Bacterial cloning was first introduced over a century ago and has since become one of the most useful procedures in biological research, perhaps paralleled in its ubiquity only by PCR and DNA sequencing. However, unlike PCR and sequencing, cloning has generally remained a manual, labor-intensive, low-throughput procedure. Here we address this issue by developing an automated, computer-aided bacterial cloning method using liquid medium that is based on the principles of (i) limiting dilution of bacteria, (ii) inference of colony forming units (CFUs) based on optical density (OD) readings, and (iii) verification of monoclonality using a mixture of differently colored fluorescently labeled bacteria for transformation. We demonstrate the high-throughput utility of this method by employing it as a cloning platform for a DNA synthesis process.

Transformation followed by monoclonal culture of bacteria (i.e., cloning) is arguably the most widely used procedure in biological research. Two methods for obtaining monoclonal bacterial cultures have been established: limiting dilution of bacteria in a liquid growth medium, and plating bacteria onto solid medium in Petri dishes (1). At present, biologists have largely neglected the original dilution-based method (1) and instead routinely plate and pick colonies one-by-one from solid medium in Petri dishes. However, increasing requirements for high-throughput cloning (2-9) call for a reexamination of how monoclonal bacterial culture is obtained. Here we demonstrate that high-throughput cloning can be realized by reverting to the use of a dilution-based technique. Recent techno-logical advancements (e.g., plate readers and liquid handling robots) enabled us to develop a high-throughput cloning methodology using liquid growth medium for standard multiwell plates that can be managed and analyzed in real time by software developed in-house, and processed using liquid-handling robots and plate readers.

Materials and methods

All of the automated procedures listed herein were executed using the Tecan Freedom EVO 200 system (Männedorf, Switzerland) with the relevant automated peripheral equipment, including a liquid handling arm (LiHa), a robotic manipulator arm (RoMa), automated plate reader, centrifuge, plate shaker, and incubator. Real-time analysis and control over the processes were executed using specialized software developed in-house. Additional methods and robot control scripts using this software for the automated procedures listed herein can be found in the Supplementary Materials.

Automated transformation

Transformations were performed using an automated procedure (See script in Supplementary Materials) into Z-competent Escherichia coli (Zymo Research, Orange, CA, USA) according to manufacturer's specifications.

Automated inoculations

Cells were taken out of the plate reader-incubator with real-time OD monitoring (Tecan Infinite 200 PRO series) at the predetermined OD₆₀₀ value of 0.2 and 5 x 10⁵ with Luria Bertani broth (LB) (see script in Supplementary Materials). Thirty-microliter inoculations were dispensed into 384-well plates containing 40 µL LB/well for a total volume of 70 µL/well.

Automated plasmid and PCR product purification

Clones were grown overnight in 1.3 mL LB and plasmids were extracted using the QuickClean 96-Well Plasmid Miniprep Kit (Genscript, Piscataway, NJ USA).PCR reactions were purified using the Zymo Research DNA Clean & Concentrator-5 kit.

Automated sequencing

Automated sequencing was performed using the BigDye Terminator v3.1 Cycle Sequencing Kit (ABI, Carlsbad, CA, USA) and sequencing reactions were purified using the 96-well Performa DTR kit (EdgeBio, Gaithersburg, MD, USA).Purified sequencing reactions were then electrophoresed on a 3730x196-capillary DNA Analyzer (ABI).

Results and discussion

Our high-throughput liquid-based cloning methodology is designed to be amenable to automation using standard off-the-shelf components. To this end, we programmed a Tecan Freedom EVO 200 system using in-house robot control software to carry out the method. Automated procedures were developed for the entire process including ligation, transformation, clonal amplification, plasmid purification and DNA sequencing. A general overview of the method, without specification of robotics or software parts, is depicted in Figure 1.

Figure 1. The five-step process for accurately performing end-to-end automated bacterial cloning in liquid. (Click to enlarge)

High-throughput cloning of bacteria in liquid medium

For automated bacterial cloning, competent bacteria not requiring heat shock or recovery were transformed in 96-well plates using an automated procedure. Following transformation, all cultures were diluted 1:860 into transparent 96-well plates in LB-Amp so that their OD₆₀₀ measurement (0.075 ± 0.005 using 300 µL LB-Amp aliquots) was not masked by the absorbance of untransformed dead cells.

