Edwin

Abstract

Characterization of gene expression is a central tenet of the synthetic biology design cycle. Sometimes it requires high-throughput approaches that allow quantification of the gene expression of different elements in diverse conditions. Recently, several large-scale studies have highlighted the importance of posttranscriptional regulation mechanisms and their impact on correlations between mRNA and protein abundance. Here, we introduce Edwin, a robotic workstation that enables the automated propagation of microbial cells and the dynamic characterization of gene expression. We developed an automated procedure that integrates customized RNA extraction and analysis into the typical high-throughput characterization of reporter gene expression. To test the system, we engineered Escherichia coli strains carrying different promoter/gfp fusions. We validated Edwin’s abilities: (1) preparation of custom cultures of microbial cells and (2) dynamic quantification of fluorescence signal and bacterial growth and simultaneous RNA extraction and analysis at different time points. We confirmed that RNA obtained during this automated process was suitable for use in qPCR analysis. Our results established that Edwin is a powerful platform for the automated analysis of microbial gene expression at the protein and RNA level. This platform could be used in a high-throughput manner to characterize not only natural regulatory elements but also synthetic ones.

Keywords

synthetic biology high-throughput robotics gene regulation laboratory automation

Background

Gene expression quantification has become an essential step to understand a multitude of biological processes.^1,2 Accurate characterization often involves high-throughput approaches, which enable the rapid analysis of different regulatory elements. In fact, gene expression profiling coupled with high-throughput methods has been used not only in research, but also for industrial³ and diagnostic purposes.⁴ Usually, expression analysis experiments are based on techniques to measure mRNA or protein abundance at population^5,6 or, more recently, at the single cell level.^7,8 These techniques can be used to measure global changes at the transcriptional (microarrays, RNA-seq) or protein (proteomic approaches based on mass spectrometry) level.^9,10 Although these approaches are good for screening purposes, the precise characterization of regulatory elements usually requires further verification using more sensitive and quantitative assays.¹¹ Other tactics can be used to measure the expression of specific elements, such as those based on the PCR, termed quantitative PCR (qPCR), or specific promoter/reporter gene fusions.^12,13 These procedures have been utilized to understand the performance of specific bacterial genes in different conditions^14,15 and characterize the strength of different promoters using them individually¹⁶ or in combination.¹⁷ Recently, several large-scale studies highlighted the importance of posttranscriptional regulation mechanisms and their impact on the correlation between mRNA and protein abundance.^8,18 Integrated data, corresponding to different levels of regulation, would be necessary to completely understand the performance of each regulatory unit at a defined condition. Consequently, technologies that can couple gene expression analysis at these two different levels could facilitate the characterization of functional regulatory elements. Besides, as gene expression constantly changes in order to adapt to environmental and intracellular perturbations,¹⁹ its quantification across a dynamic range becomes indispensable. In the case of microbial cells, gene expression is usually quantified at one or several time points of the growth curve. Therefore, automatic monitoring and/or propagation of microbial cell growth is also desirable for gene expression experiments.

In the past years, high-throughput systems that allow bacterial²⁰ or eukaryotic cell culture²¹ have been developed. For microbial cells, monitoring the growth kinetics has been greatly facilitated with the application of plate readers that use optical density (OD) values as indirect estimation of the number of bacteria at each time point.²² In some cases, these robotic platforms have been upgraded with other devices to automatically monitor the production of florescence proteins²³ or even to perform protein expression and purification.²⁴

In fact, the utilization of fluorescent proteins, or other kind of reporters to measure promoter performance,²⁵ has also contributed to the integration of plate readers in high-throughput platforms.²⁶ However, qPCR quantification of transcripts involves total RNA isolation, first-strand cDNA synthesis by reverse transcription (RT), and qPCR amplification. Each of these steps, in particular extraction of total RNA from a large number of samples, typically requires numerous pipetting steps, restricting its incorporation into automated high-throughput systems. In addition, the duration and the nature of the first step (RNA extraction) makes the simultaneous implementation of both kinds of gene expression analysis (promoter/reporter output and transcriptional analysis) into automated platforms challenging.

Here, we describe Edwin, a robotic workstation, the design of which enables the integration of both kinds of gene expression analysis with the automatic propagation and cultivation of bacterial cells. In order to achieve that, different commercial solutions used in gene expression quantification experiments were integrated with a liquid handling platform and a robotic arm that moves all the labware between the system components. All those instruments are controlled by Overlord3, a software that, in a user-friendly manner, directs the processes that are taking place in each of the commercial devices, integrating their actions in an automatic workflow. We then developed scripts and Overlord3 procedures that allow (1) automatic bacterial propagation, (2) fluorescence and OD monitoring, and (3) simultaneous customized RNA purification/RNA quality analysis at different time points of the growth curve. To validate the system, we engineered different Escherichia coli strains carrying two constitutive promoters of differing strengths and an inducible promoter controlling the expression of the green fluorescent protein coding gene (gfp). First, and to test the performance of Edwin as a robotic platform capable of propagating bacterial cell cultures in a customized approach, we prepared bacterial cultures at different target ODs. Then we validated (using the engineered E. coli reporter strains) Edwin’s ability to extract RNA, in an automatic way, during the dynamic acquisition of fluorescence and OD measurements. Finally, we tested that RNA extracted during the automated procedure described here is suitable to be used in qPCR analysis, and we compared these results with the ones obtained measuring fluorescence production. We can conclude the robotic platform presented here and procedures developed confer a powerful set of tools for automated gene expression analysis at the transcriptional level during time-lapse fluorescence quantification based on promoter and reporter gene fusions.

Methods

Bacterial Strains, Culture Conditions, and General Procedures

The strains used in this study were E. coli XL1-Blue MRF’ (Stratagene, La Jolla, CA) and E. coli DH10B (Invitrogen, Waltham, MA) carrying plasmids with different gfp/promoter fusions. E. coli were grown in Luria Bertani (LB) base media supplemented with 100 µg/mL ampicillin to ensure plasmid maintenance. All cultures were performed at 37 °C and 700 rpm in the plate reader containing 200 µL of LB per well.

