Optimized Replication of Arrayed Bacterial Mutant Libraries Increases Access to Biological Resources

ABSTRACT Biological collections, including arrayed libraries of single transposon (Tn) or deletion mutants, greatly accelerate the pace of bacterial genetic research. Despite the importance of these resources, few protocols exist for the replication and distribution of these materials. Here, we describe a protocol for creating multiple replicates of an arrayed bacterial Tn library consisting of approximately 6,800 mutants in 96-well plates (73 plates). Our protocol provides multiple checkpoints to guard against contamination and minimize genetic drift caused by freeze/thaw cycles. This approach can also be scaled for arrayed culture collections of various sizes. Overall, this protocol is a valuable resource for other researchers considering the construction and distribution of arrayed culture collection resources for the benefit of the greater scientific community. IMPORTANCE Arrayed mutant collections drive robust genetic screens, but few protocols exist for replication of these resources and subsequent quality control. Increasing the distribution of arrayed biological collections will increase the accessibility and use of these resources. Developing standardized techniques for replication of these resources is essential for ensuring their quality and usefulness to the scientific community.

M utagenesis of a given organism, followed by phenotypic selection or measurement of mutant fitness, is a cornerstone of experimental microbial genetics. High-quality, publicly available collections of mutants, such as the Escherichia coli Keio collection (1), Bacillus subtilis single-gene knockout libraries (2), and the Staphylococcus aureus USA300_FPR3757 transposon (Tn) mutant library (3), greatly accelerate the pace at which research can be performed and enhance scientific rigor and reproducibility. Construction of such resources requires significant time, labor, and resources, and it is inefficient for multiple laboratories to generate redundant biological resources. The research community would benefit from increased generation, replication, and dissemination of resources such as arrayed bacterial mutant libraries. Although robust protocols have been developed for manual or robotic arraying of colonies and mapping of arrayed collections of mutants (4)(5)(6)(7)(8), few exist for the replication of arrayed culture collections (9)(10)(11). Here, our goal was to establish a protocol and collection of best practices to minimize contamination and genetic drift of arrayed bacterial culture collections while increasing accessibility for other researchers.
We previously described the generation and application of an arrayed library of Tn mutants in Enterococcus faecalis OG1RF, consisting of ;15,000 individual clones (12). From this library, two targeted sequence-defined mariner Tn (SmarT) libraries were generated (5). The first SmarT library consists of 6,829 Tn mutants arrayed across 96-well plates (n = 73 plates), with insertions in approximately 70% of annotated genes and intergenic regions in OG1RF. The second SmarT library consists of 1,946 Tn insertions in poorly characterized genes and intergenic regions and was designed to facilitate genetic screens targeting uncharacterized regions of the genome (13). Both libraries are also available in pooled formats and have been used extensively to identify E. faecalis genes required for biofilm formation, metabolism, responses to antibiotics, phage infection, vaginal colonization in a mouse model, and polymicrobial interactions involving E. faecalis (14)(15)(16)(17)(18)(19)(20)(21).
In addition to these genetic screens, hundreds of individual Tn mutants have been distributed to domestic and international laboratories. We regularly receive requests for the entire Tn library, but it is not feasible to generate a new copy of the entire arrayed library for each individual request. Therefore, we sought to perform a large-scale replication of the larger SmarT library (6,829 mutants) to increase the accessibility of this resource for other laboratories, ensure quality control of the collection, and avoid genetic drift by decreasing the number of freeze/thaw cycles for the original library plates. Here, we present a protocol for efficient manual replication of arrayed library resources, including estimation of the time required (person hours). This protocol does not require access to robotic handling systems, making it feasible for researchers who do not have access to this specialized equipment. This protocol can be scaled to accommodate libraries of different sizes, as well as different numbers of replicates. We also describe multiple quality control checks throughout the process and compare sequencing-based verification of pooled mutants with previously published results. Additionally, because of ongoing supply chain difficulties due to the coronavirus disease 2019 (COVID-19) pandemic, we consider multiple options for consumables required throughout, as well as ergonomic considerations for technical staff.

