Here, we describe a methodology to perform a targeted siRNA “ubiquitome” screen to identify novel ubiquitin and ubiquitin-like regulators of the HIF1A-mediated cellular response to hypoxia. This can be adapted to any biological pathway where a robust read out of reporter activity is available.
Post-translational modification of proteins with ubiquitin and ubiquitin-like molecules (UBLs) is emerging as a dynamic cellular signaling network that regulates diverse biological pathways including the hypoxia response, proteostasis, the DNA damage response and transcription. To better understand how UBLs regulate pathways relevant to human disease, we have compiled a human siRNA “ubiquitome” library consisting of 1,186 siRNA duplex pools targeting all known and predicted components of UBL system pathways. This library can be screened against a range of cell lines expressing reporters of diverse biological pathways to determine which UBL components act as positive or negative regulators of the pathway in question. Here, we describe a protocol utilizing this library to identify ubiquitome-regulators of the HIF1A-mediated cellular response to hypoxia using a transcription-based luciferase reporter. An initial assay development stage is performed to establish suitable screening parameters of the cell line before performing the screen in three stages: primary, secondary and tertiary/deconvolution screening. The use of targeted over whole genome siRNA libraries is becoming increasingly popular as it offers the advantage of reporting only on members of the pathway with which the investigators are most interested. Despite inherent limitations of siRNA screening, in particular false-positives caused by siRNA off-target effects, the identification of genuine novel regulators of the pathways in question outweigh these shortcomings, which can be overcome by performing a series of carefully undertaken control experiments.
Modification of proteins with ubiquitin and ubiquitin-like molecules (UBLs) represents an expansive biochemical system that regulates diverse biological pathways and stress responses. The covalent attachment of UBLs to their target proteins can have various outcomes regulating the stability, localization, function or interactome of the substrate1. The enzymatic steps underlying UBL modification were first established for ubiquitin, and now serve as a paradigm for modification with most UBLs, including SUMO, NEDD8, ISG15 and FAT10. For modification to occur, the carboxylate group of the UBL diglycine motif is first activated by an E1 activating enzyme to form a high-energy thiol that is transferred to the active-site cysteine of an E2 conjugating enzyme. The E2 then interacts with a substrate-bound E3 ligase to mediate transfer of the UBL onto (usually) a target lysine residue creating a branched chain (isopeptide) linkage2. Successive rounds of modification can occur to build isopeptide chains onto the substrate, which for ubiquitin can occur through any of its seven lysines, or through its N-terminal methionine to create linear ubiquitin chains. These modifications form discrete topologies with diverse purposes such as creating new interaction motifs and targeting proteins for degradation prior to UBL removal by specialist proteases. In the case of ubiquitin there are two E1 enzymes, 30-40 E2 conjugating enzymes, at least 600 E3 ligases and approximately 100 deubiquitylating enzymes (DUBs). While the pathways are less expansive for the other 10 or so UBLs, the overall ubiquitome complexity affords huge diversity in the biological outcome of a particular UBL modification. However, while major advances in UBL biology have been made, the precise cellular roles of the majority of these ubiquitome components remain unknown.
The use of short interfering ribonucleic acid (siRNAs) has emerged as a powerful tool in reverse genetics due to the ability of siRNAs to specifically target cellular mRNAs for destruction, allowing the role of individual genes to be examined in different biological contexts3. Whole genome screens have been used to identify and validate new regulators of many cellular processes, and have created a wealth of useful data accessible to the wider scientific community. However, while whole genome screens have proven extremely useful, targeted screens are becoming increasingly popular as they are cheaper, faster, involve less data management and report only on members of the genome in which the investigator is most interested. Therefore, to better understand which cellular processes UBL family components are involved in, we have compiled a human siRNA library targeting all known and predicted components of the ubiquitome. This includes the UBLs, E1 activating enzymes, E2 conjugating enzymes, E3 ligases, ubiquitin-binding domain (UBD)-containing proteins and DUBs. This library can be used to screen against a wide range of reporter cell lines of distinct biological problems, thus allowing the unbiased identification of novel UBL components governing these pathways.
The following protocol describes how to perform a rigorous targeted siRNA ubiquitome screen to identify novel regulators of the HIF1A-dependent response to hypoxia. Under normal oxygen tension, HIF1A is subject to prolyl hydroxylation that causes it to be recognized and targeted for degradation by the Von Hippel Lindau (VHL) E3 ligase complex4. Hypoxia inhibits prolyl hydroxylation leading to the stabilization of HIF1A and its subsequent binding to hypoxia response elements (HREs) to drive gene expression. Here, we describe a screen using U20S osteosarcoma cells stably expressing firefly luciferase under the control of three tandem copies of the Hypoxia Response Element (U20S-HRE cells)5. This protocol can be adapted for any biological pathway if a robust read-out of reporter activity is achievable and can be coupled with appropriate positive and negative controls.