Manual cultivation of bacteria cells was performed as follows: E. coli strains were cultured overnight in 2 mL of LB in 15 mL Falcons at 37 °C. Then, OD₆₀₀ was recorded in the Nanodrop 2000c spectrophotometer (Thermo Scientific, Waltham, MA) and a new black, flat, clear-bottom 96-well microplate (Greiner, Monroe, NC) was inoculated at the specified target OD₆₀₀ by manual pipetting.

PCRs were performed as follows: 50–100 ng of the template indicated in each case was mixed in a 50 μL reaction mixture with 10 pmol of each of the primers specified and 1 unit of Pfu DNA polymerase (Stratagene). Samples were then subject to an initial step of 5 min at 98 °C and then 30 cycles of 15 s at 95 °C, 15 s at 58 °C, and 2 min at 72 °C, to finish with a final extension of 10 min at 72 °C. Clones were first checked by colony PCR²⁷ using 1.25 units of Taq DNA polymerase (Roche, Basel, Switzerland) and later confirmed by DNA sequencing.

Other gene cloning techniques and standard molecular biology procedures were carried out according to Sambrook et al.²⁷

Construction of Plasmids

Plasmids carrying the gfp gene controlled by different promoters were constructed as follows: a 736 bp fragment containing the gfp coding region downstream of a strong ribosome binding site (RBS) was amplified by PCR using as a template Biobrick part BBa_E0040 and the primers B0034GFP1F (ggACTAGTgaaagaggagaaatactagATGCG TAAAGGAGAAGAACTTT) and GFP3R (gcCTGCAG CGGCCGCTTATTTGTATAGTTCATCCATGC). Primer B0034GFP1F carried a SpeI restriction site (uppercase underlined), followed by the strong RBS (Biobrick part BBa_B0034, lowercase) and primer GFP3R a PstI (uppercase underlined) restriction site. The fragment was then cloned SpeI/PstI into Biobrick part BBa_J23100 or BBa_J23114, which contain a strong constitutive promoter (J23100, ttgacggctagctcagtcctaggtacagtgctagc) or a weak constitutive promoter (J23114, tttatggctagctcagtcctaggtacaatgctagc), respectively (both promoters are in a pJ61002 backbone). This generated the plasmid pJ23100 or pJ23114, which carry the gfp gene with a strong RBS under the control of a strong or weak constitutive promoter, respectively.²⁸

To generate a plasmid carrying a promoterless version of gfp, another PCR was performed using as a template Biobrick part BBa_E0040. This time, the primers used were B0034GFP2F (ggTCTAGA gaaagaggagaaatactagATGCGTAAAGGAGAAGAACTTT) and GFP3R, which carried the XbaII and PstI restriction sites, respectively (uppercase underlined), and the RBS (lowercase in B0034GFP2F). This new 736 bp fragment was cloned XbaI/PstI into Biobrick part BBa_J23100, replacing the strong promoter present in that backbone and yielding pGFP plasmid (carrying a promoterless version of the gfp downstream of Biobrick part BBa_B0034 RBS).

Finally, a 199 bp fragment carrying the Plac promoter flanked by XbaI and SpeI restriction sites (introduced in the primers, uppercase) was amplified using Plac1F (ggTCTAGAgcgcccaatacgcaaacc) and Plac2R (ggACTAGTtaaattgttatccgctcacaatt), using as a template pUC18. This fragment was then cloned XbaI/SpeI into pJ23100, replacing the constitutive promoter J23100 by the Plac promoter and generating pPlac plasmid.

Automated Procedures

All the automated steps performed in this study were executed using the relevant equipment within Edwin, namely, a Biomek FX^P liquid handler (Beckman Coulter, Brea, CA), FLUOstar OMEGA plate reader (BMG Labtech, Cary, NC), QIAxcel Advanced capillary electrophoresis system (QIAGEN, Hilden, Germany), and LightCycler 480 II system (Roche). The automatic process presented here involved 96-well plate inoculation with different bacterial strains, cultivation of bacterial cells and monitoring of OD and fluorescence, sample collection for RNA extraction, and RNA extraction and analysis. This process was executed using a specific procedure developed in the Overlord3 software (Peak Analysis and Automation Ltd. [PAA], Hampshire, UK). The Overlord3 procedure encompassed different steps, some of which controlled the movements of the SCARA robotic arm directing the transfer of the labware from one device to another. They also ran, in specific moments of the workflow, (1) the methods developed in Biomek FX^P software to collect and prepare liquid samples by automatic pipetting and to extract RNA samples by using the Biomek Filtration System with a Vacuum Valve Unit (Beckman Coulter), (2) the protocols developed in OMEGA software to run the OMEGA plate reader (BMG Labtech), and also, when required, (3) QIAxcel and LightCycler 480 II system methods.

Variables that define the target OD for plate inoculation, samples to be taken for RNA extraction, and columns of the Wizard SV 96 Binding Plates used in the RNA extraction protocol in each step were defined by the user using Overlord3 software. Scripts were written for Overlord3 software (Supplementary Material) in order to generate files defining pipetting patterns for Biomek FX^p software based on parameters introduced by the user.

Growth and Fluorescence Assays

Bacterial growth was estimated by measuring the optical density of the cultures at a wavelength of 600 nm (OD₆₀₀). OD₆₀₀ and florescence values were recorded in the multimode microplate reader FLUOstar OMEGA (BMG Labtech) when it was defined in the Overlord3 procedure. Bacterial cultures were propagated using 200 µL of LB per well and fluorescence (λ_ex 485 nm, λ_em 520 nm, and gain 1300), and OD₆₀₀ was measured using black, flat, clear-bottom 96-well microplates (Greiner). The relative fluorescence values were calculated by dividing the obtained values of total light emission (in arbitrary units) by the OD₆₀₀ values obtained in the same wells. All the experiments were reproduced by running the Overlord procedure on at least three independent and different occasions.

