1 Synthetic Riboswitches that Induce Gene Expression in Diverse 2 Bacterial Species
4 Shana Topp, Colleen M.K. Reynoso, Jessica C. Seeliger, Ian S. Goldlust, Shawn K. Desai,
Dorothée Murat, Aimee Shen, Aaron W. Puri, Arash Komeili, Carolyn R. Bertozzi, June R. Scott, 5 * 6 & Justin P. Gallivan 7
8 MATERIALS AND METHODS
9 Materials. Synthetic oligonucleotides were purchased from Integrated DNA Technologies 10 (Coralville, IA). Culture media was obtained from EMD Bioscience. Theophylline and o-
11 nitrophenyl-β-D-galactopyranoside (ONPG) were purchased from Sigma. Kanamycin was 12 purchased from Fisher Scientific. X-gal was purchased from US Biological. DNA polymerase 13 and restriction enzymes were purchased from New England BioLabs. Plasmid manipulations 14 were performed using E. coli MDS42 cells (Scarab Genomics) that were transformed by
15 electroporation. Purifications of plasmid DNA, PCR products, and enzymatic digestions were 16 performed by using kits from Qiagen. All new plasmids were verified by DNA sequencing 17 performed by MWG Biotech or Elim Biopharmaceuticals. Strains and plasmids are listed in 18 Table S2.
19 Riboswitch development. We first tested whether the most active riboswitch from our 20 previously reported FACS-based screen (14) could be transferred directly to 21 Acinetobacter baylyi. We subcloned this riboswitch into pBAV1K (3) and characterized its 22 performance in A. baylyi and E. coli at various theophylline concentrations. While the switch 23 provides ligand-dependent gene expression in A. baylyi, its dynamic range and activation ratio
24 are diminished (Riboswitch D in Figs. S5 and S6), suggesting that a synthetic riboswitch that has
1 been optimized for small-molecule dependent gene expression in E. coli may not enable optimal
2 function in even a closely related species.
3 To identify synthetic riboswitches that have more desirable properties in A. baylyi, we adapted a
4 previously reported high-throughput screen for use with A. baylyi cells (8, 13). Using this screen,
5 we identified a purine-rich riboswitch sequence that activates gene expression 23-fold for
A. baylyi cells induced with 2 mM theophylline (Fig. S1 and Riboswitch A in Fig. S9). The 6
7 switch also provides 30-fold induction for E. coli cells grown in 2 mM theophylline (Riboswitch
8 A in Fig. S5), suggesting that riboswitches identified in other species may be more portable than 9 those identified in E. coli. To test this idea, we assessed the performance of each riboswitch in 10 the archetypal Gram-positive species, Bacillus subtilis. While the riboswitch identified by FACS
11 in E. coli activates gene expression only 2-fold in B. subtilis (Riboswitch D in Fig. S9), the
12 purine-rich riboswitch identified in A. baylyi activates gene expression greater than 20-fold in 13 B. subtilis (Riboswitch A in Fig. S9).
14 Encouraged by the high portability of the purine-rich riboswitch from our A. baylyi screen,
15 we sought to isolate riboswitches directly in B. subtilis by combining semi-rational design with a
16 low-throughput screen (21). We envisioned that riboswitches identified in this manner would 17 also function in a broader range of bacteria than those identified in E. coli. Guided by data from
18 our previous high-throughput screens in E. coli (13, 14, 22), we designed a 256-member library
19 and isolated Riboswitch B (Fig. S2), which activated gene expression greater than 40-fold in 20 E. coli, A. baylyi, and B. subtilis (Figs. S6, S5, and S9). These data further supported our 21 hypothesis that species other than E. coli may be used to develop highly portable synthetic
1 With this in mind, we wondered if we could introduce rational changes to riboswitches identified 2 in E. coli to enable them to function in other species. Toward this end, we redesigned our
FACS-derived E. coli riboswitch so that it could be ported without further modification to a 3
