than P_CW (34,35). We cloned these promoters upstream the gfp gene in the plasmid pA369 (also a p15A-based replicon), introduced them in V. cholerae N16961 and measured fluorescence by flow cytometry (Figure 3E). In our system, the fluorescence increments given by V. cholerae N16961 harboring the plasmids PcW-gfp or PcS-gfp, relative to the empty pDProm, were 19- and 525-fold, respectively (a 27-fold difference between both constructs, which is in accordance to the literature) (Figure 3F). By comparing these values with those obtained for the Vibrionaceae gene-less cassettes, we can conclude that most of them present a promoter strength similar to that of P_CW, with the exception of cassettes C5, C8, F2 and F3, which have promoter strengths between P_CW and P_CS.

Determination of promoter activity of gene-less cassettes by RT-qPCR

To demonstrate promoter activity at a transcriptional level, we performed RT-qPCR from a selected subset of strains, chosen for their diverse range of gfp and mCherry expression levels, previously measured by flow cytometry. We specifically targeted gene-less cassettes that exhibited robust promoter activity on the sense (F2, C5 and F1) and antisense strand (F1 and F2), alongside cassettes with mild or almost no such activity on both strands (V11 and V7, respectively) (Figure 4A and 4B). Transcription levels of both reporter genes were also measured in strains harboring pDProm-P_lac-attC-gfp and pDProm-P_lac-attC-mCherry, and were compared to the levels given by the empty vector. As shown in Figure 4C and D, we observed different transcription levels of the gfp and mCherry given by the expression of the selected cassettes. In most cases, the obtained transcription levels correlate well to the expression observed by measuring fluorescence, corroborating the consistency of our previous observations.

Figure 4.

Validation of gene-less cassettes expression levels by RT-qPCR. (A) Fold change in fluorescence of GFP (sense) and (B) mCherry (antisense) given by the Vibrionaceae-selected cassettes relative to values obtained by the empty pDProm (dashed line). Error bars represent standard deviation of fluorescence measurements of three biological replicates with two technical replicates each. (C) Transcription levels of gfp and (D) mCherry determined by RT-qPCR from the selected Vibrionaceaegene-less cassettes relative to the empty pDProm. Error bars represent standard error of three independent biological replicates with three technical replicates each.

Identification of transcription start sites in gene-less cassettes

Having confirmed promoter activity in specific V. cholerae gene-less cassettes by measuring the expression of the reporter genes gfp and mCherry by both flow cytometry and RT-qPCR, we proceeded to further characterize the potential transcription start sites (TSSs) (+1). To accomplish this, we performed the 5′ rapid amplification of cDNA ends (5′ RACE) technique, selecting the same gene-less cassettes (F1, F2, C5, V7 and V11) for which transcription activity had previously been validated via RT-qPCR. Following 5′ RACE, the resulting cDNA products were amplified with cassette-specific oligos and cloned into a pTOPO plasmid. Sequencing of a variety of clones revealed the TSSs positions for each selected gene-less cassette (Figure 5). Notably, TSSs were identified at different positions within the length of the cassette, and even within the attC recombination site, as in cassette F2. While certain cassettes exhibited a concentration of TSSs within a specific region, potentially indicating a single promoter (cassettes F1, C5 or V7), others displayed TSSs distributed at different points along the cassette at equal frequencies, as cassettes F2 and V11, suggesting a potential spurious transcription effect. Of particular interest, in a previous study Krin et al. located TSSs in six of the cassettes for which we have found promoter activity (C2, C3, C5, C6, C9 and C10). Our results for C5 are in accordance with the TSS at position 359 in cassette C5 (36), both validating our results and supporting the rest of TSSs in other cassettes. Furthermore, we also performed the 5′ RACE assay on the antisense strand of cassette F1, where high expression levels were observed by measuring mCherryexpression. As depicted in Figure 5, two main TSSs on the F1 antisense strand were identified, corroborating the presence of transcription on both strands in this gene-less cassette by distinct promoters. The putative promoter sequences are listed in Table S3.

Figure 5.

