Synthetic gene regulatory networks in the opportunistic human pathogen Streptococcus pneumoniae

Streptococcus pneumoniae can cause disease in various human tissues and organs, including the ear, the brain, the blood and the lung, and thus in highly diverse and dynamic environments. It is challenging to study how pneumococci control virulence factor expression, because cues of natural environments and the presence of an immune system are difficult to simulate in vitro. Here, we apply synthetic biology methods to reverse-engineer gene expression control in S. pneumoniae. A selection platform is described that allows for straightforward identification of transcriptional regulatory elements out of combinatorial libraries. We present TetR- and LacI-regulated promoters that show expression ranges of four orders of magnitude. Based on these promoters, regulatory networks of higher complexity are assembled, such as logic AND and IMPLY gates. Finally, we demonstrate single-copy genome-integrated toggle switches that give rise to bimodal population distributions. The tools described here can be used to mimic complex expression patterns, such as the ones found for pneumococcal virulence factors, paving the way for in vivo investigations of the importance of gene expression control on the pathogenicity of S. pneumoniae.


Introduction
Human pathogens and commensals reside in highly dynamic environments where they interact 36 with host tissue, the immune system and niche competitors. Streptococcus pneumoniae 37 (pneumococcus) is a prominent example of a colonizer of such complex habitats. 38 Pneumococcus is generally found in a commensal state in the human nasopharynx; however, 39 pneumococci can also cause disease, such as otitis media, meningitis, sepsis and pneumonia, networks in model organisms such as Escherichia coli and Bacillus subtilis 13-18 , but for many 49 human pathogens these tools mostly still need to be developed. 50 Here, we apply synthetic biology approaches to engineer novel gene regulatory  To date, and to our knowledge, there is only one counterselection system described for 82 S. pneumoniae called Janus 25 . Janus cloning relies on a streptomycin-resistant strain that TET, tetracycline; PCPA, para-chlorophenylalanine. (b) Determination of the maximum concentration allowing for outgrowth (starting from OD595 0.002, reaching OD595 0.2 or higher, within 10 h for antibiotics, and within 5 h for PCPA) of strains harboring selection and counterselection marker constructs, in dependency of Zn 2+ induction; cultures were tested in duplicate with concentration series that doubled the applied dosage in each consecutive step (one order of magnitude was split into three steps of similar size: 1, 2, 5, 10); in the case of PCPA the upper limit of 2 mg ml -1 was the highest concentration tested because of solubility limitations. (c, d) Plate reader assay sets in duplicate measuring cell density (OD595) of S. pneumoniae D-PEP7PZ1 (see scheme; luc, luciferase; gfp, green fluorescent protein) growing at different induction levels of PZ1, in the presence of 0.1 µg ml -1 ERY (c), or in the presence of 2 mg ml -1 PCPA (d).
6 becomes susceptible when the wild type allele of the ribosomal gene rpsL is expressed. Janus 84 was furthermore extended by adding the B. subtilis gene sacB (Sweet Janus) 26 which confers 85 sucrose sensitivity via an unknown mechanism 27 ; however, counterselection with sacB was 86 never shown to work independently of the Janus context in S. pneumoniae 26 . The disadvantage 87 of Janus cloning is the requirement of a genetic background that carries a mutated rpsL allele.

88
The pheS counterselection system of E. coli does not rely on a mutated genetic background 28 .

89
Expression of the mutated phenylalanine-tRNA ligase PheS A294G allows for the incorporation  After the identification of a suitable selection system (erm/erythromycin) and a clear Zn 2+ -dependent growth profile was observed. In contrast, when D-PEP7PZ1 was grown 106 in presence of 2 mg ml −1 PCPA, the dose-response relationship of Zn 2+ induction was inverse; 107 the absence of Zn 2+ allowed for uninhibited growth while high Zn 2+ levels resulted in growth 108 7 arrest (Fig. 1d). However, after a lag period of approximately 6 h, Zn 2+ -induced cells restarted 109 to grow (Fig. 1d) suggesting the rapid emergence of mutants.    TetR-and LacI-regulated promoters 159 The above-mentioned results showed that our cloning vector and our selection platform could 160 be successfully applied to identify constitutive promoters of desired strength. Next, we sought   Within the PT promoter series, a single tetO site (tetO1) 41 was placed either into the core 195 region (PT1-1) or into the proximal region (PT4-1) of P2, which gave rise to similar results, 196 with expression values for the induced and the repressed state within approximately three orders 197 of magnitude (Fig. 3a, b). For PT5-3, two operator sites were placed both into the core and into demonstrating that promoter leakiness in the repressed state can be reduced by decreasing the 206 overall expression strength (Fig. 3b). PT8-6 showed even lower expression in the induced state 207 as compared to PT8-2; luminescence of the repressed state, however, did not decrease any 208 further because the lower detection limit for luciferase expression was reached (Fig. 3b). The 209 expression curve of an induction series of PT8-2, as compared to PT4-1, was found to be similar 210 but downshifted. Interestingly, PT5-3, harboring two tetO sites, showed a more hypersensitive 211 dose-response relationship for ATc induction as compared to PT4-1 and PT8-2 that harbor only 212 one tetO site (Fig. 3d). For the PL promoter series, the positioning of lacO sites (lacOsym) 42 within the core 216 region (PL1-2) resulted in weak expression in the induced state, presumably because of the 217 introduction of a −35 sequence variation (Fig. 3a, c). In contrast, positioning in the proximal 218 region (PL14-2) resulted in a strong expression of 7.1×10 6 RLU OD −1 when adding IPTG.