4 variety of Gram-positive bacteria, which often require stronger binding interactions between the 5 16S rRNA and the Shine-Dalgarno region (Riboswitch E, Fig. S4). Similarly, we applied 6 rational engineering to expand the host cell range of a second riboswitch that we previously 7 identified using an E. coli-based screen (21) (Riboswitch C, Fig. S3). We envisioned that these 8 engineered switches could be combined with Riboswitches A, B, and D (isolated in A. baylyi,
9 B. subtilis, and E. coli, respectively) to create a discrete set of five constructs that could be 10 screened in a novel species to identify the optimal riboswitch for a specific application. 11 E. coli manipulations. Electrocompetent E. coli cells (strain MDS42 (19), Scarab Genomics)
12 were prepared and transformed with the pBAV1K riboswitch constructs by electroporation. All 13 ;-galactosidase assays were performed by the method of Miller (17), using cells that were grown 14 with shaking at 37 ？C, in Luria Broth (LB) containing kanamycin (50 ；g/mL) and 0 or 2 mM
16 A. baylyi manipulations. To transform A. baylyi strain ADP1, ATCC 33305 (10, 23) (gift from
17 Dr. Ichiro Matsumura, Emory University) with the pBAV1K constructs, cells were grown 18 overnight in 5 mL LB at 30 ？C with shaking (250 rpm). In the morning, 3 mL fresh LB was
19 inoculated with 200 ；L from the overnight culture. These cells were grown at 30 ？C for 90 min,
20 at which time the culture was divided into 300 ；L aliquots. 3 ；L plasmid DNA was added to
21 each tube of cells. These cultures were grown with shaking at 30 ？C. After 3 h, 200 ；L of these
22 cultures was plated on selective media. ;-galactosidase assays were performed by the method of
1 Miller (17), using cells that were grown with shaking at 37 ？C, in LB containing kanamycin
2 (50 ；g/mL) and 0 or 2 mM theophylline.
3 B. subtilis manipulations. B. subtilis strain JH642 (2) (gift from Dr. Charles Moran, Emory
4 University) was grown at 37 ?C in LB with shaking (250 rpm). Competent B. subtilis cells were
5 prepared and transformed using the Spizizen method (6). Transformants were selected on LB 6 with 20 ；g/mL kanamycin. To assay for ;-galactosidase activity, strains harboring the
7 riboswitches were grown overnight in LB media with 20 ；g/mL kanamycin. In the morning,
8 cells were diluted 1:100 into fresh selective media containing 1 mM IPTG and either 0 or 2 mM 9 theophylline. ;-galactosidase assays were performed by the method of Miller (17) using 10 permeabilization with toluene.
11 A. baumannii manipulations. Acinetobacter baumannii ATCC 19606 (1) (gift from Dr. Eric
12 Skaar, Vanderbilt University) was grown at 37 ?C in LB with shaking. Electrocompetent cells 13 were prepared by pelleting mid-log phase bacteria at 5000 g for 5 min, washing pellets twice
14 with cold 2 mM CaCl, twice with 10% glycerol, and resuspending the cells in 10% glycerol. 2
15 The electrocompetent A. baumanii cells were transformed using a BioRad electroporator in 0.1 16 cm cuvettes, setting 1.8 kV. The cells were recovered in LB at 37 ?C and selected on LB plates 17 with 30 ；g/mL kanamycin. A. baumannii harboring pBAV1K plasmids were grown at 37 ?C in
18 LB-kanamycin (30 µg/mL) with shaking. All ;-galactosidase assays were performed by the
19 method of Miller (17), using cells that were grown with shaking at 37 ？C, in the presence of 0 or
20 2 mM theophylline.
21 A. tumefaciens manipulations. Competent A. tumefaciens cells (strain C58 (9), gift from Dr.
22 David Lynn, Emory University) were prepared and transformed by the method of Cangelosi et al.
1 (4). Cells were grown in 5 mL LB at 28 ？C with shaking (250 rpm). After 18 h, 10-100 ；L of
2 saturated culture was added to fresh LB (50-150 mL) and allowed to grow until the OD = 0.4. 600
3 The cells were then pelleted three times by centrifugation at 5000 g, washing twice with water
4 and once with 10% glycerol. 1 ；L of plasmid DNA was mixed with 50 ；L of cells, which were
transformed by electroporation at 1800 V (Eppendorf Electroporater 2510; Westbury, NY). 5
6 Electroporated cells were permitted to recover in 500 ；L SOB for 4 h 28 ？C. Finally, 10-100 ；L
of culture was plated on selective media. All ;-galactosidase assays were performed by the 7
8 method of Miller (17), using cells that were grown with shaking at 28 ？C, in the presence of 0 or
9 2 mM theophylline.