Transcription start sites (TSS) of Vibrionaceaegene-less cassettes. Schematic illustration of the gene-less cassettes F1, F2, C5, V7 and V11 from Vibrionaceae, indicating the detected TSSs identified through 5′ RACE. The height of the peaks corresponds to the frequency of detection for each TSS along the cassette. The peaks of cassette F1 below the line indicate the identified TSSs on the antisense strand. The blue point in cassette C5 represents a previously reported TSS position (36). Light green arrows represent the attC sites within each cassette.

Next, we wanted to understand whether spurious transcription is at play, and the potential role of the transcriptional terminator Rho -a terminator of spurious transcription (37). We measured gene expression in an assay using the Rho inhibitor bicyclomycin, and measured fluorescence levels by flow cytometry after 2h-incubation. Thus, if Rho is involved in the transcriptional termination in gene-less cassettes, its inhibition would result in increased fluorescence levels due to higher number of transcripts reaching the fluorescence reporter gene. Despite the inhibition of Rho, we did not observe any substantial enhancement in the GFP or mCherry fluorescence signals in comparison with the non-treated strains (Supplementary Figure S4), suggesting that Rho's role in regulating cassette expression, if any, is absent or very limited in our conditions.

Identification of promoters in gene-less cassettes cannot be performed in silico.

Given the number of gene-less cassettes with promoter activity, we wanted to know whether this could have been accurately predicted in silico. To do this, we used BPROM software (Softberry.com) (38) and De Novo DNA (39) to predict the presence of putative σ⁷⁰ promoters or transcription initiation rates, respectively, on both sense and antisense strands of the cassettes. BPROM identified promoters in both strands of all fragments tested and, in some cases, more than one promoter was predicted. The LDF (Linear Discriminant Function) score provided by BPROM is an indicator of promoter strength. The predicted -10 and -35 boxes and scores for all the cassettes are described in Supplementary Table S4. As controls, we also analyzed the previously described known promoters P_CS, P_lac and P_CW, which had LDF values of 5.97, 3.94 and 3.10, respectively. We plotted, for all the cassettes, the LDF values for the sense (Supplementary Figure S5A) and antisense (Supplementary Figure S5B) strands along with the mean of GFP and mCherry fluorescence levels, respectively. We calculated the squared Pearson correlation coefficient and could not find a strong correlation between in silico and in vivo promoter strength (insets; sense strand: R² = 0037; P= 0.3; antisense strand R²= 0.2; P= 0.01). The use of De Novo DNA software identified TSSs in silico that did not match those detected experimentally by 5′ RACE (Supplementary Figure S6). Generally, none of these algorithms would have allowed to identify bona fide promoters, confirming the need for experimental assessment.

Transcription levels of integron cassettes from V. cholerae N16961 superintegron

Our results show that chromosomal integrons in Vibrio spp. have a wealth of gene-less cassettes with promoter activity scattered along the array. For instance, V. cholerae N16961 contains nine such cassettes, eight of which show at least half of the strength of a P_CW and are therefore biologically relevant. These have to be added to the list of cassettes that encode a gene with a dedicated promoter, such as toxin-antitoxin systems, among others. It is therefore plausible that the operon-like model of expression is more complex on SCI, with many promoters along the array.

In order to investigate the levels of expression within the V. cholerae superintegron, and to contextualize the effect of gene-less cassettes, we performed RNA-seq for the V. cholerae strain N16961 in exponential and stationary growth phase conditions. We analyzed the data as described in Materials and Methods and expression levels of superintegron cassettes were measured calculating the number of transcripts per kilobase million (TPM), as depicted in Figure 6. We excluded from the analysis duplicated cassettes or those containing repeats of at least 300 bp, to avoid biased reporting of expression values of regions for which discrimination power is low. To provide an idea of physiological levels of gene expression, we also calculated the TPM values for eight housekeeping genes. As a validation of our results, the sigma factor RpoS shows higher levels of expression in stationary phase, as expected.

Figure 6.

Expression levels of V. cholerae N16961 superintegron gene cassettes. (A) Expression levels represented as transcripts per kilobase million (TPM) of gene cassettes along the superintegron array, and eight housekeeping genes used as a reference, were calculated from RNA-seq data obtained from exponential (XP) and (B) stationary (ST) growth conditions. TPM values of genes-less cassettes (magenta), toxin-antitoxin systems (mauve and pink), integrase (orange), and the chloramphenicol resistance gene catB9(yellow), are highlighted. Error bars represent the standard deviation of three independent biological replicates.

