|
|
|
|
|
Preparation
and Curation of Nucleic Acid Collections
General Guidelines for Safe Handling of Glycerol Stocked
cDNA Collections
Replication and Growth of Glycerol Stocks
Production of
Plasmid DNA for Spotting
Manual Alternatives
to Curation and Production Automation
General Guidelines
for Replacing Automation with People
General Guidelines for Safe Handling
of Glycerol Stocked cDNA Collections
Collections are stored at -80oC and great care is taken to minimize the time the cultures are thawed while working in production. Typically we have found that if cultures are left thawed for more than six hours the glycerols will suffer from instability manifested as reduced growth in replicated wells or lost or weakly growing wells in subsequent replications or inoculations. In addition to minimizing thaw times, several replicate copies of each set of vendor plates are made with the specific purpose of production use. Once 'working copies' have been produced that exhibit great reliability as measured by even and complete growth and survival over at least one freeze-thaw cycle, the master plates are no again thawed unless all working copies become non-viable or in the instance of contamination of working copies with phage or non-coli bacteria. Working copies themselves may be replicated several times a year to replace copies that have become non-viable as a consequence of repeated freeze-thaw cycles. Thus only in extreme circumstances are the original master plates re-thawed and used to again make working copies. Using this method we hope to ensure that the 'master' stocks will maintain viability for the longest period of time. Although it varies from collection to collection and from strain to strain, a good glycerol stock in 96-well format should survive at least three freeze-thaw cycles before exhibiting any instability, and often this can be extended to five or more cycles if thaw time is minimized. We also store copies of each collection, often the first copies of the daughters made from the masters, in separate freezers in off-site facilities to prevent catastrophic loss due to power outages or freezer malfunction.
Replication and Growth of Glycerol
Stocks
Prior to replication of glycerol stocks or propagation of miniprep cultures, we carefully plan our strategy to ensure that we are employing the proper plate type and media for each step in the entire production process. While plates and miniprep consumables suitable for bacterial work are available from many vendors, slight differences in plate dimensions or flow-through characteristics typically require that each plate type be "taught" to the automation instruments, which has the effect of requiring plate-specific run routines. Plate teaching and run routine development are highly tedious and error-prone processes. For example, the routine we employ to obtain reliable, high-yield minipreps using the MWG RoboPrep 2500s running Macherey-Nagel NucleoSpin Robot-96 Plasmid miniprep consuimables has several hundred steps and required several weeks of trial-and-error optimization. Thus it would be wise if not following our steps precisely, i.e. employing a different automation suite or using different consumables for miniprep and normalization, to decide upon a process, optimize it for your specific needs using a known set of plates and consumables, and then, once optimized, refrain from making changes. To replicate glycerol stocks, source plates are placed in a laminar flow hood to thaw, which requires 30 to 60 minutes. To avoid cross-contamination among neighboring wells, we recommend un-foiling plates while they are frozen rather than after they have thawed. Either before our during source plate thawing, destination plates are filled using a Genetix QFill2 or other multi-drop device capable of pipetting from 40 ul to 1.5 mls per well in both 96- and 384-well format. Plates should be filled to around the halfway, allowing plenty of room for volume expansion during freezing and reducing the possibility of sloshing bacterial cultures amongst neighboring wells. Since working copies of glycerol stocks are frequently employed as source plates for re-arrays using our Genetix "Q"Bot for which these plates have been taught, we make all of our glycerols in 96- or 384-well Genetix plates with plastic covers, for which the "Q"Bot has been pre-taught. We use LB media for making our glycerol stocks and TB media for propagating bacterial cultures for mini-preparation of plasmid DNA. The antibiotic used depends, of course, on the bacterial selectable marker presented on the collection's vector(s). Glycerol content also varies depending on E. coli strain. Typically we have found that a solution of 8% glycerol will work well for most cell lines; however, DH5 alpha and certain other cell stains will respond better to a concentration of 10% or higher. This is theorized to be the result of greater freeze thaw protection and a greater density of the stock solution preventing cells from drifting to the bottom of wells. Although it is rare to have a choice in the cell strain used to propagate a collection, if a choice is presented, employ whenever possible those E. coli strains that are phage resistant, such as DH10B-T1, although we recommend avoiding those phage resistant strains harboring F' episomal plasmids, as the inheritance of such large plasmids tends to reduce yield of plasmid DNA in our hands and also to contaminate plasmid DNA when using Macherey-Nagel consumables. Once source plates to be replicated have thawed completely, each source plate is replicated by inserting a sterile, single-use disposable plastic 96- or 384-pin replicator (Genetix), swirling the replicator around to "mix" the culture, removing the replicator carefully so as not to dribble into adjacent wells in the source plate, and then