Temperature-sensitive (ts) lethal mutants are valuable tools to identify and analyze essential functions. Here we describe methods to generate and classify ts lethal mutants in high throughput.
Systematic identification and characterization of genetic perturbations have proven useful to decipher gene function and cellular pathways. However, the conventional approaches of permanent gene deletion cannot be applied to essential genes. We have pioneered a unique collection of ~70 temperature-sensitive (ts) lethal mutants for studying cell cycle regulation in the unicellular green algae Chlamydomonas reinhardtii1. These mutations identify essential genes, and the ts alleles can be conditionally inactivated by temperature shift, providing valuable tools to identify and analyze essential functions. Mutant collections are much more valuable if they are close to comprehensive, since scattershot collections can miss important components. However, this requires the efficient collection of a large number of mutants, especially in a wide-target screen. Here, we describe a robotics-based pipeline for generating ts lethal mutants and analyzing their phenotype in Chlamydomonas. This technique can be applied to any microorganism that grows on agar. We have collected over 3000 ts mutants, probably including mutations in most or all cell-essential pathways, including about 200 new candidate cell cycle mutations. Subsequent molecular and cellular characterization of these mutants should provide new insights in plant cell biology; a comprehensive mutant collection is an essential prerequisite to ensure coverage of a broad range of biological pathways. These methods are integrated with downstream genetics and bioinformatics procedures for efficient mapping and identification of the causative mutations that are beyond the scope of this manuscript.
Phenotypic characterization of systematic mutant collections of model organisms is a proven approach for dissecting cellular complexity. The haploid unicellular green algae Chlamydomonas reinhardtii has a plant-like gene set, but it diverged from land plants before multiple genome duplications in the land plant lineage2. In principle, lack of gene duplication and a mainly haploid life cycle greatly facilitates loss-of-function genetic approaches. However, targeted disruption of genes of interest is nearly impossible due to the lack of efficient homologous genomic integration. A random insertional disruption library is under construction, combined with identification of the disrupted site, so far yielding an arrayed set of 1,935 mapped disruptions representing 1,562 genes3. However, this approach (expected in general to produce null mutations) is not applicable to essential genes. Temperature-sensitive (ts) mutations can be recovered in essential genes, and recent methods allow efficient identification of the mutated gene and the causative lesion. Phenotypic analysis at high temperature then provides immediate information about the function of the mutated gene. We reported on the isolation and characterization of ts lethal mutations in ~70 essential genes in Chlamydomonas, focusing especially on genes involved in cell cycle progression and control1,4.
Ts lethal screens have been a mainstay of genetic analysis in microorganisms for decades5,6. In principle, a desirable feature is to approach "saturation," meaning that all genes capable of mutating to ts lethality are identified by at least one mutant, allowing a complete analysis. However, in practice, several factors limit the approach to saturation. First, while almost all genes can be mutated to loss of activity at high temperature, the efficiency of recovery of such mutants varies over at least an order of magnitude7,8. Therefore, a random screen begins to pick up recurrent hits in "frequent flyers" long before saturation is approached. Second, while ts mutations usually result in reduction of function, they may not be true nulls at a restrictive temperature (and conversely, are frequently not fully functional at a permissive temperature). This problem can be dealt with to some extent by comparing multiple alleles; if they all share a common phenotype, this is more likely to reflect the result of simple inactivation of the gene. Multiple alleles are also very helpful for definitive molecular identification of the causative lesion1. However, the "frequent flyer" problem means that multiple alleles in rarely hit genes can be difficult to recover.
For these reasons, we have been developing an enhanced pipeline to isolate and phenotypically characterize ts mutants. We have collected over 3,000 ts mutants so far, including about 200 new candidate cell cycle mutations. Molecular and phenotypic analysis of this collection, which already likely includes mutations in most or all cell-essential pathways, should provide new insights and hypotheses in plant cell biology. Importantly, this pipeline can be applied to any microorganism that grows on agar to efficiently construct ts mutant collections.