219
Luminescence of repressed cultures was not completely suppressed and gave rise to a 220 measurement of 8.1×10 3 RLU OD −1 (Fig. 3c). LacI repressors are known to be able to shifting the second operator four base pairs further upstream, we could observe an 8-fold 228 reduction of luminescence in the repressed state as compared to PL14-2; unfortunately, this also 229 led to a 2-fold reduction in the induced state (Fig. 3c). Serendipitously, another clone with a 230 double lacO site was isolated, called PL8-2, which differs from PL8-1 by two spontaneous 231 insertions, one additional nucleotide in the distal operator site and one additional nucleotide in 232 the proximal region (Fig. 3a). PL8-2 showed the largest induction range of the PL promoter 233 series, spanning four orders of magnitude, with an expression strength of 9.5×10 6 RLU OD −1 234 in the induced state and 1.2×10 3 RLU OD −1 in the repressed state (Fig. 3c). PL8-2 also showed 235 the strongest hypersensitive response towards IPTG induction (Fig. 3e). response relationship was found within the induction series of each strain (Fig. 3f, g). Together, 241 these results suggest that Ptet and Plac display a dynamic induction range of four orders of 242 magnitude, making them the best controllable promoters currently available for S. pneumoniae. 243 Indeed, our laboratory has successfully used Plac in several studies to accurately express  both ATc and IPTG. To test this, the synthetic promoter Ptela was constructed that contains a 263 tetO site within the core region and a lacO site within the proximal region (Fig. 4a). Strain D-

264
LT-PEP9Ptela, which drives luc from Ptela and expresses both LacI and TetR from the prsA 265 16 locus, was found to require both ATc and IPTG to highly express luciferase (Fig. 4c, d, e). LacI 266 repression on its own (with ATc but without IPTG) was not enough to completely shut down 267 Ptela activity (Fig. 4e). However, the absence of ATc, and thus TetR repression alone, was 268 enough to decrease luminescence below the detection limit even in the presence of 1 mM of 269 IPTG (Fig. 4e). The results above show that both TetR and LacI can be functional within the same cell.  (Fig. 4b). Ptet induction 282 in this context was found to closely match the values obtained with PT5-3, without any 283 observable interference from the additionally present LacI (Fig. 4c). Plac expression, in 284 contrast, deviated from the corresponding PL8-2 pattern, with a 2-fold decreased maximum 285 luminescence in the presence of high concentrations of IPTG (Fig. 4d). Weaker luminescence 286 signals from Plac in the double-inducible system, as compared to PL8-2, could originate from 287 a decreased promoter activity caused by the inverse reading orientation (into the direction of DNA replication) or from the sequence deviation in the proximal region (BclI instead of the 289 BglII site). Alternatively, the translation efficiency might be decreased because of the alteration 290 within the 5' UTR. Nevertheless, the double-inducible system showed to work without 291 interference between the two regulators.

292
With the TetR-and LacI-regulated systems in place, we wondered if we could construct 293 a system where the expression of one repressor is controlled by the activity of the other 294 repressor, giving rise to IMPLY gates (Fig. 4f, g). To do so, we controlled the amount of LacI Toggle switch 1 (TS1) harbors the promoters Ptet (identical to PT5-3; with restriction 314 sites XbaI upstream and AseI downstream) driving lacI, and Plac (identical to PL8-2; with 315 restriction sites XbaI upstream and BglII downstream) driving tetR (Fig. 5a). Reporter genes 316 were integrated into the S. pneumoniae D39V genome at the amiF locus. Based on pPEPdi, 317 pPEP10 was created, with the erythromycin resistance marker erm, luc and gfp driven from 318 Ptet, and the kanamycin resistance marker aphA driven from Plac (Fig. 5a). Separating the 319 toggle switch from reporter genes increased the robustness of the system, and it furthermore 320 allowed for straightforward replacement of reporter genes. Toggle switch strains were 321 triggered, either with IPTG or with ATc, by plating and overnight incubation, followed by 8 h 322 cultivation in liquid medium in the presence of inducer (to allow for the establishment of stable 323 expression equilibria; Fig. 5b). Next, induced cultures were re-plated and re-grown for 8 h in 324 liquid medium without inducer to allow for the settlement of gene expression at stable states, 325 and for switching events to occur (Fig. 5b).    2). Transitions from the L-state to the T-state were found to occur even less frequently in TS3 368 cultures, with only ~ 0.01 % of cells showing kanamycin resistance (Fig. 5d). However, in TS3 369 strains, also cells of cultures that were pretreated with IPTG were found to be able to switch, in containing a transcriptional toggle switch at the prsA locus (TS) and genes of interest at the amiF locus (PEP10); gen R , gentamicin resistance marker; spt R , spectinomycin resistance marker; aphA, kanamycin resistance marker; erm, erythromycin resistance marker; grey circles indicate transcription terminators. (b) Work flow of D-TS-PEP10 induction and the subsequent identification of switching events; IPTG, 1000 µM; ATc, 100 ng ml -1 ; GEN, gentamicin, 20 µg ml -1 ; SPT, spectinomycin, 100 µg ml -1 ; KAN, kanamycin, 500 µg ml -1 ; ERY, erythromycin, 1 µg ml -1 ; KAN+ERY, kanamycin 500 µg ml -1 and erythromycin 1 µg ml -1 ; ABX, antibiotics. (c) Overlay of the fit curves corresponding to TetR-dependent Ptet expression (light blue) and LacI-dependent Plac3 expression (magenta) to indicate stable states (circles) and the threshold (circle with asterisk) of the toggle switch. this case from the T-state to the L-state, with a frequency of ~ 0.1 % of cells within the observed 371 time period (Fig. 5d). 372 Remarkably, an additional prediction that was made based on the plot in Fig. 5c