GreA and GreB enhance Escherichia coli RNA polymerase transcription rate in a reconstituted transcription-translation system

Cell-free environments are becoming viable alternatives for implementing biological networks in synthetic biology. The reconstituted cell-free expression system (PURE) allows characterization of genetic networks under defined conditions but its applicability to native bacterial promoters and endogenous genetic networks is limited due to the poor transcription rate of Escherichia coli RNA polymerase in this minimal system. We found that addition of transcription elongation factors GreA and GreB to the PURE system increased transcription rates of E. coli RNA polymerase from sigma factor 70 promoters up to 6-fold and enhanced the performance of a genetic network. Furthermore, we reconstituted activation of natural E. coli promoters controlling flagella biosynthesis by the transcriptional activator FlhDC and sigma factor 28. Addition of GreA/GreB to the PURE system allows efficient expression from natural and synthetic E. coli promoters and characterization of their regulation in minimal and defined reaction conditions making the PURE system more broadly applicable to study genetic networks and bottom-up synthetic biology. Graphical abstract


Introduction
Diverse biological processes can be reconstituted and studied in vitro using purified proteins or lysates. This approach has facilitated fundamental discoveries in molecular biology and biochemistry, such as DNA replication 1 and translation of the genetic code 2 . The cell-free approach allows construction of systems that would be difficult or impossible to develop in vivo and to perform measurements and experiments that are difficult to conduct in cells. Apart from basic research on biological networks, applications of in vitro systems thus far have include biosensors and molecular synthesis [3][4][5] . E. coli cell-lysate transcription and translation (TX-TL) systems have become popular to engineer and study genetic networks of increasing complexity [6][7][8][9][10] . Cell-lysate-based TX-TL systems produce high protein yields and allow transcription from native E. coli promoters 11 . However, lysates have the disadvantage that they contain almost all the proteins and macromolecules present in the cytoplasm at the moment of lysis. For bottom-up synthetic biology and reconstitution studies this is not ideal because the lysate still contains many unknown components and uncontrollable variables.
An appealing transcription and translation system that is commercially available as PURExpress (PURE), is reconstituted from purified components and allows experiments under minimal and defined conditions 12 . This reconstituted system contains T7 RNA polymerase (RNAP), purified ribosomes, all necessary translation factors from E. coli, tRNAs, enzymes for tRNA aminoacylation and energy regeneration, creatine phosphate as an energy source, and nucleotides and amino acids as precursors. Various genetic networks can be implemented in this system, which to date mostly relied on single subunit phage RNA polymerases for transcription 13 .
Transcription by the multisubunit E. coli RNA polymerase (EcRNAP) has been reconstituted in the PURE system and can be implemented by either adding the purified holoenzyme to the reaction mix or by co-expressing its subunits 14,15 . While transcription rates of the EcRNAP in PURE depend on the concentrations of DNA template and EcRNAP, they are generally considerably lower than for phage polymerases. For example, we observed roughly an order of magnitude lower mRNA concentrations synthesized from a consensus sequence sigma factor 70 (σ70) promoter by EcRNAP than by phage T3 RNAP 13 under similar conditions. In vivo elongation rates of EcRNAP range between 28 and 89nt/s 16 19 . GreA resolves smaller backtracking events by cleaving 2-3 nt from the 3'end of the RNA, whereas GreB can also rescue longer backtracked complexes 20 . GreA and GreB are known to increase transcription elongation rate and stimulate promoter escape in a subset of promoters [19][20][21][22] , but GreA and GreB transcription elongation factors are not present in the PURE system.
Here we show that E. coli transcription elongation factors GreA and GreB enhance EcRNAP transcription rates in the PURE system up to 6-fold to reach the rate of T7 RNAP transcription in the system. We go on to show that an increase in transcription rates can be observed for several different synthetic and natural σ70 E. coli promoters. Furthermore, we used the enhanced system to study natural E.coli promoters involved in flagella biosynthesis and their activation by two different transcriptional activators in vitro, under defined conditions.