RESULTS
Abbreviated protocol for large-scale replication and quality control of arrayed Tn resources. We manually created 15 copies of the 73-plate SmarT library over a 2-week period in April 2022. Required supplies and consumables are listed in Table 1. The approximate timeline for library replication on this scale is outlined in Tables 2 and 3. Additional protocol details can be found in File S1 in the supplemental material. To avoid repeated freeze/thaw cycles of the original library plates, all copies were created at the same time. An overview of the process is shown in Fig. 1. Frozen SmarT library stock plates were used to inoculate deep 96-well plates containing brain heart infusion (BHI) broth. Cultures were grown overnight and manually inspected for contamination of known blanks or lack of growth. If plates had either contaminated wells or wells with no growth, then the entire plate was discarded, and a new deep-well plate was inoculated. Overnight cultures were transferred from deep-well plates to prelabeled 96-well plates containing glycerol to generate individual library sets, which were stored at 280°C.
Multiple quality control checkpoints were used to prevent contamination. The entire process was carried out in a class 2 (A2) biological safety cabinet using biosafety level 2 (BSL-2) practices. Deep-well plates were filled with BHI broth and incubated overnight at 37°C prior to inoculation to ensure a lack of contamination. Replicating pins used for inoculation from freezer stock plates were sterilized in a series of ethanol baths between plates. To ensure against dilution effects that could reduce the effectiveness of alcoholbased disinfection, the ethanol baths were completely replaced after inoculation of four deep-well plates. Replicating pins were also autoclaved every night.
The original SmarT library layout included a mapped series of known blank/empty wells in each plate (5). These wells were manually inspected in the deep-well plates, and any plates with contamination were discarded and reinoculated. Any deep-well plates in which mutants did not grow as expected were also discarded and reinoculated. Because the deep-well plates do not fit in standard plate readers, optical density (OD)/absorbance was not measured. We recognize this as a limitation of the protocol, because such data could be important for identifying mutants with reduced overnight growth that is not detectable by eye. We strongly recommend that users who adapt this protocol for their own biological collections measure endpoint OD 600 values and provide these as a data set for the community. Importantly, we did not find any mutants in the entire 6,829-clone library that have lost viability since the initial library construction.
New copies of library plates were prefilled with glycerol and capped with sterile foil seals after addition of bacterial cultures. Plates were mixed by inversion after sealing (instead of pipetting) due to the number of samples. To ensure that this would not introduce crosswell contamination, we first empirically investigated the potential for cross-contamination. We filled a 96-well plate with glycerol, buffer, and bromophenol blue for visualization. This plate was covered with a foil seal, vortex-mixed at 2,500 rpm for 5 min, and incubated overnight on a rotating shaker. This agitation exceeded the brief mixing process performed with Tn library plates containing cultures and glycerol. No dye leakage or damage to the foil seal was observed ( Fig. 2A), suggesting that brief mixing of Tn library plates would not cause intraplate contamination. To quantitatively assess whether the exact process used with library plates would introduce interwell contamination, we filled a 96-well plate with E. faecalis OG1RF (;10 7 CFU/mL) or medium blanks. The plate was sealed, inverted, and incubated at 37°C overnight. Any well-to-well leakage would result in growth of OG1RF in the medium blank wells. We did not observe any contamination or growth outside the wells that were inoculated (Fig. 2B), strengthening our conclusion that the library plate mixing process does not cause intraplate contamination. This is consistent with our experience creating and handling the original arrayed Tn library. Pooling of the SmarT library from 96-well plates and Tn sequencing. The SmarT libraries are available in arrayed format and as a pooled version, in which all Tn mutants have been combined in equal amounts to facilitate Tn sequencing (Tn-Seq) or similar genetic screens. Previously, the pooled versions were created by plating aliquots of each strain on BHI agar plates, followed by scraping and combining all mutants (5). This was done to ensure that roughly the same numbers of mutants would be present in the pooled library regardless of in vitro growth defects in liquid medium and to screen each mutant stock for contamination. To determine whether we could pool liquid cultures of the SmarT library after growth in deep-well plates and still achieve a similar balance of mutants, we combined ;200 mL remaining from each deep-well plate after distribution of the cultures into the new library plates. DNA was extracted from the pooled cultures, and Tn abundance was determined using Illumina sequencing and previously established protocols (5). These results were compared to those for samples extracted from the original pooled SmarT library in a previous experiment (17).