inserting the replicator pins, each bearing a droplet of culture media, into the wells of the destination plate containing glycerol media, leaving the replicator in the wells for several minutes to permit bacteria to disperse into the media from the replicator pin. We have found also that some collections are available in cell strains that do not propagate well from mother to daughter over many cycles using these one-use sterile pin replicators, probably as a consequence of both cell settling during freeze-thawing and gradual differential growth resulting in dilute cell concentration. If a gradual increase in the number of dead wells per mother-daughter replication cycle is observed, one should switch to using liquid inoculation of glycerol stocks, either using a Genetix "Q"Bot fitted with a liquid handling head or another automated pipettor capable of autopipetting volumes greater than 5 ul using disposable tips, such as a Beckman FX. Once inoculated, destination plates are covered with their plastic lids, wrapped with Fisher Scientific wrap to prevent dessication, and incubated 18-36 hours at 37oC depending upon collection replicated and the cell density of the source stock. After growth, stocks are visually inspected plate by plate and if viable, are foiled and placed directly at -80oC for long-term storage.
Collections in which every well contains a clone encoding a full-length gene are often unavailable. The MGC collection is a good example, where at best only 70% of the clones in any given IRAK plate encode a sequence-verified full-length gene. Although screening costs per well are quite low, even a 30% non-functional gene rate may inflate screening costs per annum by tens of thousands of dollars, necessitating condensation of each collection to only those clones encoding a functional product. Re-arrays are also frequently undertaken in cases when researchers desire to screen only certain functional gene families or some other smaller subset of a larger collection of gene-containing arrayed clones. Once the bioinformatics have been completed and the locations of the full-length genes in the vendor collection have been identified for any given re-array, we use a Genetix "Q"Bot to re-array the pertinent wells into a collection for screening. The .imp file generated previously (an .imp file instructs the "Q"Bot which wells to pick from which source plates) is loaded onto the "Q"Bot's operating software and a 96-pin picking head used to transfer the bacteria carrying the genes of interest from the 384-well source plates to the 96 well destination plates. Source plates can remain foiled with the lids on during this procedure, although care must be taken to ensure that pins don't become too "gummed" up - some foils with particularly thick adhesive can transfer adhesive to the pins during firing through the foil, causing the pins to become sticky and mis-fire, resulting in either dead wells or bent pins and broken plates. Two hotels carrying a total of 72 source plates can be simultaneously loaded; if a re-array requires more than 72 source plates, then the operator will have to pause the run to switch out the hotels once the first 72 source plates have been picked. Once a particular source plate has been picked during a picking run, the "Q"Bot plate handler replaces its plastic lid, picks the plate up off the deck and re-inserts it into the source plate hotel, extracts the next source plate from the hotel, positions it on the deck, and removes its plastic lid. Since the entire plate switching process can take upwards of a minute per source plate, the "Q"Bot's advantage over a manual re-array declines as clones become more rarely distributed among source plates. For example, picking 100 clones out of 100 source plates is much more efficiently performed by hand than on the "Q"Bot, both in terms of speed and, given that the "Q"Bot requires that source plates be thawed, in terms of source plate glycerol stock viability. Destination plates are prepared as outlined in the section dealing with glycerol stock replication, i.e. re-arrays are always prepared as glycerol stocks. Once the robot has completed the re-array protocol, the destination plates are removed from the robot and grown at 37oC until the stock is confluent and then foiled and stored at -80oC. Because of our use of different E. coli strains, media, antibiotics, and growth times will vary; however, 18-22 hours of growth time is typical. Note that it is theoretically possible to re-array from one 384-well source plate into a second 384-well source plate. Nevertheless, since screening requires relatively large amounts of DNA such as we have only been able to obtain with conventional 96-well bind-and-elute miniprep consumables, clones must eventually pass from 384- into 96-well format anyway and we view the best point to do this as at the re-array stage. Once the 96-well format of the re-array has grown, one additional re-array is produced whose chief aim is to insert four wells lacking bacteria for subsequent use for spotted assay controls. This re-array, which we call the "92-well" re-array, has the same pattern as the 96-well re-array except that the bacteria in wells G11, G12, H11, and H12 are transferred in order to separate 96-well plates whose wells G11-H12 are also left blank. If screens are run in 384-well format, which is essentially four 96-well plates, the four sets of G11-H12 empty wells become a 4x4 array of empty wells for controls in the 384-well plate: M21-24, N21-24, O21-24, and P21-24. Obviously, great care must be taken at all phases of a re-array to ensure clone placement precision in order to not squander the great benefit of addressable, arrayed-well screening: that of knowing the identity of all genes that hit in a given assay immediately after the assay has been run.