RNA Extraction

Customized RNA extraction was performed automatically using specific methods developed in Biomek FX^P software and SCARA movements, both executed by Overlord3. The procedure comprised three serial stages that started once a defined time point was reached: (1) Biomek Span-8 P250 Tips (Beckmann) and a reservoir containing the required buffers (Polypropylene Deep-Well Plates, Beckmann) were loaded into the Biomek FX^P desk. (2) A 96-well plate with bacterial cultures was taken from the BMG OMEGA plate reader by the SCARA robotic arm and transferred to the Biomek FX^P desk, where 100 µL of the user-defined samples was mixed with 100 µL of lysis buffer (SV 96 Total RNA Isolation System, Promega, Madison, WI) in the reservoir. The samples taken and the number of columns of the reservoir containing lysis buffer were defined by the scripts (see Supplementary Material). The plate then was transferred back to the plate reader to continue with the recording of OD₆₀₀ and fluorescence values. (3) The last stage performed in Biomek FX^P was completed as follows: The lysates were transferred to specific columns (defined by another script described in Supplementary Material) of the Wizard SV 96 Binding Plate (Promega) sitting on the Biomek Filtration System (Beckman Coulter). After a first washing with 525 µL of washing buffer (96 Total RNA Isolation System, Promega), samples were treated in columns with 1 unit of TURBO DNase (Ambion) at room temperature for 20 min. A second washing step (525 µL of washing buffer) was performed followed by another treatment with DNase at room temperature for 20 min. After the second DNAse treatment, 500 μL of DNAse Stop solution (96 Total RNA Isolation System, Promega) was added to the columns, and after that, they were washed again with 525 μL of washing buffer. Finally, RNA was eluted in a new 96-well plate (elution plate) by adding 175 μL of water to the Wizard SV 96 Binding Plate. The Wizard SV 96 Binding Plate was subjected to 50,796 Pa of vacuum pressure for 3 min after each pipetting step using a Biomek Filtration System (Beckman Coulter). After the last washing step (before the elution), vacuum was applied for 20 min in order to remove residual ethanol from the column. All the RNA extractions were reproduced by running three independent experiments, using the Overlord procedure on different occasions.

RNA Analysis

The procedure developed in Overlord3 also included serial steps for the RNA analysis. The RNA analysis was performed by running the samples in a QIAxcel RNA Quality Control Gel Cartridge (QIAGEN). This part of the Overlord3 procedure also involved four different stages that happened automatically after the last RNA extraction: (1) In the first one, specific tips, plates, and a reservoir containing RNA QIAXcel buffers (QIAxcel RNA QC Kit v2.0, QIAGEN) were placed in the Biomek FX^P desk. Then 6 µL of each of the RNA samples was taken from the elution plate and mixed with 6 µL of QIAxcel denaturation buffer (QIAGEN) in a LightCycler 480 Multiwell Plate 96-well plate (Roche). (2) The plate was then transferred by SCARA to the LightCycler 480 II system, where it was incubated for 2 min at 70 °C. (3) A LightCycler 480 Multiwell Plate 96 was then transferred back to the Biomek FX^P, and 18 µL of dilution buffer (QIAGEN) was added to the samples. (4) Finally, the plate was transferred using a SCARA robotic arm to the QIAxcel Advanced system device and a specific method to analyze the bacterial RNA was run. All the RNA analyses were reproduced by running three independent experiments, using the Overlord procedure on different occasions.

Transcription Analysis by RT-qPCR

RT-qPCR was performed as follows: The first-strand cDNA was synthesized in 20 µL reaction volumes using 2 µL of total RNA template, 4 µM of specific reverse transcription primers (5′- TTATTATTTGTATAGTTCATC-3′ for gfp and 5′-TAAGGAGGTGATCCAAC-3′ for 16S rRNA) described elsewhere,²⁹ and the ImProm-II Reverse Transcription System (Promega). The cDNA was diluted 1:80 in nuclease-free water, and 2 µL of this dilution was used as a template in a 20 μL reaction containing 1× LightCycler 480 SYBR Green I Master (Roche) and 10 µM of gfpFw (5′-ATGGCC CTGTCCTTTTACC-3′)/gfpRv (5′-ATGTGGTCTCTCTT TTCGTTG-3′) or 16SFw (5′-GCTACAATGGCGCATACA AA-3′)/16SRv (5′-TTCATGGAGTCGAGTTGCAG-3′) primers.²⁹ The reactions were run on a LightCycler 480 II System (Roche) using the following program: 5 min at 95 °C and 30 cycles of 15 s at 95 °C, 15 s at 50.5 °C, and 20 s at 72 °C, with a final melting curve of 15 s at 95 °C, 15 s at 60 °C, 10 min gradient to 95 °C, and 15 s at 95 °C. To analyze the RT-qPCR data, the 2-ΔΔCt method³⁰ was used, taking as a calibrator Ct values corresponding to strain E. coli pGFP. The RT-qPCR values correspond to the averages of at least three biological replicates performed using as a template RNA obtained running the Overlord procedure on different occasions.

Detection of gDNA Contamination in RNA Samples

PCR to check the presence of genomic DNA (gDNA) in the RNA extracted was performed as follows: 2 µL of RNA was mixed with 4 µM of specific reverse transcription primers (described above), but without reverse transcriptase, in a total volume of 20 µL. This mixture (similar to the RT reaction but containing only RNA and the reverse transcription primers) was then diluted 1:80 in nuclease-free water as the cDNA samples prepared before. Two microliters of this dilution or 2 µL of the cDNA diluted 1:80 was used as a template in PCRs using 16SFw/16SRv primers and 1.25 units of Taq DNA polymerase (Roche).

Statistical Comparison of Growth Curves

Parameters derived from growth curves were compared as follows: three individual growth curves produced in different experiments (different performances of the Overlord3 procedure) were fitted to the Gompertz equation³¹ using the DMFit program,³² and maximum specific growth rate (µ), length of lag phase (λ), and maximum cell growth (A) parameters were calculated. The significance of the difference between the averages of these parameters was assessed by using Student t test.

Results and Discussion

Requirements for Automatic Monitoring of Microbial Gene Expression Dynamics

Edwin is a robotic platform designed to measure bacterial gene expression at the population level and integrates in one single procedure two different processes that, individually, are often used in gene expression studies: (1) monitoring of the output of promoter and reporter gene fusions³³ and (2) extraction of suitable RNA for analysis of the transcription by RT-qPCR.¹² The first process has been integrated in a number of different high-throughput platforms due to its nondisruptive nature.²⁶ However, gene expression analysis of bacterial populations at the transcriptional level requires multiple steps, the first being the disruption of the cells to extract the RNA. This type of analysis requires customized collection of specific samples at different time points during bacterial growth that are usually stored for later RNA extraction, RNA quality control, and stand-alone RT-qPCR analysis. Both approaches described above require, before starting the gene expression analysis, the inoculation of new bacterial cultures with a similar number of cells.²³ The workflow to do this on an automated platform would encompass several sequential and parallel steps: (1) inoculation of different strains in a new plate at a specific OD₆₀₀ that corresponds to a defined number of cells, (2) OD and reporter gene output monitoring, (3) RNA extraction, (4) usually RNA validation, and finally, (5) RT-qPCR. The automation of all these steps would then require the integration of numerous commercial devices, each designed specifically to perform an individual process, and a central control of workflow and data management through specific software.