A note on equipment: two robots are very important for efficiency in this procedure (a colony picker and a combination replica plater/cherry picker). Pickers typically have metal pins. Picked colonies are held in air on these pins for some period (seconds to minutes, depending on the model). Chlamydomonas dies in about 20 sec on a metal pin in air. This restricts model choice for the organism. Accuracy matters. Our colony picker is reasonably accurate; however, it is somewhat violent in its action and has some possibility for spray-based cross-contamination. It is not used at higher than the 384 density (4.5 mm center-center spacing); at this density, the accuracy is completely acceptable. The replica plating/cherry picking robot (different attachments used for these applications) is much slower at picking, but is accurate enough for the 1,536 density (6,144 is a challenge, even for this very accurate robot; we have evaluated this density but elected not to use it because of its various difficulties). The robot will not work well if plates are loaded unevenly, etc. It is important to spot-check to be sure the right things are happening; of course, the robot will run unattended and in general, all will go well if the first few plates were correct.
1. UV Mutagenesis
2. Identification of ts Mutant Candidates: First Screen
3. Identifying ts Mutants: Second Screen
4. Initial Phenotype Determination
5. Complementation and Linkage Testing of New Mutants to "Frequent Flyers"
We show an accelerated pipeline for isolating ts mutants in Chlamydomonas. Cells are dispensed on agar plates, and shortly after quick validation under the microscope for single-cell density, plates are UV irradiated (Figure 1). Typical irradiated colonies are identified and picked into an arrayed format after 10 days of growth at a permissive temperature (Figure 2). The resultant plates in the 384 format are merged to a 1,536-array (Figure 3). From these first steps of collecting irradiated colonies, we have arrayed ~ 200,000 colonies so far. The density of the suspension is adjusted based on the planned UV dose so that, accounting for death, 200 – 600 survivor colonies will form on the plates. Three UV exposure times (1.5 min, 1 min, and 0.5 min) and three densities were tested accordingly (Table 1). Empirically, the 1.5 min exposure time yielded most of the ts- mutants (~ 50%); however, by far, 1 min yielded most of the cell cycle candidates. Therefore, future UV mutagenesis rounds were done with 1 min only. An important concern is the splashing during initial picking and the cross contamination (especially in the high-density formats). This may result in duplication and further phenotyping of the same mutant twice. To minimize the probability for such a scenario, take care to collect colonies at the optimal size, and make sure that adjacent colonies are not picked in the initial assay.
Next, two sequential ts- phenotype assays (Figure 5) were performed, and around 3,000 ts- mutants were isolated and phenotypically characterized by time-lapse microscopy (Figure 8). Due to the biological focus in studying the cell cycle of interest in phenotypes that meet certain criteria, we are not interested in collecting more than two alleles in each target gene. Therefore, we performed complementation and linkage testing with already characterized genes with more than two alleles (Query) against newly collected candidates to remove highly recurrent genes from the downstream pipeline (Figure 9).
Figure 1: Uniform Spreading of Unmutagenized Cells and UV Mutagenesis. (a) 384 x 2 µl drops are dispensed on an agar plate using a reagent dispenser. (b) Random plates are tested under the microscope to verify single-cell density and are UV irradiated with a 1-min exposure. Scale bar = 50 µm. Please click here to view a larger version of this figure.
Figure 2. UV-mutagenized Colonies Picked for the 384-arrays. (a) UV-irradiated single cells are grown for ~10 days into visible colonies. Scale bar = 2 cm. (b) Image analysis program recognizes separable colonies according to certain parameters and robotically arrays them into 384 plates. Please click here to view a larger version of this figure.
Figure 3. Merging to the 1,536-array. Four 384-format plates are merged to one 1,536-plate; once grown, they are replicated again to test for the ts- phenotype. Please click here to view a larger version of this figure.
Figure 4. Imaging for Quantification. (a) Plates are held in a frame under a document photography stand so that all plates in a series are photographed at identical magnification and positioning. (b) The first image is of a 1536-well microtiter plate with red dots placed at corners and midpoints. This "grid-plate" image defines the position of the array. (c) Example of a 1,536-array after incubation. Please click here to view a larger version of this figure.