Production of Plasmid DNA for spotting:
After the final 92-well format re-array has been produced (note the use of the term '92-well' rather than '96-well'), the process is ready to move to the actual production phase. 96-well deep well culture blocks are filled with 1.5ml of terrific broth (TB) media containing the pertinent antibiotic using a QFill2 fitted with a high, 12-channel manifold. Filled deep-well blocks are inoculated in one of three ways: by using a "Q"Bot equipped with a 96 well picking head and "Christmas Tree" pins (conical pins serrated with ridges to facilitate more glycerol transfer), using the liquid handling head of the "Q"Bot for direct liquid transfer, or with a flame inoculation protocol using a 96 pin hand tool. Labs lacking automation would best benefit from using and 6-pin metal hand innoculator. Warning: if used, a metal-pin, 96-pin innoculator must be sterilized between plates by ethanol flaming and both the amount of ethanol required and the resulting flame represent a significant fire hazard. This procedure should only be performed in a hood in the absence of any flammable material, either in the hood on the floor below. Prior to live use of a 96-pin hand innoculator, we highly recommend practicing its ethanol sterilization in a well-isolated, flameproof location. After inoculation, the blocks are covered with gas-permeable air-pore tape, which is normally provided along with the deep-well blocks supplied with 96-well miniprep kits, but which can also be purchased separated. The deep well blocks are then loaded into a Gene Machines HiGro four-tower rotating incubator and are grown at 37oC, with oxygen flow for a half second burst every thirty seconds, and with towers rotating at 350-400 rpm. Typical growth time of deep-well block production culture is 16-18 hours but varies depending upon antibiotic, plasmid backbone (high-copy ori versus low-copy ori), E. coil strain, media type and the volume of initial inoculate used. Production of high-quality plasmid DNA at maximal yields requires adequate but not excessive bacterial growth, necessitating visual inspection of each deep-well block after 16 hours of growth in the HiGro. Indications of sufficient growth are typically a dark yellow culture with perhaps a small amount of cells accumulated at the bottom of the wells. No growth should have occurred in the 4 blanks wells reserved for subsequent use as positive controls. Based upon insert size of the plasmid or variances in the amount of starting innoculate, cultures may vary in density from well to well. Care should be taken to minimize this variance, as bacterial culture density is positively correlated with DNA yields, and variation in DNA concentration is in turn positively correlated with variation in transfection efficiency. It is common for several wells in each block to fail to grow after inoculation using an apparently live glycerol stock, particularly when inoculating using christmas tree pins on the "Q"Bot or the metal pin hand-innoculator. One to two dead wells per block are tolerable under high-throughput processes. If this number exceeds five wells then we recommend employing the more tedious step of performing liquid inoculations, ensuring when so doing to only use fresh, sterile, single-use PCR tips and to pipette vortex the glycerol media. Immediately after growth, deep-well blocks are centrifuged at >1500 x G in a plate centrifuge to pellet the bacteria and the bacteria-free media discarded by inverting the blocks over a sink. Insufficient centrifugation will result in pellet loss upon block inversion and by no means should blocks be left inverted to drain, as pellet will commonly dis-adhere from the well bottom and either slide out of the well or to the top of the well, where pellets can become cross contaminated. Should the blocks have insufficiently grown at this stage, one can simply vortex the blocks briefly, re-tape, and incubate longer in the HiGro. Once sufficient bacterial growth has been achieved and the bacteria have been pelleted, miniprep re-suspension buffer including RNase is added by Genetix Qfill2, blocks are foiled with adhesive plate sealing foil, and the bacterial pellets re-suspended using a plate vortexer. Care must be taken to ensure complete re-suspension of the bacterial pellet as residual bacterial clumps will clog the fixed probes of the automated miniprep instruments, stopping the prep run for that and all subsequent plates on the mimprep robots. Once re-suspended in miniprep re-suspension buffer, bacteria in blocks can be either miniprep'd immediately or stored at -20oC. If starting a miniprep run from frozen, re-suspended bacteria, the frozen deep-well blocks should be allowed to thaw overnight at 4oC and then, so as to prevent buildup on condensation on the sides of the block that could potentially interfere with block handling on the automation, be allowed to acclimate further to room temperature briefly.