Diluted transformations were then cultured at 37°C in a plate reader with automated real-time OD₆₀₀ monitoring (with automatic subtraction of the background OD₆₀₀ reading of 0.075). The CFU of each transformation was determined based on timing each culture's growth to the OD₆₀₀ value of 0.1 (see “Analog CFU inference” section, and Supplementary Materials). After CFU determination, if it appeared that the ODs of all the transformed cultures were not synchronized (due to different CFUs), they were then synchronized using an automated dilution-based procedure. Based on the OD readings, robotic scripts were automatically generated using in-house software to equalize ODs by the appropriate dilutions. Once the robot executed the scripts, the cultures were placed back in the plate reader to verify that the ODs were synchronized. If OD readings indicated that they were still not synchronized, the procedure was repeated. This synchronization step allows for the use of a uniform dilution factor for all transformations in the single-cell inoculation step.

The synchronized transformations were cultured until they reached an OD₆₀₀ of 0.2 ± 0.02 (which equals 4 × 10⁶ cells/mL), whereupon they were all diluted by a factor of 5 × 10⁵ to obtain a concentration of 8 viable transformed cells per mL. Inoculation of 30-µL aliquots into 384-well plates resulted in ~25% of wells containing a single cell. In principle, other OD values still within the range of linear correlation to cell density can also be used, albeit with a different dilution factor. Once grown overnight, the 384-well plates displayed a predefined ratio of positive (growth) to negative (no growth) wells of ~1:3, as determined by a plate reader OD scan. The dilution that resulted in ~25% of wells containing single cells from OD₆₀₀ = 0.2 was determined beforehand in pilot experiments (See Supplementary Material, “Description of pilot experiments to determine the dilution to achieve single cells” section and spreadsheet). Diluting to single cells using this method is independent of transformation efficiency (See Supplementary Figure 5)

Verification of clonality using DNA sequencing

Culture clonality in individual wells was accomplished through a combination of an appropriately low positive to negative well ratio, as well as through the optional use of a mixture of fluorescently labeled bacteria of different colors during transformation, as described in the “Verification of monoclonality using fluorescent bacteria” section. The maximum number of monoclonal wells was obtained when the average number of viable cells/aliquot is 1; however, this does not minimize the number of polyclonal wells. The cost and effort of sequencing false positive (i.e., polyclonal) colonies outweighs that of plating more negative wells, and therefore we aim for a low ratio of positive: negative wells. We chose a ratio of 1:3, which statistically favors the monoclonality of the positive wells. Nevertheless, since we could not rule out possible false positives caused by clumping of bacteria or contamination, the frequency of polyclonal wells was determined by DNA sequencing.

Using our automated liquid cloning method, we cloned a synthetic DNA library of a 768-bp fragment that was mutated at a high frequency. Previous comprehensive sequencing analysis showed that clones from this library have an average of 4–5 mutations randomly positioned along the 768-bp sequence (10). As a result, this library is, in effect, barcoded, with each clone having a unique mutational pattern. Monoclonal and polyclonal cultures from this cloned library are easily distinguishable using DNA sequencing, since polyclonal cultures always harbor more than a single plasmid sequence due to the high error-rate of the library (See Figure 2B). Four positive wells from our cloning procedure were each manually plated onto four separate Petri dishes, and three colonies from each Petri dish were manually picked and sequenced. Sequence analysis showed that colonies picked from the same Petri dish (i.e., inoculated with cells from a single positive well) reproducibly propagated plasmids with the exact same pattern of mutations. Conversely, groups of colonies picked from different Petri dishes (i.e., those groups inoculated from different positive wells) reproducibly propagated plasmids exhibiting a completely different pattern of mutations from each other. These results attest to the monoclonality of the positive wells and exclude false positive results caused by cell aggregates or contamination. Therefore, false-positive clones were not produced at any significant rate when a low ratio of positive to negative wells was used.