210 M. smegmatis manipulations. M. smegmatis mc155 (20) was grown at 37 ?C in 7H9 liquid
11 media or on 7H11 agar (Difco) containing 0.5% glycerol, 0.5% glucose, 0.05% Tween 80 and 20 12 μg/mL kanamycin unless otherwise noted. Plasmids were electroporated into M. smegmatis and
13 plated on selective media for 3 days. Transformants were grown overnight in 7H9 to an optical 14 cell density in 1 cm at 600 nm (OD) of 1-2. Cells were exchanged into fresh 7H9 containing 0 600
15 or 2 mM theophylline to OD 0.3 in 1 mL per media condition in 24-well plates and incubated 600
16 for 6 hrs (2 doubling times) at 215 rpm. The OD was measured prior to resuspending the 600
17 pelleted cells in 200 μL PBS containing 0.05% Tween 80. Fluorescence was measured in a plate 18 reader at 510 nm with 450 nm excitation and 495 nm cutoff (SpectraMax Gemini XPS, 19 Molecular Devices) and normalized to OD. Cells transformed with the plasmid pMV261 were 600
20 used as control for background fluorescence. Each growth condition was performed in triplicate. 21 S. pyogenes manipulations. GAS strain JRS1278 (T.C. Barnett and J.R. Scott, unpublished data) 22 was transformed with the riboswitch plasmids using a previously reported method (5). JRS1278
1 is ΔcovR::cat in MGAS315 generated using pJRS1349, as previously described (7). To assay for
2 ;-glucuronidase activity, riboswitch-harboring strains were grown overnight in Todd-Hewitt 3 yeast extract broth (THY broth) containing 100 μg/mL spectinomycin. This overnight culture
4 was then divided into 1.25 mL aliquots into conical flasks containing 23.75 mL pre?warmed 5 THY broth containing 0 or 2 mM theophylline. These dilutions were divided equally into
three6 15-mL conical flasks (8 mL each), and were grown in a 37 ?C water bath without agitation 7 until 2 hours into stationary phase, as determined by following the optical density. Cultures were 8 chilled on ice for 10 min and were then pelleted at 3250 g for 10 min at 4 ºC. The supernatant
9 was discarded and the pellets were resuspended in remaining supernatant and transferred to 10 1.5 mL Eppendorf tubes. The resuspended cells were pelleted, the supernatant was removed, and 11 the dry pellet was stored at 4? C. For the assay, cell pellets were resuspended in 1 mL of ice-cold 12 Z buffer and lysed in tubes with a glass bead matrix via vortexing at maximum speed for 30 13 minutes. Cell debris was pelleted and cell lysates were transferred to fresh 1.5 mL tubes.
14 Lysates were assayed by addition of 4 mg/ml solutionof p-nitrophenyl β-D-glucuronide in Z
15 buffer and the ODkinetic curve was recorded. Protein concentrations of lysates were 420
16 determined using a BCA protein assay kit (Thermo). Gus activity was determined by dividing 17 the OD by protein concentration (in μg/mL), multiplying by 1000 and dividing by the time in 420
19 M. magneticum manipulations. All riboswitches for M. magneticum were cloned within the 5？-
20 UTR of pAK22 (11), which features the Ptac promoter and expresses a MamK-GFP translational
21 fusion. Cells of M. magneticum strain AMB-1 (16) were transformed with the constructs A-E by 22 conjugation as previously described (18). The strains were grown at 30 ºC in MG growth media 23 in the presence of kanamycin (10 ；g/mL) in a chamber with the oxygen concentration
1 maintained below 10%. Starting from a culture that had been grown 24 h and had reached 2 exponential phase (OD = 0.1), two 10 mL sub-cultures were inoculated at an initial OD of 400400
0.05, in MG media in the absence or presence of 1 mM theophylline. The cells were grown in 3
4 micro-aerophillic conditions and 100 ；L of cells were spun down at different time points after
5 inoculation (50 minutes to 24 hours), spotted on agarose pads prepared with 1% agarose in
growth media, and imaged under phase contrast and fluorescence microscopy as described (18). 6
7 All fluorescent images were exposed for six seconds and the cells were visualized with the 100 X 8 objective.
9 Guide for adapting riboswitch constructs for use in new bacterial species. We anticipate that
10 in many bacterial species, Riboswitches A-E may be characterized in the context of the broad-11 host range vector, pBAV1K, using the T5 promoter and the lacZ reporter gene. (Five of the eight
12 species described in this study were transformed and assayed for ;-galactosidase activity using
13 these unmodified plasmids.) However, the true utility of these genetic control elements is that 14 they function not only in a broad range of bacterial species, but that they can also be used in 15 concert with a variety of plasmids, promoters, and reporter genes. These riboswitches may also 16 be inserted at a chromosomal locus of interest. Here
17 we describe technical considerations for constructing
18 Riboswitches A-E in the context of a different
19 plasmid, promoter, or reporter gene.