Hence, Edwin integrates a liquid handler with a shaking-incubating plate reader, a qPCR system, and a QIAxcel Advanced system for RNA quality control monitoring ( Fig. 1a ). We have further increased the high-throughput nature of this system by pairing all these instruments with a robotically assisted arm ( Fig. 1a , SCARA, PAA) that can transfer samples from one device to another. We also have developed procedures using the custom global governor software, Overlord3 (PAA), which is used to control all the different vendor software and to manage the collection of results. The procedure presented in this work (a schematic version of the procedure and the interplay of the commercial devices is shown in Fig. 1b ) integrates into one single run the automatic inoculation of bacterial strains, monitoring of OD and fluorescence, and personalized RNA extraction and analysis. The process can be customized through the introduction of several user-defined variables in Overlord3 before starting. The user can define the number of RNA extractions, select which samples will be collected for RNA extraction/analysis, and state the specific time points when this will happen. Finally, the RT-qPCR can be done as a stand-alone procedure.

Figure 1.

Edwin workflow and parts. (a) Hardware arrangement of Edwin: Biomek FX^P liquid handler (1), OMEGA BMG plate reader (2), QIAxcel Advanced system (3) LightCycler 480 II (4), and SCARA robotic arm in detail in left panel (5). (b) Schematic representation (not to scale) of the Overlord3 procedure and the interplay between the different devices integrated in Edwin: OMEGA BMG plate reader (green boxes), Biomek FX^P liquid handler (grey boxes), LightCycler 480 II (blue boxes), and QIAxcel Advanced system (purple boxes). Grey arrows represent the movements of labware mediated by the SCARA robotic arm. Racks, where the labware is loaded and unloaded, are represented by orange triangles. A plate with bacterial cultures is transferred to the plate reader from the racks and cultured overnight. The following day, the cultures are diluted into a new plate (dil), which is moved to the plate reader where the OD of the samples is recorded. After this, the new plate is transferred back to the BiomekFX^P and different volumes of each well are taken to inoculate a new plate with a specific target OD (dil2). Then this plate is transported back to the plate reader, where the monitoring of growth and fluorescence starts. If the user defines RNA extractions at the specific time points (1 or 2), steps for RNA extraction will start (r1, r2, r3). Finally, after finishing the RNA extractions, steps for analyzing the RNA quality will take place (a1, a2, a3, a4).

Automated Inoculation and Cultivation of Microbial Cells

A required step to study bacterial gene expression using plate readers is to prepare a new plate with the strains and conditions that are going to be monitored (assay plate). Thus, Edwin was designed to propagate bacterial cultures by specific dilution. The procedure described in this work includes different steps ( Fig. 1b ): (1) to transfer a plate with overnight cultures to the Biomek FX^P deck, (2) to prepare a new plate by diluting the overnight samples with medium in order to decrease the number of cells per milliliter (dilution plate), and (3) to record the OD₆₀₀ measurements of the dilution plate in the plate reader. Then, a bespoke script written for Overlord3 software defines a pipetting data file by calculating from OD₆₀₀ values (Script Absorbance converter, Supplementary Material) the volume to be sampled from each well to achieve a user-defined target OD upon dilution. The Biomek FX^P software can then read that pipetting data file, taking those volumes from the dilution plate to prepare a new one (assay plate) at the target OD₆₀₀ defined by the user ( Fig. 2a ), ensuring the same amount of bacteria in each well.

Figure 2.

Automated inoculation of bacterial plates. (a) Workflow used by Edwin to inoculate a new plate. It contains examples of the files produced by the plate reader software (half 96-well plate in the example) and by Overlord3 software. Overlord3 transforms the plate reader output into a Biomek FX^P readable file with volumes and positions. Cultures of E. coli prepared at different target ODs, 0.01 (b) or 0.05 (c), by manual handling or using the Edwin automated culture. Averages and standard errors of three different biological replicates are shown.

To test the ability of Edwin for custom inoculation bacterial cultures, we used an E. coli strain carrying a plasmid with an antibiotic resistance cassette (pPlac, described below). We decided to run the Edwin procedure to inoculate culture plates using two different sizes of inocula and compare those growth curves with the ones obtained by manual handling. In order to do that, we cultured the E. coli strain overnight (see below) and inoculated it into new 96-well plates using two different target OD₆₀₀ values, 0.01 ( Fig. 2b ) and 0.05 ( Fig. 2c ), by using manual handling or Edwin. Growth curves were fitted to the Gompertz equation³¹ using the DMFit program,³² and maximum specific growth rate (µ), length of lag phase (λ), and maximum cell growth (A) parameters were calculated and compared. The results showed that the µ, λ, and A generated with different target ODs by using Edwin were not significantly different from the ones obtained by manual handling (Suppl. Table 1). These results indicated that by using an automation procedure specifically designed for Edwin, it is possible to prepare assay plates with different target ODs and culture them. Besides, the obtained ODs fell within the range where the OD₆₀₀ and number of cells follow a linear correlation (Suppl. Fig. 1).

Customizing the Automated Procedure for Simultaneous Fluorescence/Growth Monitoring and RNA Purification/Analysis

Edwin has been designed to monitor the dynamics of gene expression at two levels, the protein level using the promoter/reporter gene fusions and the transcriptional level by extracting RNA and quantifying it by RT-qPCR. Automation of gene expression detection by using promoter/reporter gene fusions can be easily achieved using plate readers and/or incubators and multiwell plates ( Fig. 3 ). In general, in these experiments multiwell plates contain technical replicates of each strain/condition to ensure robust measurement of reporter output ( Fig. 3 ). By contrast, integrated qPCR quantification of gene expression during time-course experiments is more challenging, principally due to the first step of the process (total RNA extraction of multiple samples).³⁴ Protocols/kits to extract total RNA involve numerous pipetting steps, starting with the lysis of the cells. Besides, all of them involve pelleting the cells before mixing them with the lysis buffer.³⁵ This implies that, in the case of the extraction of all the samples of the same plate, more replicates would be required to continue monitoring reporter gene expression during the rest of the time course.