We first compared the number of Tn mutants identified from sequencing with the known number of mutants in the arrayed library (n 5 6,829). In the new pooled library, 250 mutants were missing (0 reads) in all extracted replicates (358, 336, and 347 mutants in the individual replicates) (see Table S1 in the supplemental material). A total of 192 mutants had 0 reads in all previously sequenced samples (275, 259, 268, and 267 mutants in the individual replicates) (17). A total of 166 mutants had 0 reads mapped in all replicates of the old and new pooled libraries. These mutants might be missing due to incomplete lysis of cells (perhaps due to physiological changes due to the disrupted gene), instability of the Tn  insertion in the chromosome, or loss of DNA during preparation and sequencing. Because we did not find any mutants that did not grow in deep-well plates during library replication, we do not think that these mutants are missing from the sequencing results due to a loss of viability. We next examined the similarity in relative abundances of mutants in the original pooled library and the new library pooled from liquid cultures (Fig. 3A). In addition to a greater number of mutants with 0 reads, the new pooled library had a broader distribution of relative abundance frequencies (Fig. 3B). Low-abundance mutants in the new pooled library (relative abundance, 0 to 0.00001) had higher relative abundances in the original input library (Fig. 3C). However, mutants that were absent from or had relatively low abundances in the original pooled library also had low abundances in the new pooled library (Fig. 3D). Overall, we conclude that the original pooled library created by collecting cells grown on agar plates had a more even distribution of mutants than the new pooled library created by combining liquid cultures. Based on this, we strongly recommend generating new pooled libraries using our original method of plating and scraping individual mutants.

DISCUSSION
Culture collections and arrayed mutant libraries are valuable biological resources that increase the throughput, rigor, and reproducibility of experiments across an entire scientific field. To avoid redundancy and wasted resources, these collections and libraries should be broadly available to researchers. While core facilities or private companies may have resources to generate arrayed library copies using robotic arraying and liquid-dispensing equipment, this remains beyond the reach of most academic research laboratories at many institutions. Therefore, we sought to establish a protocol for manual replication of arrayed library collections that would increase the accessibility of these biological resources while maintaining high quality control standards and preventing genetic drift due to repeated freeze/thaw cycles of arrayed culture collections.
Using this approach, we created 15 copies of a large arrayed E. faecalis OG1RF Tn library; we have already distributed most of these library sets to other research groups. We also pooled Tn mutants grown during library replication and used Tn-Seq to compare mutant distribution from this pool to the results for a previously generated pooled library in which individual mutants were scraped from agar plates. Although we found that a majority of Tn mutants (.96%) were present in our new pooled library, we observed greater representation of low-abundance mutants using the previous approach of scraping and pooling mutants from agar plates. The liquid pooling approach described here could have been strengthened by measuring endpoint OD 600 values for each library plate after overnight growth and adjusting the volume pooled for each mutant accordingly. Although the plating and scraping approach is more labor-intensive, we strongly recommend using that approach to generate new pooled libraries, instead of pooling together liquid cultures. This methodology can guide the creation and distribution of arrayed mutant collections in a variety of microorganisms.