Figure 5. Identification of the ts- phenotype by Semi-automated Image Analysis. Plate images are analyzed by custom MATLAB software (provided in the S.I.). The image is automatically segmented into cell arrays (96, 384, or 1,536), and the signal in each cell is quantified. Growth at 21 °C is marked in blue, and growth at 33 °C is marked in yellow. Colonies exhibiting significantly higher growth at 21 °C compared to 33 °C (standardized to ts+) are selected and marked in black squares with a green dot. These choices can be edited manually. Final selections are transferred to a file used by the colony-picking robot. Please click here to view a larger version of this figure.
Figure 6. Robotic Picking of First-round ts Lethal Candidates. The cherry-picking robot follows instructions from a file generated as described in step 2.1 to pick candidates for a new array. The top panel shows the loading of a sterilized pin. The bottom panel shows the precise picking from a certain colony in the source plate. Please click here to view a larger version of this figure.
Figure 7. Time-lapse Screening for Initial Sorting of ts Lethal Phenotypes. Clones that passed two rounds of the screening protocol illustrated in Figure 5 are replicated to plates at a cell density such that individual cells can be resolved microscopically. A modified tetrad dissection microscope is used to identify positions at which photomicrographs are taken after 0 hr, 10 hr, and 20 hr incubations at 33 °C. Please click here to view a larger version of this figure.
Figure 8: The Chlamydomonas Cell Cycle Mutants Identified by Time-Lapse Microscopy. (a) Illustration of the Chlamydomonas unique cell cycle describes the long G1 phase characterized by cellular growth, followed by fission S-M cycles, which end up with the hatching of newborn daughter cells. (b) Microscopic images demonstrate WT cells during the cell cycle and the identification of typical cell cycle mutants that fail to complete mitosis. Scale bar = 50 µm. Please click here to view a larger version of this figure.
Figure 9: Complementation and Linkage Test. (a) An antibiotic resistant (e.g., Hygro) query with a frequently repeated mutated gene is crossed in a 96-well plate with a set of mutants of the opposite mating type that harbor a second antibiotic resistance cassette (e.g., Paro). Complementation plates yield high fraction of diploids as shown by the nucleic acid staining and analysis in flow cytometry (lower left panel in a). (b) The mating mixture is tested both for complementation in diploids and for linkage in double-resistant meiotic progeny. The positive control is the query itself in the opposite mating type, and, as expected, shows the ts- phenotype at 33 °C, whereas the negative control is WT and shows the ts+ phenotype. The circled colonies show no complementation with the query and are therefore new ts- alleles for the query gene. These are generally excluded from further characterization, since only "frequent flyers" are in the complementation test set. Please click here to view a larger version of this figure.
Time | O.D. | Colonies picked (#) | Ts- phenotype | Cell-cycle candidates |
1.5’ | 0.012 | 6000 (Avg=~200) | 200 (3.3%) | 7 (3.5%) |
1’ | 0.003 | 11,000 | 131 (1.19%) | 15 (11.4%) |
(Avg=~360) | ||||
0.5’ | 6E-04 | 9000 | 40 (0.45%) | 1 (2.5%) |
(Avg=~300) |
Table 1: Calibration of Irradiation Times to Maximize the Yield of Cell Cycle Mutant Candidates. Three irradiation times were tested (30 plates each) with corresponding ODs to ensure surviving colony numbers (200 – 600).
The pipeline described here for high yield isolation of ts lethal mutants ensures that presumably all cellular-essential pathways of the Chlamydomonas genome are represented. The two most critical steps for efficient collection of potential cell cycle genes and for the elimination of repetitive "frequent flyer" alleles are: 1) the coherent definition of arrest phenotype characteristics for incomplete cell cycles and 2) the parallel complementation assay against already-identified query genes to enlarge the collection with newly isolated ones.