Figure 3.

Schematic representation of the Edwin capabilities. Edwin is able to extract and analyze total RNA at different time points, during the typical real-time experiment, to measure reporter gene expression on a plate reader (upper part). The user can choose the wells (columns) that will be taken for RNA extraction among the ones present in the typical plate layout prepared for reporter gene detection (an example layout is sketched). In the following time point, the next column will be selected for RNA extraction. Finally, after the RNA extractions Edwin will perform the RNA analysis, running it automatically in the QIAxcel high-resolution capillary electrophoresis.

In order to overcome this limitation, and to integrate total RNA extraction during a typical time-course reporter gene expression experiment, we developed specific scripts that can be read by Overlord3 software (scripts for automatic RNA extraction, Supplementary Material). Users then can define, at the beginning of the process, which and how many columns of the assay plate (and also of the filter plate used in the RNA extraction process) they want to use for RNA extraction (exemplified in Fig. 3 ). Overlord3 generates files that organize the collection of samples for RNA extraction and the initial steps of this process. The samples present in the selected wells will be collected for RNA extraction when the time point defined by the user is reached. At that specific time, and after recording the reporter gene output, the plate will be automatically transferred to the Biomek FX^p deck. The defined samples will then be collected (e.g., a set of technical replicates) and the plate will be transferred back to the plate reader to keep on recording the reporter output while the RNA extraction takes place. In this procedure ( Fig. 1b ), the number of time points where the samples are taken for RNA extraction is limited by the duration of each extraction (approximately 1 h, 20 min). Nevertheless, it is possible to perform an automatic RNA extraction during the growth curve in, at least, two time points ( Figs. 1b and 3 ). The Overlord3 procedure encompasses steps to automatically modify the variables used for the first RNA extraction and to again execute the scripts. This allows subsequent RNA extraction at a later time point of another set of replicates of the assay plate ( Fig. 3 ) using different binding columns in an automatic way. Besides, one important advantage of the procedure we have developed is that total RNA is extracted from liquid samples. These samples are taken directly from the 96-well plate, avoiding the need for freezing and/or pelleting the cells, and therefore decreasing the complexity and duration of the automated process.³⁵

In addition to the automatic extraction of RNA, we included (in the same Overlord3 workflow) steps that allow automatic analysis of the RNA samples. This process is performed after finishing all the RNA extractions determined by the user ( Figs. 1 and 3 ) by running the samples in a QIAxcel RNA Quality Control Gel Cartridge (QIAGEN) as described in Material and Methods.

Validating Edwin’s Ability to Measure Fluorescence and OD₆₀₀ in an Automatic Procedure

The design previously mentioned allows an integrated and automatic RNA extraction and analysis during the typical experiment of fluorescence and OD measurement. In order to test this procedure, we decided to engineer different E. coli strains that we could use as a reference. These E. coli strains, named pJ23114, pJ23100, and pPlac, carry the gfp reporter gene under the control of a constitutive weak promoter (Biobrick BBa_J23114), a constitutive strong promoter (Biobrick BBa_J23100), and the IPTG inducible Plac promoter. Before performing any RNA extraction, we decided to first test the ability of Edwin to grow bacterial cultures while acquiring a fluorescence time course. We then inoculated and cultured these strains by using the described procedure, defining no RNA extractions and using LB (and also LB + 10 mM IPTG in the case of the strain carrying Plac→gfp) as a medium to propagate them.

We recorded the fluorescence and optical density values during the time course. The results are shown in Figure 4 . The constitutive promoters ( Fig. 4a ) rendered values of specific fluorescence that agreed with the strength previously described for these two promoters.³⁶ Also, in the case of the Plac inducible promoter ( Fig. 4b ), we could observe an increase of the production of the fluorescence when we cultured the strain in the presence of IPTG, an inducer of the Plac promoter. These results confirmed that Edwin is suitable not only to inoculate and culture microbes in 96-well plates but also to monitor the gene expression using reporter gene/promoter fusions in a time-course experiment. Besides, the obtained fluorescence values fell within the range where fluorescence and amount of fluorophore follow a linear correlation (Suppl. Fig. 2).

Figure 4.

Time course of GFP expression obtained using automated culture shown by E. coli carrying the gfp under the control of a constitutive strong (pJ23100 strain) and constitutive weak (pJ23114 strain) promoter growing in LB (a) or E. coli carrying Plac→gfp fusion growing in LB or LB + 10 mM IPTG (b). Averages and standard errors of at least three different biological replicates are shown. All pJ23100 values were significantly different from the pJ23114 ones (Student t test, p < 0.05). All the LB + 10 mM IPTG values were significantly different from the LB ones except time points 0, 1, and 2 (Student t test, p < 0.05).

Validating the Procedure Steps to Extract RNA by Analyzing Its Quality

As we mentioned above, we introduced in the procedure the possibility of performing customized RNA extractions of the samples that are being monitored in the plate reader at different time points. The procedure also allows the user to include an analysis of the quality of the extracted RNA by sensitive, high-resolution capillary electrophoresis in the QIAxcel Advanced system. Based on the image of the electrophoresis, the QIAxcel Advanced system produces two indicators of RNA quality assessment, the RNA Integrity Score (RIS) and the 23S/16S rRNA ratio. To validate these steps, we again ran the procedure with the strains and conditions detailed above, together with a strain carrying the promoterless version of GFP to calibrate in RT-qPCR analysis. We also included the E. coli DB10B strain carrying Plac→gfp as a positive control of Plac expression. This strain does not contain the LacI^q repressor of the Plac promoter. Therefore, in this genetic background, the expression mediated by Plac is not induced by the addition of IPTG, which relieves LacI^q repression and behaves like a constitutive promoter. We then defined RNA extractions of all these strains at two different time points of the growth curve (times 3 and 6), which represented the middle and late exponential phase. We ran the procedure, including the automatic RNA analysis, after the extraction, which was performed automatically in the QIAxcel Advanced system (a representative image is shown in Fig. 5a ). The obtained capillary electrophoresis images were used to calculate total RNA concentration and the RNA quality indicators. An average of these values for each strain at two different time points is shown in Table 1 . The results indicated that the 23S/16S rRNA ratio of most of the samples is close to 1.82 (SD ±0.16) and that the RIS index about 6.04 (SD ±0.27). These results showed that RNA quality obtained by running the Edwin automatic procedure was suitable for downstream applications.^37,38

Figure 5.