Plasmid DNA from re-suspended bacterial pellets is prepared according to the protocol dictated by the manufacturer of the miniprep consumbables we favor, those of Macherey-Nagel GmBH. In extensive field testing we have found the Macherey-Nagel miniprep consumables to provide the best combination of yield, quality (transfectability), and reliability. We have found the MN post-lysis plasmid filter plate to be the only truly clog-free plate on the market, permitting routine use of the richest bacterial medias such as TB and thereby increasing bacterial density and plasmid DNA yields. The mini preps are processed on an automated miniprep/normalization suite comprised of a pair of MWG Biotech RoboPrep 2500s linked in tandem to an R16 arm, a PowerWave spectrophotometer (BioTek), a barcode reader, an MWG Biotech RoboSeq 4204, and an ABGene plate sealer. Plasmid DNA is eluted into Costar 96-well UV plates. Since the well-to-well concentration of DNA in the minipreps can vary and DNA yields are typically too high to permit direct spotting into 384-well assay plates, DNA concentrations of the raw minipreps are then normalized using a two-step dilution protocol either on the integrated MWG 4204 liquid handling robot off-line using a much faster MWG Biotech THEONYX liquid handler. The diluent is the same buffer as the elution buffer from the Macherey-Nagel kit (5 mM Tris-HCl, ph 8.5). Both the integrated miniprep/normalization suite and the off-line rapid normalization suite are currently equipped with integrated BioTek PowerWave XS 96-well plate spectrometers that enable intervention-free importation of UV data to the liquid handlers, which are equipped with database software that tracks DNA concentrations and calculates normalization volumes. Typically, cDNA collections to be used for screening are normalized to a concentration of 40 ng/ul with a final volume of 80 ul. Round-bottom 96-well plates are employed as destination plates during normalization, which has the effect of concentrating the DNA in a concave, well-centered bubble, enabling the spotting robot to maximally spot the DNA from the normalized, round-bottomed plates. Once normalized, UV data from the UV plates is used to generate a data file identifying wells containing DNA concentrations meeting or exceeding the defined normalized concentration, in our case typically 40 ng/ul. DNA in the UV plates at lower concentrations does not get transferred to the round-bottom normalization plate that will be used for spotting, i.e. these wells contain no DNA at all in the normalized plate. Normalized DNA can be stored for extended periods in foiled plates at -20oC prior to spotting into assay plates. Once an entire collection of plasmid DNA has been prepared and normalized, the DNA is ready to be spotted into assay plates. Typically we will spot all the plasmid DNA at once into many screening copies, thereby providing ready access to screening copies as needed, preventing multiple freeze-thaw cycles of the source DNA (although this is a minor concern with plasmid DNA, it becomes a major concern for packaged viral stocks, which we ultimately hope to produce in a similar fashion), and minimizing spotting time, expense, and operator involvement. Generally, the type of destination plates used for screening are customized for the particular types of cell-based assays envisioned for the collection and their detection methodologies; we typically rely most heavily upon luminescence endpoint determinations and hence mostly use white or black Greiner 384-well assay plates, although for phenotypic endpoints we have also spotted into Greiner 384-well clear bottom plates. For assays to date, the amount of DNA spotted per well has depended on the type of collection screened, the quality of the miniprep DNA, and the desired ratio of DNA to transfection reagent which, of course, depends heavily on the type of transfection reagent employed, cell types to be screened, etc; for cDNA over-expression assays employing Fugene as transfection reagent, each well of 384-well plate usually receives 62 ng DNA/well, whereas for shRNA knock-down assays, each well usually receives 20 ng of