Our data show that—in addition to TA systems that contain their own promoter—, many cassettes are expressed at detectable levels. TPM values show broad variation along the array, with 210-fold and 500-fold differences between the most and least expressed cassettes in exponential (Figure 6A), and stationary phase, respectively (Figure 6B). Regions or ‘blocks’ of gene cassettes that present similar expression levels are also detected throughout the length of the superintegron. The expression levels of many of the cassettes are in the same range as those of housekeeping genes, proving the physiological relevance of our measures. Our results were further verified by analyzing the data from an independent RNAseq from the same strain available in the databases (BioProject PRJNA420494) (22). We found a similar distribution of reads along the array (Supplementary Figure S7), allowing to reach the same general conclusions.

To reveal the role of gene-less cassettes in their natural context we took a closer look at the mapping of transcripts of the 6 cassettes with expression levels similar to PcW (C2 to C6 and C8). Our analysis found technical and biological limitations. Cassettes C5 and C8 contained repeated regions and were discarded to avoid mapping biases; and cassette C3 could not be clearly interpreted (Figure 7B) because it is located immediately upstream of the parD/E3 toxin-antitoxin system encoding a promoter in the opposite orientation. Of the remaining cassettes, C4 and C6 showed only mild increases in transcript levels (Supplementary Figure S8), while a clear increase in transcription could be observed inside the sequence of C2 (Figure 7A), supporting its role as promoter of downstream cassettes. Our results indicate that the superintegron is not a ‘silent’ structure and that the expression of gene cassettes in the array does not rely solely on the P_C promoter. We also find evidence of a gene-less cassette acting as a promoter in its native context.

Figure 7.

Representation of reads from the RNAseq of V. cholerae N16961 superintegron. (A) The number and orientation of reads, and the coverage upstream and downstream gene-less cassettes C2 and (B) C3 were analyzed by Integrative Genomics Viewer (IGV) software. Blue color designates genes and reads from the plus (sense) strand; red color designates genes and reads from the minus (antisense) strand; uncolored reads indicate unknown status or reads repetitions.

Promoter-carrying gene-less cassettes on the sense strand can trigger antibiotic resistance

Given the possibility that mobile integrons recruit cassettes from SCIs with promoter activity, we aimed to investigate their potential role in promoting antibiotic resistance when located upstream of a silent resistance gene. To test this, we selected the plasmids containing the gene-less cassettes with the highest promoter activity from each Vibrio species (C8, P4, V13 and F3) and replaced the gfpgene with the promoterless trimethoprim resistance cassette dfrB9. As negative and positive controls, we constructed pDProm-dfrB9and pDProm-P_lac-dfrB9, respectively (Figure 8A). Last, we also tested the C7 cassette from V. cholerae N16961, that encodes a putative 78-amino acid long protein. As we did not detect fluorescence increase in either the C7 sense or antisense strands (Figure 3A), we used this promoterless cassette as a control. We introduced all plasmids in V. cholerae N16961 and performed MIC assays to trimethoprim. We found that the four selected gene-less cassettes increased trimethoprim resistance from two- and four-fold (cassettes C8 and V13) to more than 500-fold in the case of P4 and F3 (Figure 8B). The same result was obtained for the positive control P_lac-dfrB9, while strains with C7 or the empty pDProm-dfrB9showed the same level of resistance as the strain with no plasmid.

Figure 8.

Promoter-carrying gene-less cassettes induce trimethoprim resistance. (A) Fragments of the pDProm-dfrB9 plasmids where the gfp gene is replaced by the trimethoprim resistance cassette dfrB9 in pDProm-P_lac and pDProm-‘gene-less’ plasmids. (B) Minimum inhibitory concentration (MIC) to trimethoprim and resistance fold change of V. cholerae N16961 and (C) E. coli DH5α containing pDProm-dfrB9 plasmids with the cassettes yielding the highest fluorescence intensity from each Vibrio (C8, P4, V13 and F3). Strains containing P_lac-dfrB9 or pDProm-dfrB9 (dfrB9, empty vector) were used positive and negative controls of trimethoprim resistance, respectively. C7-dfrB9 was used as promoterless control. The resistance levels of V. cholerae N16961 and E. coli DH5α strains with no plasmids (-) are indicated with a dashed line. MIC values above 128 μg/ml are represented with (>). The MIC of each strain was averaged from two biological replicates with two technical replicates each.