MATERIALS AND METHODS
Bacterial strains and culture conditions. The 6,829-clone E. faecalis arrayed Tn library was previously generated and stored at 280°C (5,12). BHI broth (BD Difco) was used for overnight growth. Prior to inoculation, plates were filled with BHI broth, incubated at 37°C, and manually inspected for contamination prior to inoculation. Tn mutants were inoculated, using a metal replicating pin (Boekel Scientific), into 2-mL deep 96well plates (Biotix) containing 1.8 mL BHI broth. Plates were grown overnight at 37°C without shaking and were manually inspected for contamination prior to distribution of cultures to new library plates.
Preparation of library plates. Sterile flat-bottomed 96-well plates (Thermo Fisher Scientific) were labeled on the lid and two sides of the plate with printed cryovial labels (LabTAG). Labels contained the library copy number (from 1 to 15) and plate number (from 1 to 73). One hundred microliters of autoclaved 50% glycerol (VWR) was dispensed into each well using multichannel electronic pipettes.
Generation of new library copies. From each deep-well plate, 100 mL of overnight culture was distributed into 15 prelabeled 96-well plates (prefilled with 100 mL 50% glycerol) using multichannel electronic pipettes. Sterile AlumaSeal adhesive foil seals (Life Science Co.) were applied with a small rubber roller (Speedball). Plates were sorted and stored at 280°C.
Evaluation of intraplate leakage. A 96-well plate with 0.1% bromophenol blue in phosphate-buffered saline (pH 8) and 10% glycerol was set up in a checkerboard pattern, with a total volume of 200 mL per well. The plate was capped with an AlumaSeal adhesive foil seal (Life Science Co.) and sealed with a small rubber roller (Speedball). The plate was mixed by inversion, vortex-mixed for 5 min at 2,500 rpm on a multitube vortexmixer (Benchmark Scientific), and incubated overnight on a Belly Dancer orbital shaker at ;50 rpm. Leakage of dye between wells was evaluated by eye. For quantitative measurement of intraplate leakage, a 96-well plate was inoculated with alternating wells of E. faecalis OG1RF (;10 7 CFU/mL) in BHI broth plus 10% glycerol or medium blanks. The plate was capped with a foil seal, mixed by inversion, and incubated at 37°C overnight, after A test plate with glycerol and bromophenol blue was sealed with foil, mixed by inversion, vortex-mixed, and incubated overnight on a shaking platform to ensure that the inversion process used to mix glycerol and bacterial cultures would not create contamination between wells. (B) A 96-well plate was inoculated with alternating wells of E. faecalis OG1RF (;10 7 CFU/mL) or medium blanks. To mimic library plate processing, the plate was capped with a foil seal and mixed by inversion. The plate was incubated at 37°C overnight, after which an OD 600 measurement was taken to evaluate growth. The heatmap values are the averages of three independent biological replicates.

Replication of Arrayed Strain Libraries
Microbiology Spectrum which an OD 600 measurement was taken using a Biotek Synergy H1 plate reader. Data shown are the average of three independent biological replicates. Preparation of pooled Tn samples for Tn-Seq and Tn-Seq analysis. Two hundred microliters was pooled from each well of each deep-well plate after the distribution of mutants to new library plates. Samples were pooled in 50-mL Falcon tubes, pelleted at 6,000 rpm for 10 min in a Beckman Coulter Avanti JXN-30 floor centrifuge, and stored at 220°C until further use. DNA was extracted using a Qiagen DNeasy blood and tissue kit with a lysozyme pretreatment step, as described previously (17), and submitted to the University of Minnesota Genomics Center. Libraries were prepared using a NEBNext Ultra II FS DNA library preparation kit for Illumina and a Nextera XT Index kit v2 set A. Libraries were sequenced using a NextSeq P1 flow cell (150-bp paired-end reads), with ;2.1 million reads per sample (;300 reads/mutant). Mutant abundances were quantified using custom scripts, as described previously (5,17).
Data availability. Tn-Seq data have been deposited in the NCBI GEO database under accession number GSE233193.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.