Applying an Inducible Expression System to Study Interference of Bacterial Virulence Factors with Intracellular Signaling

Many pathogenic bacteria harbor secretion systems to secrete or translocate bacterial effector proteins into the host cell. These effector proteins have the capacity to modulate processes and pathways in the host cell, allowing the bacteria to survive and replicate within their respective intracellular niche. Understanding the biochemical activities and the molecular mechanisms of the effector proteins will help towards a better understanding of pathogenicity and may help to develop new therapeutic tools to combat disease. Furthermore, as the effector proteins frequently exert their function by mimicking host cell activities, they can also be used to learn more about the biology of eukaryotic cells¹. However, studying the biochemical activity of a single effector protein has proven to be complicated for several reasons: 1) functional redundancy among effector proteins, 2) the temporal regulation of engagement of the effector proteins, and 3) the effector protein working in context with another effector protein, interfering with its function. Therefore, the activity of an effector protein is commonly studied by overexpression studies. Unfortunately, the usage of transient overexpression of the effector protein also has several drawbacks. Thus, only single cell analysis assays can be employed, and the activity of the effector protein cannot be analyzed if the activity of the effector protein depends on the activity of another effector protein. To overcome at least the first drawback, stable cell lines can be used. In case the expression of the effector protein is toxic for the cell, this is not suitable. In this scenario, stable and inducible cell lines can be established using a similar system as we describe here for transient expression of apoptosis-inducing proteins (see below). However, our results (Figure 5) demonstrate that the expression of CaeB protects the cells from undergoing apoptosis and is therefore not toxic. There are a few options in dissecting at which step within the apoptotic cascade CaeB interferes with apoptosis induction. One possibility is to determine the host cell binding partner of CaeB. However, this is very challenging and the identification of the binding partner might not help to explain the point of action of CaeB within the apoptotic cascade. Another possibility is to use different apoptosis inducers, but CaeB inhibits apoptosis induction by two different apoptosis inducers, UV-light and staurosporine⁷, indicating that this approach might also not lead to the identification of the point of action of CaeB. Another possibility in narrowing down the point of action of CaeB, which was exploited here, is to use a system where the apoptosis signaling cascade can be induced at defined steps.

In case expression of the bacterial effector protein that was selected as protein of interest already interferes with cell physiology, its expression can also be placed under control of an inducible expression system, e.g., such as the one used to express the target protein. Then, expression of both protein of interest and target protein will be concomitantly induced when the ligand is added to the cell culture allowing the investigator to monitor the cells for differential effects on cell physiology and signaling system output.

Clearly, expressing a protein of interest and trying to identify its point of action in a specific intracellular pathway by inducible expression of selected pathway proteins works best if the protein of interest interferes with the physiological activity of the selected pathway. In this case, readout is simply the lack of a signal normally associated with the pathway, such as cell death in the case of apoptotic signaling or activation of transcription in the case of Wnt signaling. In principle, proteins that activate a specific pathway can also be probed with this type of assay system. It only has to be set up differently, for example, using inducible expression of a si/sh/miRNA against a pathway protein and testing if the protein of interest can rescue pathway signaling. Alternatively, small molecule inhibitors of specific proteins can be used to interrupt signal transduction within a pathway and then to check for pathway rescue by inducible expression of a protein of interest. However, this type of approach will be experimentally more challenging than the protocol presented above.

The selection of the target gene to be probed in the interference assay is, in principle, straightforward. It is best to test as many of the proteins in the selected signaling pathway as possible to fine tune the identification of the protein of interest’s site of interaction with the pathway. This should be done in an iterative fashion, because it is not always experimentally feasible or even possible to monitor all potential interaction partners in a single experiment. Therefore, it is best to first test several proteins spaced out evenly along the entire pathway and then to refine the experiment using the information gained in the first round. In signaling pathways, many proteins are modified post-translationally, either by a covalent modification, such as phosphorylation or sumoylation, or by a processing event, such as proteolytic cleavage. Consequently, the respective protein is either in a ground-state or in an activated form. If possible, it is best to test both forms of the protein for two reasons. One is that the activated form delivers a higher output signal³¹, which presents a more stringent challenge for the protein of interest. In our proof-of-principle example, apoptotic signaling, the caspase 3 is activated by proteolytic cleavage and dimerization. This activation event can be simulated artificially by rearranging the unprocessed pro-caspase gene. The small subunit is artificially placed in front of the large subunit and both are joined by a linker. This is termed a “reverse caspase” and represents the activated form of the enzyme when expressed³⁵. It was chosen due to its higher degree of activity. The second reason for probing both ground-state and activated forms of a pathway protein is because it allows one to distinguish if the protein of interest interferes with activation of the target protein, or if it acts downstream of the protein in the pathway. In the first case, cell death will be inhibited if the ground-state form is expressed, but not if the activated form is present. In the second case, neither protein will lead to apoptosis. Thus, testing both forms of a modified protein improves the resolution of the analytic system.