Since ICs can be mobilised and transferred through horizontal gene transfer among unrelated species, we were interested in the effect that these gene-less cassettes may have in the resistance levels when present in a different species. To address this, we transformed these plasmids in E. coliand measured the resistance levels to trimethoprim. In accordance to what we observed in V. cholerae, cassettes P4, F3 and the P_lac-dfrB9control showed 2000-fold increases in trimethoprim resistance, while strains with the C7 control, or the empty pDProm-dfrB9plasmid, showed no differences with the empty strain. These results suggest that promoter-carrying gene-less cassettes, when located upstream a resistance gene cassette in a plasmid, are able to induce high levels of resistance not only in vibrios, but also in more phylogenetically distant species as E. coli.

Promoter-carrying gene-less cassette on the antisense strand reduce antibiotic resistance levels

Given that promoter activity was also detected on the antisense strand of some cassettes, we decided to test their influence on the expression of upstream cassettes. We hypothesize that if antisense promoters are functional, we would observe a decrease in trimethoprim resistance if cloned downstream dfrB9. We selected cassette F1 from V. fischeri, which showed a 14-fold increase in mCherry fluorescence and where two TSSs have been detected, and cloned it in pDProm-C8-dfrB9 downstream dfrB9. It is of note that in this setup dfrB9 expression is dependent on the C8 cassette. As a control, we also cloned the cassette C7 downstream dfrB9 (Figure 9A), which did not show promoter activity. We introduced both plasmids in V. cholerae and measured the MIC to trimethoprim. After 24 h (Figure 9B), we observed that the MICs of strains where dfrB9 was expressed (pDProm-C8-dfrB9 and pDProm-C8-dfrB9-C7) were 2- to 4-fold higher than those of strains without plasmid or carrying a silent dfrB9. In the presence of F1, the MIC decreased to the same levels as in the case where dfrB9 is silent or absent (Figure 9B). This effect was accentuated after 48 h of incubation, when strains expressing dfrB9 reached above 8-fold higher MICs, while controls and the strain carrying the pDProm-C8-dfrB9-F1 retained the same values (Figure 9C). To better characterize these differences, we also performed growth curves of V. cholerae with all plasmids in the presence of inhibitory concentrations of trimethoprim (0.25 and 0.5 μg/ml). As observed in Figure 9D, all strains grew similarly in MH, but only those carrying pDProm-C8-dfrB9 or pDProm-C8-dfrB9-C7 were able to grow in the presence of the antibiotic. Only after 32 h of incubation, one out of four replicates of V. cholerae with pDProm-C8-dfrB9-F1 started growing, probably due to the presence of spontaneous mutations. These data show, for the first time, that the presence of a promoter on the antisense strand of a gene-less cassette can suppress the expression of a gene, leading, in this case, to a reduction of antibiotic resistance.

Figure 9.

Gene-less cassette carrying a promoter on the antisense strand act as a transcriptional terminator. (A) Fragments of the plasmid pDProm-C8-dfrB9 where cassette F1, as a cassette containing a promoter on the antisense strand, and C7, as a promoterless cassette control, are cloned downstream dfrB9 (pDProm-C8-dfrB9-F1 and pDProm-C8-dfrB9-C7). (B) Minimum inhibitory concentration (MIC) to trimethoprim and resistance fold change at 24 h and (C) 48 h of V. cholerae N16961 containing plasmids pDProm-C8-dfrB9, pDProm-C8-dfrB9-F1 or pDProm-C8-dfrB9-C7. The strain containing pDProm-dfrB9 (dfrB9, empty vector) was used a negative controls of trimethoprim resistance. The resistance levels of V. cholerae N16961 with no plasmids (–) are indicated with a dashed line. MIC values above 8 μg/ml are represented with (>). (D) Growth curves of V. cholerae N16961 containing plasmids pDProm-C8-dfrB9, pDProm-C8-dfrB9-F1 or pDProm-C8-dfrB9-C7 in the absence (MH) or presence of trimethoprim at 0.25 and 0.5 μg/ml. Error bars represent standard deviation of at least three independent biological replicates.