20 Riboswitches A-E were cloned into the broad host
21 range vector pBAV1K to enable modular subcloning
22 of various promoter, riboswitch, and reporter gene
1 sequences. The plasmid map shown at right, and the diagram shown below, highlight several 2 unique restriction sites that may be useful to modify various features. The promoter and constant 3 5？-UTR sequence (described in the text of Table S1) is positioned between XbaI and KpnI sites;
4 the riboswitch sequence (Table S1) is positioned between the KpnI site and the start codon; and
5 the stop codon of the reporter gene (IS10-lacZ, in this case) precedes the SpeI recognition site.
6 An intrinsic transcriptional termination sequence is positioned 3？ to the ApaI restriction site, and
7 pBAV1K also features transcriptional terminators in either direction flanking the multiple 8 cloning sites to prevent undesired transcription from promoters elsewhere on the plasmid through 9 the mRNA of interest.
11 With these details in mind, we now present specific recommendations for changing the plasmid, 12 promoter, or gene of interest to test Riboswitches A-E in new bacterial species. 13 1) Plasmid – Riboswitches can be inserted at a chromosomal locus or can be cloned onto 14 any plasmid that can be replicated and maintained in the species of interest. We 15 recommend that transcriptional terminators be placed 5？ and 3？ to the expression
16 cassette, to prevent transcriptional read-through initiated by promoters elsewhere on 17 the plasmid or chromosome.
18 2) Promoter – The riboswitches described here act at the translational, not the 19 transcriptional level. Thus, the choice of promoter should be flexible (4 different 20 promoters are used in this work). In cases where the promoter has been well 21 characterized, we recommend removing long leader sequences and regulatory
1 elements that are not essential for a given study. We typically position the first 15-30 2 bases that are transcribed by the native promoter before the constant sequence
(ATACGACTCACTATA), as indicated in the figure above and in Table S1. We 3
4 applied this approach to construct the plasmids reported in this study for use in 5 Mycobacteria smegmatis. In cases where the promoter sequence is poorly
6 characterized, contains important regulatory elements, or requires a long leader 7 sequence, we suggest using the entire native 5？-UTR up until the native RBS, which
8 must be removed to enable the riboswitches to function properly. The constant 9 sequence (ATACGACTCACTATA), riboswitch sequence, and start codon should 10 then be positioned as indicated in the figure above and in Table S1. We applied this 11 approach to construct the plasmids reported in this study for use in Streptococcus
13 3) Gene of Interest – Because the regulatory elements are located entirely in the 5?-UTR, 14 these can likely be used to regulate the expression of most genes (we reported 4 15 examples here: lacZ, gus, GFP, and mamK-GFP), though it is formally possible that
16 certain sequences in the coding region may interfere with the riboswitch; such 17 sequences can be modified using silent mutations. It is critical to position the 18 translational start codon as shown in Table S1, and to use the indicated RBS and 19 flanking sequences.
20 To achieve optimal expression levels for a given application, we recommend that all 5 switches 21 be subcloned and tested in parallel. If this is not possible, we suggest testing a subset of switches 22 that function well in related species. For example, Riboswitches E and E* (Table S1) are likely 23 to be good choices for applications that require high levels of gene expression in Gram-positive
1 species. In all cases, we recommend considering the optimal identity of the Shine-Dalgarno 2 sequence, as well as its distance from the start codon.
3 We have previously observed that genes that are controlled by a synthetic riboswitch often 4 display lower levels of protein expression than similar sequences lacking the riboswitch (12). 5 Therefore, to regulate gene expression at a chromosomal locus using a native promoter, we
recommend testing each of the 5 riboswitches. Because the riboswitch series provides access to 6
7 a wide dynamic range of gene expression, and each riboswitch features a dose-dependent 8 expression profile, we believe that previously inaccessible studies in species lacking inducible 9 expression systems can now be performed.