When synchronized by light-dark cycles, Chlamydomonas grows photosynthetically during daylight hours and increases in cell size > 10x without any DNA replication or cell division13. Approximately coincident with the onset of night, cells then undergo multiple cycles of alternating DNA replication, mitosis, and cell division (Figure 8). This regulatory scheme provides a natural distinction between genes primarily required for cell growth and integrity and genes required specifically for the cell division cycle. We found that the 10-hr and 20-hr time points are very informative for an initial rough phenotypic cut1. The broad classes of ts lethal mutants that we recognize currently, based on these images (see S.I. in Tulin and Cross, 2014)1, are: Notch, Popcorn, Round, Small, Medium, early lysis, and multiple-cycle (Figure 8).
The three most relevant categories we focus on are Notch, Popcorn, and Round. The "Notch" and "Popcorn" phenotypes were shown previously to be characteristic of most cell cycle-specific lesions (e.g., mitotic cyclin-dependent kinase, DNA replication machinery, and Topoisomerase II)1. The appearance of one (Notch) or multiple (Popcorn) apparent planes of incipient but unsuccessful cell division is a convenient morphological indicator of cell cycle initiation. These mutants generally exhibit little or no growth defects, with increases in cell volume similar to WT at the 10-hr mark. The Notch and Popcorn phenotypes are evident at 10 hr and are fully developed (frequently associated with cell lysis) by 20 hr. "Round" cells grow similarly to WT but with much-reduced production of apparent incipient division planes, thus yielding large, round arrested cells. Previous mutants in this category have fallen into components of the anaphase-promoting complex14 or in genes required for microtubule function (tubulin-folding cofactors, gamma-tubulin ring complex)1. At later times, these cells frequently exhibit pronounced cell lysis.
"Small" and "Medium" cells grow either negligibly (Small) or significantly less than WT (Medium). Many of these mutants identified to date have lesions in genes whose annotations suggest roles in basic cellular growth processes (translation or membrane biogenesis). The main microscopic discrimination between Medium and Round rests on the amount of growth at 10 hr (Round: like WT; Medium: reduced). Because the Small and Medium categories are quite large and probably reflect lesions in a great range of cellular pathways, we are not attempting to saturate these categories; however, we do want to molecularly identify representatives of the class to understand phenotypes of loss in diverse pathways. Two unstudied categories are: 1) the early-lysing mutants that lose integrity (loss of green color, loss of refractility) by the 10-hr mark, with little evidence of prior cell growth and 2) the multiple cycles. Cells proliferate similarly to WT at 10 and 20 hr, though they exhibit a complete inability to carry out longer-term proliferation.
We are mostly characterizing "notch," "popcorn," and "round" and exclude small and medium round cells, as well as leaky mutants that complete few cell divisions. This is principally to ensure that basic cellular features, such as growth and membrane integrity, are functional and enrich the probability for division-related genes. This approach is found to be empirically efficient; however, it may be that a cell cycle gene is pleiotropic and has additional roles earlier in G1, before actual division. Such cases, which we expect to be rare, are missed. More generally, we aim for homogenous arrest, which in high probability is due to one causative mutation that is a completely dysfunctional protein. However, for the same reason as just described, there may be several arrest points and therefore, some flexibility is advisable in choosing candidates.
In order to enrich the collection with newly identified genes, the chosen candidates are assayed for complementation. We require ts- in the positive control (query against query mutation) and ts+ in the negative control (query against WT). New mutants in the same complementation group as the query are ts-. Membership in the same complementation group almost invariably reflects a molecular lesion in the same gene (this has been the case for every such gene we have tested). Therefore, for "frequent flyers," this criterion is exclusionary for further characterization. Mutants that were not at the same complementation groups as the tested queries are candidates for new genes and are further characterized by bioinformatics and experimental tools. Highly variable recovery of ts- alleles into different complementation groups is a well-known phenomenon-that is, variability is markedly greater than Poisson noise, due to the great intrinsic variability of mutability to ts- between different genes. Causes could include intrinsic thermolability differences; different protein sizes; the presence of a protein as a monomer versus as a large, stabilized complex; and mutagenic hot spots. This is almost a pure nuisance. However, one resulting favorable outcome is that the "frequent flyer" list is not long (with only a few targets occupying most of the list), so the labor-intensive complementation testing is not a massive undertaking until the later stages of the project.