1. Assay Development Stage
Note: prior to initiating the siRNA screen, an assay development stage is critical to set out important parameters for screening with the reporter cell line. It is essential to invest significant effort at this stage as this will underpin the future success of the screen.
2. Primary Screen
Note: once these basic conditions from the assay development stage are in place, the primary screen can be performed in triplicate in 96 well-plate format using the following protocol.
3. Secondary Screen
Note: a confirmatory secondary screen is carried out based on a maximum of 80 siRNAs of interest from the primary screen. This number can be conveniently carried out with each replicate plated on a single 96 well plate with full controls as per the primary screen (see 2.2). It is very useful to confirm that the regulators identified in the primary screen reproducibly elicit the same phenotype, and this will assist in refining the triage decisions on which hits should be carried through to the final tertiary/deconvolution screen.
4. Tertiary/Deconvolution Screen
Note: tertiary or deconvolution screening is performed on a maximum of 20 siRNAs from the secondary screen. This step is to examine the effect of knockdown using each individual siRNA from the original pool of four. Normally, at least two individual siRNA duplexes from each pool should illicit the same phenotype to have a reasonable degree of confidence that the observed phenotype is not due to siRNA off-target effects. To further increase confidence at this stage, additional individual siRNA duplexes targeting the gene in question can be designed and tested for their ability to elicit the given phenotype. Results from these experiments may then be used to calculate the H score, where H = 0.6 or over (i.e. where at least 3 out of 5 individual siRNAs elicit the phenotype) is considered acceptable6.
Note: Ideally the threshold for individual siRNA duplexes should be set at the same stringency cut-off as for the pool. However, it may be acceptable to relax the threshold by 10-20% for individual siRNAs, especially where at least one other duplex falls within the set threshold. It is worthwhile bearing in mind that the individual effect of siRNAs may be less than that of the pool, and conversely, in some cases individual siRNAs may show a stronger effect on the cellular phenotype in isolation than when it exists in the pool (even when concentration is accounted for).
Prior to screening, the hypoxia-responsiveness of U20S-HRE cells is established. U20S-HRE cells express a reporter construct consisting of firefly luciferase fused downstream of three tandem copies of the hypoxia response element, which is bound by the HIF1A/HIF1B heterodimer upon exposure to hypoxia (Figure 1A). Cells are placed in a hypoxia workstation for a range of times to establish which hypoxia exposure produces the most effective response for screening. U20S cells express low levels of HIF2A, therefore luciferase readings are largely dependent on hypoxia-mediated HIF1A stabilization7. Exposure of U20S-HRE cells to hypoxia for 0 hr, 2 hr, 4 hr, 6 hr and 24 hr results in a hypoxia-dependent induction of luciferase activity (Figure 1B). Exposure of U20S-HRE cells to hypoxia for 24 hr results in a ~10 fold induction of activity (Figure 1B), therefore this represents an excellent response for siRNA screening.
To establish suitable high and low controls, siRNAs to HIF1A and FIH1 are used to reverse transfect U20S-HRE cells. Cells are transfected with the controls for 24 hr and exposed to hypoxia for a further 24 hr before luciferase assays are performed. HIF1A siRNA diminishes the hypoxia signal to around 10% whereas the FIH1 siRNA enhances the hypoxia signal to around 170% relative to non-target siRNA (Figure 2). These controls result in an average Z factor of 0.8, which is an excellent score for screening. The screening parameters have now been successfully established for a robust screen.
The primary screen is carried out in 96-well plate format using the workflow demonstrated in Figure 3. Control and library siRNAs are stamped onto assay plates (Figure 3A), and U20S-HRE cells are reverse transfected then incubated at 37 °C for 24 hr. Assay plates are then exposed to hypoxia for 24 hr (Figure 3B) before luciferase assays are performed and the data analyzed (Figure 3C). Quality control analysis of the screen includes determining the Z factor for each plate, where a Z factor above 0.5 is desirable8, 9. The Z factor is calculated for each plate of the triplicate primary screen using the formula in Figure 4A. The Z factors for each plate are visualized using a scatter plot, where it can be seen each plate has a Z factor greater than 0.5 (Figure 4B). Calculating the Z factor is important as it reports on how good a separation window exists between controls and can highlight potential problems with individual screen assay plates. The CV of the high and low controls are also calculated, and the averages are 5.95% +/- 0.4 and 8.04% +/- 1.2 respectively. The primary screen results in a distribution of siRNA duplexes forming a continuum between the low and high controls. Typically, there will be approximately equal numbers of positive and negative regulators, with the majority of siRNAs having no effect on the reporter (Figure 5). Regulators of interest are taken through to secondary then tertiary screening to establish a list of high-confidence regulators for follow-up analysis and validation.