Cloning the target gene in the inducible expression vector requires several decisions. First, shall expression of the target gene occur only transiently or is it better if the vector is stably integrated into the genome of the host cell line? If only transient expression is required, a simple vector such as pUHC13-3²⁴, which contains a Tet-dependent promoter and a polyA site, or a bidirectional vector such as pBi-2 or pBi-3³⁶, which couples expression of the gene of interest to the expression of a reporter gene for easier monitoring of target gene expression, will be sufficient. If stable expression is preferred, a vector such as pWHE655³³ (Figure 3) should be used. It contains insulators that flank the transgene expression unit to prevent position effects at the genomic integration site from interfering with inducible transgene expression. It also contains a linked, but independently expressed, resistance cassette for G418 to facilitate the selection of vector integration events. The net result of these modifications is an increase in the number of clones with excellent regulatory properties^33,37,38. cDNA sequences of target genes can be cloned into this vector by using the unique restriction sites for EcoRI and EcoRV³³. Cloning of the vector expressing the reverse Caspase-3 cDNA was described in detail in reference 33. Cloning of the vector expressing human Bax was performed analogously. The human bax cDNA was amplified by PCR from the plasmid pVenus-Bax³⁹ using oligonucleotide primers that introduce restriction sites for EcoRI and EcoRV. After purification of the PCR fragment and restriction with the two enzymes, the fragment was ligated with likewise-restricted pWHE655, yielding pWHE655TREtight-Bax. Second, which minimal promoter is most appropriate for transgene expression? The original minimal promoter P_hCMV*-1²⁴ displays some cell-type specific leakiness³⁰, due in part to the presence of functional interferon-α inducible response elements²⁹. This led to the development of second generation Tet-dependent promoters, termed ΔMtetO, TREtight and SG-TRE, with altered spacing and sequences between the tetO elements and with different minimal promoter sequences^33,40-42. They display greatly reduced leakiness in the absence of doxycycline. Among these second generation promoters, ΔMtetO has the lowest transgene expression level in the presence of doxycycline³³. TREtight mediates an intermediate and SG-TRE the highest level of transgene expression³³. Further modifications have minimized background expression of the tetracycline responsive promoter even more^43,44. Thus, the choice of a tetracycline-responsive promoter can be made according to the requirements or limitations of the respective transgene.

Independent of the type of experimental approach, tightness of expression of the target protein is important. For example, induction of apoptosis with activated forms of pro-apoptotic proteins leads to cell death, even when only minimal amounts of the respective target protein are expressed^31-33. The required stringency is achieved by adding a tetracycline-dependent transsilencer to the conditional regulatory system³⁰. It ensures that transcription from the inducible promoter is actively repressed, which is especially important when transient transfections are performed, since they are typically rather leaky in their transgene expression³⁰. If signaling pathways that do not induce cell death are probed, the requirement for stringent regulation of target gene expression may be relaxed and inducible systems without a transsilencer could be used alternatively^43,44.

The inducible regulatory system used to probe the signaling pathway does not have to be the Tet system. Its major advantage is that it has been intensively used in mammalian cells for more than 20 years now and that it is the most highly developed and best characterized inducible control system. Nevertheless, there are at least 25 other conditional regulatory systems^45,46 that could be employed alternatively, or in addition to the Tet system. Two or three different inducible systems might be useful in probing convergent signaling pathways or for testing redundancy of either the target gene or the protein of interest.

The procedure presented here should work with transfection efficiencies that result in a robust and reliable output signal from the respective signaling pathway that is probed. Because the stable cell lines express the protein of interest in a homogenous manner (Figure 4B), all cells that are transiently transfected with the inducible plasmid expressing the protein of interest are probed for interference of the protein of interest with pathway signaling.