As a complementary approach, we performed a linkage assay. In this assay, double-resistant progenies are selected and are tested for the ts- phenotype. A ts- phenotype is expected (and observed) for mutants in the same complementation group as the query or for tightly linked mutations. For each of the tested genes, a WT progeny is expected to appear (ts+ phenotype) in a certain probability depending on the genetic distance. We estimate that there are around 100 zygospores for each mating in these spots. Assuming 100% meiotic efficiency, this will result in around 100 double-drug-resistant progeny from the unlinked drug resistance cassettes (25% of the meiotic progeny, four per meiosis, due to Mendelian inheritance). This would also be the case for ts- mutations, where 25% of the progenies will be double-mutant, and 25% will be WT if the query and test mutations are unlinked. Therefore, out of the double-drug-resistant progeny, 25% will be WT (around 25 cells). This is the case for completely unlinked mutations; however, moderate linkage (within ~ 20 cM, about 2 Mb, or 2% of the genome115) will strongly reduce or eliminate the ts+ signal. In the case of linkage of the tested mutation to the antibiotic cassette, ts+ haploids that are double-drug-resistant are present in very low amounts. This manifests as apparent failure to recombine with all mutants tested, despite complementing all mutants tested, an aberrant result that is easily noted; in such cases, backcrossing will solve the problem.
Both from prior knowledge and from sequence analysis, we expect cell cycle genes in the Chlamydomonas to be around 500 genes2, although most, but probably not all, are essential. We will evaluate the necessity for additional mutagenesis rounds as more mutants are collected and the level of saturation rises.
This procedure is uniquely designed for studying essential biological processes and the genes and proteins that carry them out. Other methodologies to generate perturbations in essential genes exist (e.g., transformation of randomly mutagenized alleles16, conditionally transcribed alleles17, or hypomorphic alleles18). However, they all require homologous recombination, which is strongly suppressed in vegetative Chlamydomonas. The clustered, regularly interspaced short palindromic repeat (CRISPR)/Cas9 system has been established as a powerful tool for gene modification19; however, it is yet to work efficiently in Chlamydomonas20. Critically, all of these methods require prior knowledge of the target. This is a severe restriction if one wishes to have the possibility to learn something new! Our approach will yield mutations identifying essential genes, independent of any prior knowledge. Therefore, at the present level of technology, isolation of random ts mutations followed by gene identification by deep sequencing may be the most efficient method of gaining rapid entry into microbial cell biology in the plant superkingdom.
Identification of causative mutations (from among ~100 coding-sequence-changing mutations in each clone) is beyond the scope of this paper. Deep sequencing of bulked segregant pools1 is effective but labor-intensive. A combinatorial pool strategy for the determination of all mutations in a large number of strains, after sequencing a small number of pools, is very cost- and labor-effective. A new strategy for combinatorial bulked segregant sequencing is under development that will allow the identification of causative mutations in dozens of mutants simultaneously in a single sequencing run (in preparation). These efficiencies are very important to allow the critical gene identification step to keep pace with the very rapid accumulation of mutants that is made possible by the procedures described here.
The authors have nothing to disclose.
We thank the Cross lab members for advice and useful discussion. This work was supported by PHS 5RO1-GM078153 and by a Junior Fellow award from the Simons Foundation to Michal Breker.
Equipment: | |||
Hudson RapidPick colony picker | Hudson Robotics | ||
MultiDrop Combi Reagent Dispenser | Thermo Scientific | 5840300 | |
Small tube metal tip cassette | Thermo Scientific | 24073295 | |
Singer RotoR replica-plating robot | Singer Instruments | Very essential for the process. For lower scale screenings you can use an in-house manual tool | |
Singer single-colony picking attachment (‘Stinger’) | Singer Instruments | Can be picked manually, however for large scales it is nearly impossible | |
Name | Company | Catalog Number | Comments |
Materials: | |||
SYTOX Green Nucleic Acid Stain | ThermoFisher Scientific | S7020 |