Figure 1. Hypoxia exposure induces U20S-HRE luciferase reporter. (A) Schematic representing the hypoxia reporter construct in U20S-HRE cells used for siRNA screening. Three tandem copies of the hypoxia response element (HRE) are bound by the HIF1A/HIF1B heterodimer to drive the expression of firefly luciferase. (B) Increasing exposure of U20S-HRE cells to 2 hr, 6 hr, 10 hr and 24 hr to hypoxia results in increasing amounts of reporter luciferase activity. 24 hr hypoxia exposure results in ~10 fold increase in luciferase activity compared to 0 hr control (normoxia). Scale bars represent the standard error of the mean.
Figure 2. Establishing high and low siRNA controls for screening. U20S-HRE cells reverse transfected with siRNA to HIF1A (low control) and FIH1 (high control) decrease and increase the response to 24 hr hypoxia exposure respectively. These controls give a Z factor of 0.8, which is considered an excellent score for screening. The effect of non-target (NT) siRNA is shown for comparison. Error bars represent the standard errors of the mean.
Figure 3. Screening workflow to identify regulators of the HIF1A hypoxia response. (A) Arrangement of 96 well plates for siRNA screening. Control siRNAs are stamped onto the outside columns as indicated and the ubiquitome siRNA library is stamped onto the internal columns, spread over 17 plates. U20S-HRE cells are reverse-transfected and incubated at normoxia for 24 hr. (B) Stacks of assay plates are transferred in a sterile environment to a hypoxia workstation set at 1% oxygen for 24 hr. (C) Plates are removed from the hypoxia workstation and luciferase activity assays are performed on all assay plates. Luminescence is recorded on a luminometer and data analysis performed to establish the reporter activity of cells transfected with each library siRNA.
Figure 4. Quality control is performed by calculating the Z factor of each plate. (A) The Z factor for each plate is calculated using the formula given, where s represents the standard deviation, m represents the mean, p the positive control and n the negative control. (B) Scatter plot displaying the calculated Z factor of each plate from the three replicates of the ubiquitome screen. A cut-off of 0.5 is applied, and in this case all plates display a Z factor > 0.5 and therefore pass the quality control.
Figure 5. Chart showing the distribution of the siRNA library. Chart displaying the frequency of siRNAs from the triplicate ubiquitome siRNA screen that result in the corresponding percent activation. Low and high controls are set to 0% and 100% respectively. The siRNA library is distributed evenly between the high and low control, whereby the majority of siRNAs have no effect on the cellular phenotype. A small proportion of siRNAs are observed to increase and decrease the response, and therefore represent the potential novel regulators of the pathway.
The use of genome-wide siRNA screens in mammalian cells has proven to be extremely valuable in identifying novel regulators of distinct biological pathways. Here, we have described the use of a targeted ubiquitome siRNA screen to identify regulators of the HIF1A-mediated cellular response to hypoxia. Targeted screens are becoming increasingly attractive as they are generally cheaper, quicker, easier to manage and report only on the pathway components in which the investigators are interested 7, 10, 11.
It is critical to put a major effort into the assay development stage to ensure the basic cell reporter is suitable for screening, and that reproducibility and low-variability can be achieved. It is also important to select robust positive and negative controls to create a window of separation within which the library regulators will fall. In addition, these act as an internal control in each assay plate of the screen allowing for the rapid identification of any problems with individual plates.
One common problem in high throughput screening is the occurrence of plate edge effects, which can be caused by a local difference in humidity at the outside wells of the plate relative to the internal wells. Having the appropriate controls in place and creating a heat map of the plate output can help to identify this problem. We have found that edge effects can be overcome by using moist tissue paper at the base of the plates, and placing a stiff, transparent plastic bag over the stack of plates to make the microenvironment of each plate inside the incubator more even.