RNA quality obtained by using Edwin’s procedure. (a) Representative picture of the electrophoresis performed in QIAxcel loading RNA extracted automatically at time point 3 (t3) and time point 6 (t6) from the following strains: E. coli DH10B growing in LB carrying the plasmids pGFP (1), pJ23114 (2), and pJ23100 (3); E. coli DB10B and E. coli XL1-Blue MRF’ carrying Plac→gfp fusion (pPlac strains) and growing in LB (4 and 6) or LB + 10 mM IPTG (5 and 7). RNA size marker is represented as M. Two bands corresponding to 23S and 16S ribosomal RNA appear in each lane. (b) PCR using the 16S rRNA specific primers and using as a template RNA (lower part) or the corresponding cDNA (upper part) obtained from the samples described above (1–7) at time points 3 (t3) and 6 (t6). As a positive control, gDNA of E. coli DH10B (+) was used, and as a negative control, water (–). The DNA ladder is represented as M.

Table 1.

Quality of RNA Extracted Using Edwin.

	Strain	Total RNA Concentration (ng/µL)	Ratio 23S/16S	RIS
Time point 3	pGFP	130.64 (±48.21)	1.75 (±0.06)	5.53 (±0.60)
	pJ23114	121.73 (±40.94)	1.90 (±0.01)	6.33 (±0.75)
	pJ23100	105.43 (±49.63)	1.84 (±0.10)	5.83 (±0.21)
	DH10B pPLac	57.35 (±11.04)	1.78 (±0.10)	6.13 (±1.01)
	DH10B pPLac + 10 mM IPTG	93.55(±32.16)	1.85 (±0.10)	5.77 (±0.31)
	XL1 pPLac	57.61 (±4.72)	1.75 (±0.08)	6.33 (±0.46)
	XL1 pPLac + 10 mM IPTG	37.57 (±14.51)	1.79 (±0.06)	6.30 (±0.52)
Time point 6	pGFP	45.55 (±10.95)	1.64 (±0.23)	6.00 (±0.44)
	pJ23114	40.05 (±10.2)	1.90 (±0.08)	5.97 (±0.38)
	pJ23100	44.54 (±12.61)	1.80 (±0.04)	5.93 (±0.31)
	DH10B pPLac	36.73 (±13.44)	1.96 (±0.11)	5.90 (±0.30)
	DH10B pPLac + 10 mM IPTG	21.10 (±9.49)	2.28 (±0.77)	6.00 (±0.28)
	XL1 pPLac	37.21 (±17.64)	1.57 (±0.05)	5.93 (±0.50)
	XL1 pPLac + 10 mM IPTG	34.94 (±15.96)	1.75 (±0.05)	6.60 (±0.87)

Averages of total concentration, 23S/16S ratio, and RIS values of RNA samples obtained from different E. coli strains at time points 3 and 6 running the automatic Ewin procedure on three different occasions. Standard deviations are indicated in parentheses.

Finally, and because the presence of gDNA in RNA samples can lead to nonspecific amplification³⁹ in the RT-qPCR analysis, we decided to check that the extracted RNA was free of gDNA. In order to do that, we amplified by PCR a fragment of the 16S rRNA housekeeping gene using as a template aliquots of the RNA extracted or the corresponding cDNA obtained after the reverse transcription reaction. The results showed a 101 bp band corresponding to 16S rRNA only when the cDNA was used as a template, indicating that RNA is free of gDNA ( Fig. 5b ).

RNA Extracted by Edwin Is Suitable for RT-qPCR Analysis

Our results indicated that total RNA extracted during the automated process was not degraded; it was clean and free of gDNA. Nevertheless, in order to confirm that it was suitable for transcriptional analysis, we performed RT-qPCR. We used as a reference the values obtained for the E. coli pGFP (strain carrying the promoterless gfp) to calculate relative levels of mRNA for each promoter. The RT-qPCR results ( Fig. 6a ) for the constitutive promoters (J23114 and J23100) indicated that both promoters maintain the same rate of transcription at two different time points, with J23100 being approximately fourfold stronger than J23114. Although the rate of induction (between 3.5 and 4 times) observed at the transcriptional level is different from that obtained from monitoring the fluorescence ( Fig. 4a , about 15-fold higher), the transcriptional results confirmed the constitutive nature of both promoters and their relative strengths.

Figure 6.

Analysis of expression at the transcriptional level using RNA samples obtained by automation. RT-qPCR analysis of the gfp expression at time points 3 (t3) and 6 (t6) of E. coli strains pJ23114 and pJ23100 growing in LB (a) and E. coli DH10B (b) or E. coli XL1-Blue MRF’ (c) carrying pPlac plasmid growing in LB or LB + 10 mM IPTG. Averages and standard errors of at least three different biological replicates are shown. Student t test was used for statistical analysis.

With regard to the Plac inducible promoter, it was not possible to observe differences between noninducing (LB) and inducing conditions (LB + IPTG) in E. coli strain DH10B corresponding to the lack of the repressor LacI^q ( Fig. 6b ). The RT-qPCR analyses also showed a clear induction of Plac promoter in the presence of IPTG in the E. coli XL1-Blue MRF’ strain at time point 3, confirming the suitability of the procedure to also detect expression driven by inducible promoters ( Fig. 6c , regulated system). Nevertheless, our data showed that the levels of expression of the Plac promoter in the presence of IPTG at time point 6 were lower than at time point 3 and not significantly different from noninducing conditions, suggesting a decrease of the promoter activity. Also, in this case, the differences observed in the levels of mRNA for the inducible promoter did not correlate with the fluorescence values obtained with the plate reader ( Fig. 4b ). This discrepancy could be explained by the different degradation rates of the mRNA and the GFP, leading to an accumulation of the protein after induction of promoter. Taken together, these results indicated that the RNA automatically extracted during the process is suitable for RT-qPCR analysis, and also they support previous work that points to a reduced correlation between mRNA and protein abundance in some cases at single cell and population levels.^8,18,40