Results and discussion
EcRNAP can be added to the PURE system to allow transcription of DNA templates carrying E. coli promoters 14,15 but mRNA synthesis and subsequent protein production is more efficient using phage polymerases such as T7 or T3 RNAP 13 . In bacterial cells multiple proteins can increase RNAP activity, which are not present in the minimal PURE system. The transcription elongation factors GreA and GreB from E. coli increase overall transcription elongation rates and stimulate promoter escape in a subset of promoters by re-activating backtracked elongation complexes [19][20][21][22] . We added transcription elongation factors GreA and GreB to a PURE reaction containing EcRNAP with σ70 (holoenzyme) and a DNA template expressing EGFP under control of a consensus sequence E. coli σ70 promoter. We measured concentrations of full-length mRNA using fluorescent FRET probes that bind to a target region at the 3' end of the mRNA 23 (Fig. 1A) and found that in the presence of GreA and GreB mRNA concentrations increased faster and to higher concentrations than without the elongation factors ( Figure 1). The transcription rate increase mediated by GreA and GreB followed hyperbolic kinetics and plateaued at concentrations above 5 µM for both GreA and GreB. At the plateau GreA and GreB increased transcription rates about 3-fold and final EGFP protein synthesized about 2-fold for the DNA template concentration tested (Fig. 1B, C). When GreA and GreB were added in combination, we did not observe a significant synergistic effect on mRNA or protein synthesis ( Supplementary Fig. 1). We nonetheless used both proteins in an enhanced PURE (ePURE) reaction containing 10 µM of GreA and 10 µM of GreB in addition to the E. coli RNAP holoenzyme (0.2 µM EcRNAP, 1 µM σ70) for all subsequent experiments. Apart from these protein additions we did not further modify the commercial PURE system (see Materials and Methods). During a TX-TL reaction in ePURE we observed up to 10-fold higher mRNA concentrations, which also translated to an increased EGFP synthesis (Fig. 1D, E).
By testing various DNA template concentrations we observed an up to 6-fold increase of transcription rate in ePURE compared to a PURE reaction supplemented with E. coli RNAP holoenzyme alone ( Fig. 2A). Increased mRNA synthesis led to 3-fold higher final EGFP levels, and the advantage of using the ePURE reaction was strongest for lower DNA template concentrations (Fig. 2B). In the ePURE system we observed comparable mRNA and EGFP synthesis from an E. coli σ70 promoter and a T7 RNAP promoter (Fig. 2). The ePURE improvement should thus facilitate using E. coli RNAP for transcription under defined conditions. We next asked whether GreA and GreB can improve transcriptional elongation rates for genes other than EGFP. In order to test this, we measured synthesis of full-length mRNA from two additional genes: asr (309 bp), and chiP (1407 bp) which are roughly half and twice as long as the EGFP gene (717 bp), respectively. Transcription rates of both genes were increased in the ePURE system in the presence of GreA and GreB ( Supplementary Fig. 2). In the non-optimized PURE system transcription rates were below the detection limit for the longer gene and increased to measurable levels in the presence of GreA and GreB. The transcription rate of the shorter gene increased 8-fold ( Supplementary Fig. 2).
In order to determine if the ePURE system also increases the transcription rate for promoters other than the strong σ70tet promoter we characterized in Figures 1 and 2, we tested the system on 16 additional synthetic and natural σ70 promoters (Fig. 3). The panel was composed of nine constitutive promoters from the registry of standard biological parts (http://parts.igem.org), the BBa_J231xx-series, which are well-characterized in vivo and in vitro [24][25][26] , two constitutive promoters proC and proD 27 , several natural repressible promoters (promoter of the trp operon, lac UV5 promoter, the phage λ PR promoter), and three synthetic repressible promoters (P tet 28 , and the σ70 consensus sequence promoters P σ70 lac and P σ70 tet 13 ). In the ePURE system EGFP synthesis increased for 14 of the 17 promoters we characterized (Fig. 3A). Our results on the constitutive J231xx-promoters compare well with relative promoter strengths measured in lysate-based TX-TL reactions 25,26 . We found that the presence of GreA and GreB in the ePURE system enhanced expression particularly for strong promoters. These results suggest that at least for strong promoters transcriptional pausing during elongation severely limited transcription by EcRNAP in the PURE system. However, we cannot rule out effects on initiation of transcription as it has been shown that GreA and GreB can influence promoter escape differently for different promoters 22 . By increasing the range of synthesis rates that can be attained with E. coli promoters the ePURE system will be useful in implementing and characterizing genetic networks based on native E. coli promoters.
As an example of this we assembled a simple synthetic 2-gene cascade that consisted of T7 RNAP transcribing the rpoD gene to synthesize σ70, which then, together with EcRNAP core enzyme, activated EGFP expression (Fig. 3B). In order to study the effect of DNA template concentration on the performance of the network, we additionally analyzed the circuit at a low and a high DNA template concentration of 10 and 30 nM, respectively ( We analyzed eight flagellar promoters coupled to an EGFP reporter in the ePURE system and show their activation by FlhDC and σ28 in the defined TX-TL system (Fig. 4A).
We based our expectations on the EcoCyc E. coli database 36 , which contains information on experimentally known and bioinformatically predicted transcriptional activation. To synthesize our DNA templates we fused 150 to 250 bp long promoter regions that contained the annotated σ70 and σ28 promoters as well as FlhDC binding sites to identical, strong ribosomal binding sites followed by the EGFP reporter gene. To test their activation, we separately pre-synthesized the FlhDC and the σ28 activators and then added these to an ePURE reaction containing GreA, GreB, EcRNAP, and a DNA template with the respective flagellar promoters.
All eight promoters tested showed no detectable activity in the absence of FlhDC and σ28. When the reaction contained either of the two activators, we observed the expected activation pattern with widely differing promoter strengths (Fig. 4A). Both activators in combination generally did not improve expression compared to only one activator. Most of the time the presence of both activators even led to decreased expression. We attribute this effect to competition between both activators for binding to DNA 31 and to the RNAP core enzyme, which binds σ28 with a higher affinity than σ70 34 . We hypothesize that concentrations of EcRNAP and activators might be different in cells than in our in vitro assay, which could explain why the promoters did not show additive activation as observed in vivo 32 .
When considering activation by a single activator, two out of eight promoters, fliE and flgK, deviated from the annotated regulation pattern in EcoCyc. The fliE promoter was predicted to be activated by both σ28 37 and FlhDC 35 but we only detected low activation by FlhDC. For the flgK promoter only activation by σ28 was annotated on EcoCyc and previously shown 34 .
We observed strong activation by σ28 but also a low but significant activation by Using the genes for FlhDC and σ28 and two promoters that showed activation by both activators, we built a synthetic gene network (Fig. 4B), which was implemented and characterized in a microfluidic nano-reactor device 13  GreB proteins should furthermore be useful to increase protein yields in the PURE system when it is desirable to use E. coli promoters.