A limitation of siRNA screening is the presence of false-positives and false negatives. False negatives can occur due to a number of factors including insufficient knockdown of the corresponding mRNA, or the ability of residual mRNA to produce enough active protein to efficiently perform its function. False-positives are more problematic, as they can cause an investigator to invest significant time following up a hit that may not actually govern the pathway in question. False-positives can be generated through a number of mechanisms including via siRNA off-target effects. This is where the siRNA knocks down not only the mRNA in question but also other, unidentified targets that are the true pathway regulators12, 13. A number of approaches are commonly used and accepted as confirmation that the observed phenotype is a true result of knocking down the target gene in question14. For example, the problem of off-target effects can be overcome in part by tertiary screening, whereby the likelihood of more than one individual siRNA from the starting pool of four having the same off-target effect is decreased. In addition, successfully performing rescue experiments with siRNA-resistant cDNAs corresponding to the hit will also rule out off-target effects. It should be noted that this approach in itself is not without caveats, for example exogenously overexpressed proteins will not necessarily act identically to the endogenous counterpart due to altered stoichiometry of the relevant complexes or perturbed post-translational modifications etc. An alternative approach to hit validation therefore is to generate or acquire knockout or knock-in cell lines corresponding to the hits, and determine if these cell lines display the same phenotype as the siRNA knockdown. Thus, limitations of siRNA screening can be overcome with careful experimental follow-up. Furthermore, the unbiased identification of genuine modifiers of the pathways under scrutiny far outweighs the problems associated with potentially identifying false positives.
In summary, siRNA screening with a targeted “ubiquitome” siRNA library is a powerful method to identify novel members of ubiquitin and ubiquitin-like modifiers regulating distinct biological pathways. The protocol described here can be applied to any biological problem if there is a robust read-out of reporter activity. The final list of high confidence hits from the screen represents the starting point from where to establish the mechanistic basis of how each hit governs the pathway in question. Ultimately, this will add to the growing knowledge of how ubiquitin and ubiquitin-like modifiers regulate distinct biological pathways.
The authors have nothing to disclose.
This work was supported by The Wellcome Trust, Glaxosmithkline (GSK) and the Scottish Institute for Cell Signalling (now part of the MRC Protein Phosphorylation and Ubiquitylation unit).
Automated Liquid Dispenser | Fluid-X | XPP-721 | http://www.fluidx.eu/BIOTRACK/xpp-721-liquid-handling-system.html |
Automated cell counter | Nexcelom Bioscience | Cellometer Auto T4 | http://www.nexcelom.com/Cellometer-Auto-T4/index.html |
Hypoxia Workstation | Ruskin | In vivo2 300 | http://www.ruskinn.com/products/invivo2300 |
Automated Cell Dispenser | Thermo Scientific | Matrix Wellmate | http://www.matrixtechcorp.com/automated/pipetting.aspx?id=11 |
Plate Shaker | Heidolph | Titramax 1000 | http://www.heidolph-instruments.co.uk/products/shakers-mixers/platform-shakers/vibrating/titramax/titramax-1000/ |
Luminometer | Perkin Elmer | Envision 2104 Multilabel Reader | http://www.perkinelmer.co.uk/Catalog/Family/ID/EnVision%20Multilabel%20Plate%20Readers |
White Walled Assay Plate | Greiner Bio One | 655083 | http://www.greinerbioone.com/en/row/articles/catalogue/article/37_11/13221/ |
Clear Plate Film | Perkin Elmer | 1450-461 | http://www.perkinelmer.co.uk/Catalog/Product/ID/1450-461 |
Name of the Reagent | |||
siRNA library | Thermo Scientific | On-Target Plus | http://www.thermoscientificbio.com/rnai-and-custom-rna-synthesis/sirna/on-targetplus-sirna/search-gene/ |
Transfection reagent | Invitrogen | Lipofectamine RNAimax | http://www.invitrogen.com/site/us/en/home/Products-and-Services/Applications/Protein-Expression-and-Analysis/Transfection-Selection/lipofectamine-rnaimx.html |
Reduced Serum Medium | Invitrogen | Optimem | http://products.invitrogen.com/ivgn/product/31985062?ICID=search-product |
DMEM | Invitrogen | 41965-039 | http://products.invitrogen.com/ivgn/product/41965039# |
FBS | Invitrogen | 16000-044 | https://products.invitrogen.com/ivgn/product/16000044?ICID=search-product# |
Tryspin-EDTA | Invitrogen | 25300-054 | https://products.invitrogen.com/ivgn/product/25300054?ICID=search-product# |