Conclusions

Edwin, the robotic platform presented here, has been designed to measure gene expression at the protein and RNA level in microbial systems. This kind of characterization has become an essential step of the design cycle workflow used in synthetic biology.²⁶ Recently high-throughput platforms have been used to characterize biological parts; however, they typically incorporate liquid handling robots coupled with plate readers,^16,17 which only allows quantification of gene expression based on the amount of reporter protein. Monitoring gene expression at the RNA level, using standard techniques, requires several process steps and raises the difficulty of integrating this kind of analysis in high-throughput platforms.³⁴ Transcriptional analysis by qPCR requires the extraction of RNA. In general, this step involves the collection of samples and several pipetting steps, thus limiting its incorporation in time-course experiments.¹⁷ Therefore, we developed Edwin’s scripts and tools to facilitate automated RNA extraction analysis within a typical high-throughput procedure for the dynamic assessment of gene expression. We tested the system with two constitutive promoters and the Plac inducible promoter. Our results confirmed that Edwin can be used to monitor gene expression using promoter/reporter fusions and, at the same time, extract and analyze the RNA at different time points for a later RT-qPCR analysis. Our results also support previous evidence that suggests a low correlation between the abundance of mRNA and protein levels.⁸ Therefore, the analysis of the expression at both levels ought to be valued not only to decipher putative posttranscriptional mechanisms in natural promoters, but also to get a deeper characterization of the regulatory regions and lead to an optimization of parameters used in systems biology. The tools presented here can be used to (1) perform the characterization of natural regulatory elements at transcriptional and translational levels in different conditions (e.g., regulated systems in the presence of the inducers and/or repressors); (2) study the relations between the mRNA and protein levels in order to investigate regulatory events acting at the posttranscriptional level, such as riboswitches or small RNA (sRNA)¹⁸; or (3) characterize new synthetic regulatory elements.^41–43

Footnotes

Acknowledgements

We thank our Flowers Consortium colleagues Dr. Geoff Baldwin and Prof. Paul Freemont (Imperial College London) for helpful discussions. We also thank Dr. Lorraine Kerr and Eliane Salvo-Chirnside for their helpful advice about RNA extraction. Also, we recognize the efforts of Andy Mash, Andrew Melrose, and Joe Warren of Peak Analysis and Automation (PAA) in the development of Edwin.

Authors’ Contributions

ADLH performed the experiments, built and ran automation procedures, and analyzed the data. WX wrote Overlord3 scripts for custom RNA extraction.

ADLH, WX, and VS set up and adjusted physical parameters to optimize the robotic platform processes. AE conceived of and designed the robotic platform. ADLH and AE designed the experiments and wrote the article.

Availability of Supporting Data

The data sets supporting the results of this article are available in the Edinburgh DataShare repository: .

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work is delivered as an activity of the Flowers Consortium, for which we recognize the financial support of the EPSRC (EP/J02175X).

References

Henegar

Cancello

Rome

. Clustering Biological Annotations and Gene Expression Data to Identify Putatively Co-Regulated Biological Processes. J. Bioinform. Comput. Biol. 2006, 4, 833–852.

Komili

Silver

P. A.

Coupling and Coordination in Gene Expression Processes: A Systems Biology View. Nat. Rev. Genet. 2008, 9, 38–48.

Ismail

K. S.

Sakamoto

Hatanaka

. Gene Expression Cross-Profiling in Genetically Modified Industrial Saccharomyces cerevisiae Strains during High-Temperature Ethanol Production from Xylose. J. Biotechnol. 2013, 163, 50–60.

Kara

Yumrutas

Ozcan

. Differential Expressions of Cancer-Associated Genes and Their Regulatory miRNAs in Colorectal Carcinoma. Gene 2015, 567, 81–86.

Kuntumalla

Zhang

Braisted

J. C.

. In Vivo versus In Vitro Protein Abundance Analysis of Shigella dysenteriae Type 1 Reveals Changes in the Expression of Proteins Involved in Virulence, Stress and Energy Metabolism. BMC Microbiol. 2011, 11, 147.

Toledo-Arana

Dussurget

Nikitas

. The Listeria Transcriptional Landscape from Saprophytism to Virulence. Nature 2009, 459, 950–956.

Wang

Chen

. RNA-seq Based Transcriptomic Analysis of Single Bacterial Cells. Integr. Biol. 2015, 7, 1466–1476.

Taniguchi

Choi

P. J.

G. W.

. Quantifying E. coli Proteome and Transcriptome with Single-Molecule Sensitivity in Single Cells. Science 2010, 329, 533–538.

Lockhart

D. J.

Winzeler

E. A.

Genomics, Gene Expression and DNA Arrays. Nature 2000, 405, 827–836.

10.

Pandey

Mann

Proteomics to Study Genes and Genomes. Nature 2000, 405, 837–846.

11.

Hua Hu

. High-Throughput Technologies for Gene Expression Analyses: What We Have Learned for Noise-Induced Cochlear Degeneration. J. Otol. 2013, 8, 25–31.

12.

VanGuilder

H. D.

Vrana

K. E.

Freeman

W. M.

Twenty-Five Years of Quantitative PCR for Gene Expression Analysis. Biotechniques 2008, 44, 619–626.

13.

de Las Heras

Carreno

C. A.

Martinez-Garcia

. Engineering Input/Output Nodes in Prokaryotic Regulatory Circuits. FEMS Microbiol. Rev. 2010, 34, 842–865.

14.

Tan

Chen

. Differential Expression of Virulence and Stress Fitness Genes during Interaction between Listeria monocytogenes and Bifidobacterium longum. Biosci. Biotechnol. Biochem. 2012, 76, 699–704.

15.

Silva-Rocha

de Lorenzo

A Composite Feed-Forward Loop I4-FFL Involving IHF and Crc Stabilizes Expression of the XylR Regulator of Pseudomonas putida mt-2 from Growth Phase Perturbations. Mol. Biosyst. 2011, 7, 2982–2990.

16.

Keren

Zackay

Lotan-Pompan

. Promoters Maintain Their Relative Activity Levels under Different Growth Conditions. Mol. Syst. Biol. 2013, 9, 701.

17.

Kosuri

Goodman

D. B.

Cambray

. Composability of Regulatory Sequences Controlling Transcription and Translation in Escherichia coli. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 14024–14029.

