Here, we describe a method for the bacterial co-expression of differentially tagged proteins using a set of compatible vectors, followed by the conventional pulldown techniques to study protein complexes that cannot assemble in vitro.
Pulldown is an easy and widely used protein-protein interaction assay. However, it has limitations in studying protein complexes that do not assemble effectively in vitro. Such complexes may require co-translational assembly and the presence of molecular chaperones; either they form stable oligomers which cannot dissociate and re-associate in vitro or are unstable without a binding partner. To overcome these problems, it is possible to use a method based on the bacterial co-expression of differentially tagged proteins using a set of compatible vectors followed by the conventional pulldown techniques. The workflow is more time-efficient compared to traditional pulldown because it lacks the time-consuming steps of separate purification of interacting proteins and their following incubation. Another advantage is a higher reproducibility due to a significantly smaller number of steps and a shorter period of time in which proteins that exist within the in vitro environment are exposed to proteolysis and oxidation. The method was successfully applied for studying a number of protein-protein interactions when other in vitro techniques were found to be unsuitable. The method can be used for batch testing protein-protein interactions. Representative results are shown for studies of interactions between BTB domain and intrinsically disordered proteins, and of heterodimers of zinc-finger-associated domains.
Conventional pulldown is widely used to study protein-protein interactions1. However, purified proteins often do not interact effectively in vitro2,3, and some of them are insoluble without their binding partner4,5. Such proteins might require co-translational assembly or the presence of molecular chaperones5,6,7,8,9. Another limitation of conventional pulldown is the testing of possible heteromultimerization activity between domains that can exist as stable homo-oligomers assembled co-translationally8,10, as many of them cannot dissociate and re-associate in vitro during the incubation time. Co-expression was found to be useful in overcoming such problems3,11. Co-expression using compatible vectors in bacteria was successfully used to purify large multi-subunit macromolecular complexes, inclduing polycomb repressive complex PRC212, RNA polymerase II mediator head module13, bacteriophage T4 baseplate14, SAGA complex deubiquitinylase module15,16, and ferritin17. Replication origins commonly used for co-expression are ColE1, p15A18, CloDF1319, and RSF20. In the commercially available Duet expression system, these origins are combined with different antibiotic resistance genes and convenient multiple cloning sites to produce polycistronic vectors, allowing the expression of up to eight proteins. These origins have different copy numbers and can be used in varying combinations to achieve balanced expression levels of target proteins21. To test protein-protein interactions, various affinity tags are used; the most common are 6xHistidine, glutathione-S-transferase (GST), and maltose-binding protein (MBP), each of which has a specific affinity to the corresponding resin. GST and MBP also enhance the solubility and stability of tagged proteins22.
A number of methods involving protein co-expression in eukaryotic cells have also been developed, the most prominent of which is yeast two-hybrid assay (Y2H)23. Y2H assay is cheap, easy, and allows the testing of multiple interactions; however, its workflow takes more than 1 week to complete. There are also a few less frequently used mammalian cell-based assays, for example, fluorescent two-hybrid assay (F2H)24 and cell array protein-protein interaction assay (CAPPIA)25. F2H assay is relatively fast, allowing to observe protein interactions in their native cellular environment, but involves using expensive imaging equipment. All these methods have an advantage over prokaryotic expression providing the native eukaryotic translation and folding environment; however, they detect interaction indirectly, either by transcriptional activation or by fluorescent energy transfer, which often produces artifacts. Also, eukaryotic cells may contain other interaction partners of proteins of interest, which can interfere with the testing of binary interactions between proteins of higher eukaryotes.
The present study describes a method for the bacterial co-expression of differentially tagged proteins followed by conventional pulldown techniques. The method allows studying interactions between target proteins that require co-expression. It is more time-efficient compared to traditional pulldown, allowing batch testing of multiple targets, which makes it advantageous in most cases. Co-expression using compatible vectors is more convenient than polycistronic co-expression since it does not require a laborious cloning step.
The schematic representation of the method workflow is shown in Figure 1.
1. Co-transformation of E. coli
2. Expression
3. Pulldown assay
NOTE: The detailed procedures are described for proteins tagged with either 6xHis or MBP/GST. All procedures are performed at 4°C.
The described method was used routinely with many different targets. Presented here are some representative results which likely cannot be obtained using conventional pulldown techniques. The first is the study of specific ZAD (Zinc-finger-associated domain) dimerization11. ZADs form stable and specific dimers, with heterodimers possible only between closely related domains within paralogous groups. The dimers formed by these domains are stable and do not dissociate for at least a few days; thus, mixing purified ZADs does not produce any detectable binding. At the same time, the co-expression of MBP and Thioredoxin-6xHis-fused ZADs shows good and reproducible homo-dimerization activity (Figure 2A), appearing as an additional band in the SDS-PAGE results of the MBP-pulldown assay. A small portion of heterodimers can be seen with M1BP co-expressed with another domain; this interaction was not confirmed with Y2A and most likely is a result of non-specific association due to high protein concentration, since these domains are cysteine-rich and extremely aggregation-prone. Notably, in this case, 6xHis-pulldown is inappropriate as ZADs are metal-coordinating domains that non-specifically bind to the metal-chelating resin. Such activity should be carefully examined in a parallel experiment.