Discussion

The Vibrionaceae family comprises bacteria that are ubiquitously distributed throughout aquatic environments and includes important pathogens. V. cholerae is the causative agent of Cholera disease, that still causes approximately 100 000 deaths worldwide annually (40); while V. parahaemolyticus and V. vulnificus cause acute gastroenteritis, wound necrosis and/or sepsis in humans due to contaminated water or seafood consumption (41,42). A notable characteristic of the Vibrionaceae family is the presence of large SCIs in their genomes, carrying a highly diverse repertoire of ICs (43,44). These have likely been important agents in the evolution of vibrios, a phenomenon for which knowledge is scarce (45). It is therefore important to decipher the functions of ICs by revealing the role of the proteins they encode, yet, previous studies have shown that a considerable proportion do not appear to encode any (from 6% in V. cholerae to 49% in V. vulnificus CMCP6) (46). In this study, we have experimentally detected promoter activity in gene-less cassettes from the sedentary chromosomal integrons of four Vibrio species.

The general integron model posits that the P_C promoter drives the expression of cassettes and that the increasing distance to the P_C causes their silencing. Some exceptions to this rule are known, mainly in the form of cassettes encoding genes with dedicated promoters. The presence of promoters in cassettes within SCIs, and their impact on adjacent cassettes have seldom been investigated, although their presence has been previously suggested (47). In a study of the array of V. rotiferianus, the authors detected through RT-qPCR patterns of en bloc transcription along the array, suggesting the presence of internal promoters (48). In Krin et al.’s work, the authors performed the mapping of the transcription start sites (TSSs) in V. cholerae N16961, showing that many TSSs are scattered along the superintegron (36). In both studies, the origin of promoter activity was not linked to a specific type of cassette. Tansirichaiya et al. (19) were the firsts to report promoter activity in gene-less cassettes of Treponema spp. from the human oral metagenome (49). It is worth noting that this work only characterized three cassettes and that SCIs in Treponema are atypical because they encode the integrase in the same strand as the cassettes (50,51). Therefore, it is difficult to know if the presence of promoter cassettes is extrapolable to integrons in general. Here, we aimed to provide additional experimental evidence on the presence of such cassettes in canonical integrons. We have done so by tackling three limitations of the previous work. First, we test a larger number of cassettes (30 versus 3) from a variety of species, providing a better grasp on the universality of the phenomenon. Second, in the previous study, the promoter activity of Treponema spp. cassettes was tested in E. coli, a very distantly related species. It is hence possible that the promoters identified might not be acting as such in the original host. Here, we test promoter activity in V. cholerae, which is the host of several cassettes, and it is a species closely related to the other hosts. Third and last, we confirm our fluorescence results through RNA-seq, RT-qPCR and 5′ RACE, and we explore the biological consequences of the presence of such promoters. We demonstrate their importance in modulating antimicrobial resistance provided by the dfrB9 IC and prove that this effect is conserved in enterobacteria, where mobile integrons are commonplace and important actors in multidrug resistance. It is of note that we show for the first time that antisense promoters can interfere with the expression of upstream cassettes, acting as array silencers. These data also confirm the previous statement that attC sites do not act as transcriptional terminators (52). These results evidence the impact that such cassettes could have on AR if mobilized. Interestingly, some MIs contain gene cassettes of unknown function (gcu), many of which are very small and probably do not contain genes. Thus, it is possible that promoter cassettes are already modulating resistance genes in clinical settings, a matter that needs further investigation.