18.

Picard

Dressaire

Girbal

. Examination of Post-Transcriptional Regulations in Prokaryotes by Integrative Biology. C. R. Biologies 2009, 332, 958–973.

19.

Yosef

Regev

Impulse Control: Temporal Dynamics in Gene Transcription. Cell 2011, 144, 886–896.

20.

Isett

George

Herber

. Twenty-Four-Well Plate Miniature Bioreactor High-Throughput System: Assessment for Microbial Cultivations. Biotechnol. Bioeng. 2007, 98, 1017–1028.

21.

Zhao

Bishop

Clay

J. E.

. Automation of Large Scale Transient Protein Expression in Mammalian Cells. J. Struct. Biol. 2011, 175, 209–215.

22.

Warringer

Blomberg

Automated Screening in Environmental Arrays Allows Analysis of Quantitative Phenotypic Profiles in Saccharomyces cerevisiae. Yeast 2003, 20, 53–67.

23.

Huber

Ritter

Hering

. Robo-Lector—A Novel Platform for Automated High-Throughput Cultivations in Microtiter Plates with High Information Content. Microb. Cell Fact. 2009, 8, 42.

24.

Wollerton

M. C.

Wales

Bullock

J. A.

. Automation and Optimization of Protein Expression and Purification on a Novel Robotic Platform. J. Lab. Autom. 2006, 11, 291–303.

25.

Silva-Rocha

de Lorenzo

A GFP-lacZ Bicistronic Reporter System for Promoter Analysis in Environmental Gram-Negative Bacteria. PLoS One 2012, 7, e34675.

26.

Kelwick

MacDonald

J. T.

Webb

A. J.

. Developments in the Tools and Methodologies of Synthetic Biology. Front. Bioeng. Biotechnol. 2014, 2, 60.

27.

Sambrook

Fritsch

E. F.

Maniatis

Molecular Cloning: A Laboratory Manual; Cold Spring Harbor: New York, 1989.

28.

Anderson

J. C.

Anderson

: Anderson Promoter Collection. http://parts.igem.org/Promoters/Catalog/Anderson.

29.

Chappell

Freemont

In Vivo and In Vitro Characterization of σ70 Constitutive Promoters by Real-Time PCR and Fluorescent Measurements. Methods Mol. Biol. 2013, 1073, 61–74.

30.

Schmittgen

T. D.

Livak

K. J.

Analyzing Real-Time PCR Data by the Comparative C(T) Method. Nat. Protoc. 2008, 3, 1101–1108.

31.

Zwietering

M. H.

Jongenburger

Rombouts

F. M.

. Modeling of the Bacterial Growth Curve. Appl. Environ. Microbiol. 1990, 56, 1875–1881.

32.

Baranyi

Roberts

T. A.

A Dynamic Approach to Predicting Bacterial Growth in Food. Int. J. Food Microbiol. 1994, 23, 277–294.

33.

Mutalik

V. K.

Guimaraes

J. C.

Cambray

. Quantitative Estimation of Activity and Quality for Collections of Functional Genetic Elements. Nat. Methods 2013, 10, 347–353.

34.

Cherry

J. M.

Stewart

Frankenberger

. Automated Process of Q-PCR/Gene Expression. J. Lab. Autom. 2004, 9, 128–134.

35.

Fang

Willis

R. C.

Hoang

. High-Throughput Sample Preparation for Gene Expression Profiling and In Vitro Target Validation. J. Lab. Autom. 2004, 9, 140–145.

36.

Chappell

Jensen

Freemont

P. S.

Validation of an Entirely In Vitro Approach for Rapid Prototyping of DNA Regulatory Elements for Synthetic Biology. Nucleic Acids Res. 2013, 41, 3471–3481.

37.

Mathot

Wallin

Sjoblom

Automated Serial Extraction of DNA and RNA from Biobanked Tissue Specimens. BMC Biotechnol. 2013, 13, 66.

38.

Fleige

Pfaffl

M. W.

RNA Integrity and the Effect on the Real-Time qRT-PCR Performance. Mol. Aspects Med. 2006, 27, 126–139.

39.

Laurell

Iacovoni

J. S.

Abot

. Correction of RT-qPCR Data for Genomic DNA-Derived Signals with ValidPrime. Nucleic Acids Res. 2012, 40, e51.

40.

de Godoy

L. M. F.

Olsen

J. V.

Cox

. Comprehensive Mass-Spectrometry-Based Proteome Quantification of Haploid versus Diploid Yeast. Nature 2008, 455, 1251–1254.

41.

Amores

G. R.

Guazzaroni

M. E.

Silva-Rocha

Engineering Synthetic cis-Regulatory Elements for Simultaneous Recognition of Three Transcriptional Factors in Bacteria. ACS Synth. Biol. 2015, 4, 1287–1294.

42.

Guazzaroni

M. E.

Silva-Rocha

Expanding the Logic of Bacterial Promoters Using Engineered Overlapping Operators for Global Regulators. ACS Synth. Biol. 2014, 3, 666–675.

43.

Schlabach

M. R.

J. K.

. Synthetic Design of Strong Promoters. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 2538–2543.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.11 MB

Edwin

Abstract

Keywords

Background

Methods

Bacterial Strains, Culture Conditions, and General Procedures

Construction of Plasmids

Automated Procedures

Growth and Fluorescence Assays

RNA Extraction

RNA Analysis

Transcription Analysis by RT-qPCR

Detection of gDNA Contamination in RNA Samples

Statistical Comparison of Growth Curves

Results and Discussion

Requirements for Automatic Monitoring of Microbial Gene Expression Dynamics

Automated Inoculation and Cultivation of Microbial Cells

Customizing the Automated Procedure for Simultaneous Fluorescence/Growth Monitoring and RNA Purification/Analysis

Validating Edwin’s Ability to Measure Fluorescence and OD600 in an Automatic Procedure

Validating the Procedure Steps to Extract RNA by Analyzing Its Quality

RNA Extracted by Edwin Is Suitable for RT-qPCR Analysis

Conclusions

Footnotes

Acknowledgements

Authors’ Contributions

Availability of Supporting Data

Declaration of Conflicting Interests

Funding

References

Supplementary Material

Validating Edwin’s Ability to Measure Fluorescence and OD₆₀₀ in an Automatic Procedure