DNA/well. Once DNA quantity and plate types have been determined, physically spotting the DNA into the assay plates is performed using a Beckman MultiTrak fitted with disposable tips. One 96-well round-bottomed plate containing normalize plasmid DNA is placed on the deck and small aliquots of DNA from each well spotted into the first quadrant of each 384-well assay plate. We call this the A1 quadrant. Thus, DNA from well A1 of the A1-quadrant 96-well source plate goes into well A1 of the 384-well plate. DNA from well A2 of the A1-quadrant 96-well source plate goes into well A3 of the 384-well plate and so on. Once the A1-quadrant 96-well source plate has been spotted into all copies of that 384-well assay plate, the assay plates are re-stacked and the A2-quadrant 96-well plate is spotted into them, with well A1 of the quadrant A2 96-well source plate spotted into well A2 on each copy of that 384-well assay plate. Next comes spotting of the B1-quadrant 96-well plate (A1 of 96 to B1 of 384) and finally the B2-quadrant 96-well plate (A1 of 96 to B2 of 384). It is absolutely critical that the individual(s) performing the spotting keep track of which 96-well plates went to which quadrant of which 384-well assay plates, so we recommend keeping a tracking log with this data. Once a series of spotted 384-well assay plates have been completed, each 384-well plate is foiled, and subsequent series are spotted and foiled from similar groups of four 96-well plates until all plates have been done. 384-well assay plates can then be stored indefinitely at -20 oC or -80 oC for use in assays as needed.
Manual Alternatives to Curation and
Production Automation
We view the sorts of automated instruments employed here as essential to the process of screening. Nevertheless, many of the processes we've described can be performed manually, albeit, with a commensurate increase in the amount of labor that must be devoted. The only process that cannot be straightforwardly performed manually is DNA normalization, although it is possible to make plate-by-plate UV measurements in a 96-well spectrophotometer, download each plate's UV data into Microsoft Excel, calculate normalization volumes, and then upload a normalization file to a single- or multi-probe automated liquid handler or pipettor where a single-plate normalization is performed. One could even circumvent the need to normalize by lowering one's expectations with regard to the noise in one's assays, i.e. normalized DNA gives normalized transfection efficiency; inclusion of a transfection efficiency reporter may aid in this process. Bacterial production cultures can be grown in a shaking incubator, although in our experience this tends to halve the resulting plasmid DNA yield. 96-well minipreps can be performed using vacuum manifolds at nearly the same speed as using a pair of RoboPrep 2500s, although this would require continuous user intervention.
Some general guidelines for replacing automation
with people:
· Employ digital pipettors wherever possible; robots aren't subject to carpal-tunnel syndrome but people are. · Use PCR filter tips for all pipetting processes in order to protect the collections from introduction of exogenous, contaminant bacteria. · Invest in good quality multi-channel pipettors, again, digital, if possible. · Vacuum manifold minipreps are best performed using Finnpipette digital multi-channel pipettors fitted with the 8-channel 1500 ul dispenser head. · If you purchase any automation, it should be a Qfill2 or other device capable of 8- or 16-channel pipetting 40-1500 ul into shallow and deep-well blocks. · All bacterial media should be autoclaved and the functioning of the autoclaved routinely tested by pre-innoculating a sentinel bottle of media with live E. coli, autoclaving, and then streaking out that media two days later and looking for live bacteria. Dr. Anthony Orth -OR- Brendan Smith
Additional supply information:
Genetix - plates and hand replicators
384 and 96 well, "Q"Bot and QFill2 MWG - RoboPrep 2500, RoboSeq
4204, and THEONYX liquid handling system |
| Copyright ©
2003 Genomics Institute of the Novartis Research Foundation. All Rights Reserved. |