Another example is the competition assay between the ENY2 protein and its binding partners Sgf11 (1-83aa) and the zinc-finger domain (460-631 aa) of the CTCF protein26. When expressed alone, the ENY2 protein forms dimers that prevent it from interacting with its native binding partners. Presumably, both the Sgf11 and CTCF proteins bind to the same molecular surface of ENY2, making their interaction mutually exclusive. In the co-expression assay, 6xHis-tagged ENY2 interacted both with GST-tagged Sgf11 and MBP-CTCF, but no GST-Sgf11 was present in MBP-pulldowns and vice versa, (Figure 2B). These results suggest that no triple complex can be formed, and interactions are mutually exclusive. These data were independently confirmed with other assays and support the different functional roles of ENY2 in these complexes. Attention should be brought to the fact that large affinity tags can impose steric hindrances by themselves, preventing complex formation; therefore, the conclusion should not be based solely on co-expression data.
A step-by-step comparison of the workflows of conventional and coupled co-expression pulldowns is shown in Figure 3A. Co-expression coupled pulldown is at least two times more time-efficient, even with a small number of samples, and allows good scalability. The results of utilizing both techniques to study the same interaction between the BTB domain of the CP190 protein (1-126aa) and the GST-tagged C-terminal domain (610-818aa) of Drosophila CTCF protein (dCTCF) are shown in Figure 3B. Both methods show good efficiency and reproducibility (assays were performed in three replicates); for this case, co-expression coupled pulldown showed lower non-specific binding, as can be seen in the control samples with GST protein alone.
Figure 1: The schematic representation of the protocol. The schematic shows the method workflow employed in this study. Please click here to view a larger version of this figure.
Figure 2: Representative results. (A) Study of the Zinc-finger associated domain (ZAD) homodimerization in MBP- and 6xHis-pulldown assays. ZADs fused either with MBP (40 kDa) or with 6xHis-Thioredoxin (20 kDa) were co-expressed in bacteria cells and affinity purified with amylose resin (binds MBP-tagged proteins) or with Ni-NTA resin (binds 6xHis-tagged proteins). Co-purified proteins were visualized with SDS-PAGE followed by Coomassie staining. MBP-pulldown results are shown in upper panels, 6xHis-pulldown results are in lower panels (used only as protein expression control since many ZADs bind non-specifically to Ni-NTA). (B) Study of mutually exclusive interactions between ENY2 and Sgf11/CTCF proteins. GST-tagged Sgf11 (1-81aa), MBP-tagged CTCF zinc-finger domain (460-631aa), and 6xHis-tagged ENY2 proteins were co-expressed in various combinations and affinity purified with amylose resin, glutathione resin (binds GST-tagged proteins), or with Ni-NTA resin.Co-purified proteins were visualized with SDS-PAGE followed by Coomassie staining. The figure in panels A and B have been modified with permission from Bonchuk et al.11 and Bonchuk et al.26. Please click here to view a larger version of this figure.
Figure 3: Comparison of workflows of conventional and coupled co-expression pulldown assays. (A) Step-by-step comparison of required time intervals in the workflow of the conventional pulldown assay compared to pulldown coupled to co-expression. (B) Comparison of two different pulldown techniques in studying the interaction between the dCTCF C-terminal domain (610-818aa) and the BTB domain of the CP190 protein (1-126aa) in GST-pulldown assays. dCTCF (610-818aa) fused with GST (25kDa) or GST alone were either co-expressed in bacteria cells or incubated in vitro with Thioredoxin-6xHis-tagged CP190 BTB and affinity purified with glutathione resin. Three independent replicates of each assay are shown. Co-purified proteins were visualized with SDS-PAGE followed by Coomassie staining. Please click here to view a larger version of this figure.
The described method allows the testing of protein-protein interactions that cannot be efficiently assembled in vitro and require co-expression. The method is one of the few suitable approaches for studying heterodimerizing proteins, which are also capable of homodimerization since, when purified separately, such proteins form stable homodimers which most often cannot dissociate and re-associate during the experiment3,11.
The workflow of the described method is more time-efficient compared to traditional pulldown because it lacks the time-consuming steps of separate purification of interacting proteins and their following incubation. Another advantage is a higher reproducibility, due to a significantly smaller number of steps and a shorter period of time in which proteins exist within an artificial in vitro environment while being exposed to proteolysis and oxidation. The method was successfully applied for studying several protein-protein interactions when other in vitro techniques were found to be unsuitable3,11,27. Parallel co-precipitation with different affinity tags provides protein expression level control. Since the method is easier and faster than conventional pulldown, it can be used instead of it, even if co-expression is not absolutely required. The speed of cell disruption is a non-obvious bottleneck which can significantly increase the total time and can lead to a deviation in results, due to the various amount of time between the first and last samples being exposed to in vitro conditions and proteolysis. Thus, using high-throughput multi-tip sonicator probes is strongly recommended; see the Table of Materials for examples.