DNA template preparation
Linear DNA templates were produced by two-step PCR as described previously 13,23 using the primers listed in Table S1. Flagella promoters were PCR amplified from E. coli BL21(DE3) genomic DNA and replaced the 5'extension primer during two-step PCR. All linear DNA templates prepared for this study are listed in Table S2.

Preparation of GreA, GreB and EcRNAP holoenzyme
EcRNAP subunits (expression plasmid pVS10) were co-expressed in E. coli Xjb(DE3) cells (Zymo Research, Irvine, CA, USA) and the α 2 ββ´ω assembly (β´ subunit contained Cterminal His 6 -tag) was purified by a combination of immobilized-metal affinity, heparin and anion exchange chromatographies as described 42 . E. coli σ70 protein containing N-terminal Platereader batch TX-TL reactions, and measurement of the mRNA concentration were performed as previously described 23 . The initial mRNA synthesis, or transcription rate (TX), was determined by fitting the mRNA concentration (m) of the first 40 min of the reaction to: !"# * (1 + !!"# * ! ), where t is time and deg signifies the mRNA degradation rate (fixed at 0.0085 min -1 ) 23 . The final EGFP concentration was determined at the plateau of the protein synthesis reaction.
To test activation of flagellar promoters by FlhDC and σ28, we pre-synthesized the activators from T7 RNAP templates in a standard PURE reaction without EcRNAP and Gre proteins.
FlhDC was produced by combining flhD and flhC templates at 10nM each, and σ28 was produced from 10 nM fliA template for 100 min at 37ºC. The activators then were stored in aliquots at -80ºC until use. These were prepared by combining the FlhDC and σ28 presynthesis reactions 1:1 for testing activation by both activators and by diluting 1:1 with Tris buffer for testing of each activator separately.

Flagellar gene network in a nano-reactor device
We assembled the genetic network from 5 individual DNA templates. Final concentrations were 1 nM for P T7 -flhD, 2.5 nM for P flgK -Citrine, and P T7 -flhC, and 2 nM for P fliA -fliA and P fliA -Cerulean. The templates P fliA -fliA and P fliA -Cerulean were only added transiently. The microfluidic chip was prepared and used as described 13

Supporting Information
Supporting figures S1 and S1. Supporting tables S1 and S2.

Author Information
Corresponding author: E-mail Sebastian.maerkl@epfl.ch Author contributions: LLM and HN contributed equally to this work