An important aspect of this and previous works is the definition of gene-less (or ORF-less) cassettes. Here, we have defined them as cassettes lacking annotated genes and not containing ORFs longer than 210 bp. A limitation of our work (and others) is that we have not experimentally proven that smaller ORFs do not encode proteins (nor small RNAs for that matter). However, there are several reasons why we believe our definition is sensible. First, cassette architecture is generally very compact, with ORFs occupying most -if not all- of the cassette sequence, sometimes even overlapping the attC site. The cassettes selected here range in size from 491 to 666 bp and attC sites of 101–127 bp. This leaves coding regions of sizes between 335 and 515 bp. Hence, the presence of protein coding genes smaller than 210 bp would be at odds with the general compaction of cassette coding. Second, we performed a search with BLASTP suite using the complete sequence of all cassettes that allowed us to confirm the absence of small peptides with high degrees of conservation. Surprisingly, in a few cases, we found that some gene-less cassettes contained truncated proteins that are still conserved intact in other genomes in the databases. The clearest example is cassette P3, that yields hits with 90% identity to a 128 aa-long protein from V. alginolyticus, but the sequence in P3 contains three premature stop codons in that frame. This BLASTP search supports our premise that cassettes selected here are gene-less and also suggests that -at least some of them- derive from gene-encoding cassettes that have degenerated. Additionally, the fact that TSSs in our results are not found at the 5′ end of cassettes suggests that these were not promoters dedicated to the expression of a gene in the cassette prior to pseudogenization. Altogether, this is surprising due to the fact that it is generally accepted that cassettes do not contain pseudogenes, and this is interpreted as proof of selective pressure acting upon them frequently enough to avoid pseudogenization (8). Because the loss of useless cassettes is especially easy in integrons (through excision reactions), it is possible that the function as promoters provides an adaptive value to these cassettes that saves them from purifying selection.

We have attempted to show the promoter activity of cassettes in their natural context by analyzing transcription rates of these cassettes in V. cholerae’s superintegron through RNA-seq. Despite technical and biological limitations, we could clearly observe the promoter activity of C2 (Figure 7A). Interestingly, the activity of C3 was possibly masked by the presence of the parD/E3 toxin-antitoxin cassette downstream, encoding a promoter in opposite orientation (Figure 7B). Thus, our RNA-seq data, and other datasets available, support that the array is not generally silent, with expression levels being very variable, and therefore, that there must be promoters along the array.

Our 5′-RACE results show that some cassettes contain a small subset of well-defined TSSs, while others exhibit multiple TSSs distributed throughout the cassette structure, including the attC site. The latter observation could be reflecting a spurious transcription phenomenon, possibly influenced by a high AT-content, rendering the DNA sequences similar to the promoters –10 elements (Supplementary figure S6) (53). Promoter presence has also been observed in AT-rich intragenic regions from horizontally acquired genes (54,55). Indeed, cassettes are known to be AT-rich compared to the rest of the genome, a common feature among mobile genetic elements (56). Inhibition of Rho did not change the expression patterns, so the question on how transcription starts and is regulated in these cassettes remains uncertain. Nevertheless, comparison with mock DNA controls show that, irrespective of the mechanism, these findings are biologically relevant and have strong implications in the working model of long SCIs. Regarding cassette F1, with antisense promoter activity, we identified two specific locations where the majority of TSSs colocalize, but not a bidirectional promoter. The presence of promoter activity in both strands, puts forward the possibility of cassettes silencing the array upstream and adds a layer of complexity to the regulatory landscape of integrons (57,58).

In summary, our results indicate that gene-less cassettes in Vibrionaceae species encode promoters, extending previous observations on Treponema ICs to canonical integrons. This has several implications: first, it opens a new functional category in ICs -‘promoter cassettes’-, expanding our understanding of cassette functions. Second, it confirms that several regions of SCIs are likely expressed in every array, enhancing the chances of finding adaptive cassettes in the array when needed, but likely increasing the cost of the structure (48). Third, these promoter cassettes can have a major impact in antibiotic resistance in different species if mobilized, a phenomenon that might have been neglected.

Data availability

The RNA-seq data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (59) and are accessible through GEO Series accession number GSE229776 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE229776).

Supplementary data

Supplementary Data are available at NAR Online.

Funding

The work in the MBA laboratory is supported by the European Research Council (ERC) through a Starting Grant [ERC grant no. 803375-KRYPTONINT]; Ministerio de Ciencia, Innovación y Universidades [BIO2017-85056-P]; Ministerio de Ciencia e Innovación [PID2020-117499RB-100]; J.A.E. is supported by the Atracción de Talento Program of the Comunidad de Madrid [2016-T1/BIO-1105 and 2020-5A/BIO-19726]; P.B. is supported by the Juan de la Cierva program [FJC 2020-043017-I]; A.H. is supported by the PhD program at UCM; F.T.R. is supported by the Portuguese Fundação para Ciência e a Tecnologia [SFRH/BD/144108/2019]; L.T.C. and A.S.M. are supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme [ERC grant no. 757440-PLASREVOLUTION]. Funding for open access charge: Starting Grant [803375-KRYPTONINT].

Conflict of interest statement. None declared.