Using magnetic beads could decrease the non-specific binding and further speed up the method, reducing the time required for washing steps. Using vectors with compatible origins has an advantage over polycistronic expression as they provide a convenient combinatorial approach to test multiple protein interactions with the same target and do not require producing a new vector for each combination.
A disadvantage of the method is the relatively high probability of false-positive results, as proteins are co-expressed in bacteria at high concentrations. Thus, unexpected interactions discovered by this method should be treated with caution and tested with an independent assay, such as the Y2A or cellular techniques24, which also uses co-expression3,11,28. High-throughput bioinformatic approaches can also be used to analyze and validate complex protein-protein interaction networks29. Another peculiar obstacle is that protein complexes may have completely different biochemical properties compared to separate proteins; the complex might be insoluble while its components are present in the solution. This problem can typically be solved by fusing the proteins to the appropriate solubility tag. MBP is the most effective, while NusA is another good all-around alternative22. GST-tag was found to be efficient with many zinc-coordinating domains, although since it is dimeric, it should be avoided when working with oligomeric domains. On the contrary, small monomeric domains such as Thioredoxin and SUMO work well with multimeric protein domains.
Critical steps of the described method are the proper time of protein expression (a shorter time is required if non-specific binding is observed, while longer incubation times may be required to improve poor protein expression), fast cell disruption, proper choice of buffers, and maintaining the constant temperature during the protocol. Excessive incubation times after induction may result in protein aggregation in bacteria, leading to false positive results. On the other hand, some proteins require more time to be expressed in sufficient quantities. Temperature fluctuations during the sonication and wash steps may lead to changes in buffer pH and protein precipitation. Improper buffer choice also may result in non-specific protein aggregation.
In the case of poor protein expression, different solubility tags should be attempted, as well as an increased incubation time after induction. If a strong non-specific association is observed, all necessary negative controls should be run to determine the possible non-specific association of proteins with resin or the affinity tag. In case the controls do not work, different combinations of affinity tags should be attempted and different buffers used. Many proteins tend to non-specifically bind to the metal-chelating resin-like ZADs mentioned in this article; in such cases, MBP/GST-pulldown, in general, would be sufficient without a reciprocal experiment that can be used only for protein expression control. If negative controls work well, the protein expression time should be decreased, and the buffer system and reducing agent changed or the reducing agent concentration increased, especially when working with cysteine-rich proteins. In the case of poor reproducibility of results, attention should be paid to the cell disruption step, which should be performed fast but without overheating. The temperature should be monitored throughout the whole experiment.
This method can be easily modified to study multiprotein complexes or ribonucleoproteins. Another possible application is the batch testing of protein complex expression and purification conditions for subsequent studies.
The authors have nothing to disclose.
This work was supported by the Russian Science Foundation projects 19-74-30026 (method development and validation) and 19-74-10099 (protein-protein interaction assays); and by the Ministry of Science and Higher Education of the Russian Federation-grant 075-15-2019-1661 (analysis of representative protein-protein interactions).
8-ELEMENT probe | Sonics | 630-0586 | The high throughput 8-element sonicator probes |
Agar | AppliChem | A0949 | |
Amylose resin | New England Biolabs | E8021 | Resin for purification of MBP-tagged proteins |
Antibiotics | AppliChem | A4789 (kanamycin); A0839 (ampicillin) | |
Beta-mercaptoethanol | AppliChem | A1108 | |
BL21(DE3) | Novagen | 69450-M | |
CaCl2 | AppliChem | A4689 | |
Centrifuge | Eppendorf | 5415R (Z605212) | |
Glutathione | AppliChem | A9782 | |
Glutathione agarose | Pierce | 16100 | Resin for purification of GST-tagged proteins |
Glycerol | AppliChem | A2926 | |
HEPES | AppliChem | A3724 | |
HisPur Ni-NTA Superflow Agarose | Thermo Scientific | 25214 | Resin for purification of 6xHis-tagged proteins |
Imidazole | AppliChem | A1378 | |
IPTG | AppliChem | A4773 | |
KCl | AppliChem | A2939 | |
LB | AppliChem | 414753 | |
Maltose | AppliChem | A3891 | |
MOPS | AppliChem | A2947 | |
NaCl | AppliChem | A2942 | |
NP40 | Roche | 11754599001 | |
pACYCDuet-1 | Sigma-Aldrich | 71147 | Vector for co-expression of proteins with p15A replication origin |
pCDFDuet-1 | Sigma-Aldrich | 71340 | Vector for co-expression of proteins with CloDF13 replication origin |
PMSF | AppliChem | A0999 | |
Protease Inhibitor Cocktail VII | Calbiochem | 539138 | Protease Inhibitor Cocktail |
pRSFDuet-1 | Sigma-Aldrich | 71341 | Vector for co-expression of proteins with RSF replication origin |
SDS | AppliChem | A2263 | |
Tris | AppliChem | A2264 | |
VC505 sonicator | Sonics | CV334 | Ultrasonic liquid processor |
ZnCl2 | AppliChem | A6285 |