|Home | About | Journals | Submit | Contact Us | Français|
The stability of proteins is tuned by evolution to enable them to perform their cellular functions for the success of an organism. Yet, most of the arsenal of biophysical techniques at our disposal to characterize the thermodynamic stability of proteins is limited to in vitro samples. We describe an approach that we have developed to observe a protein directly in a cell and to monitor a fluorescence signal that reports the unfolding transition of the protein, yielding quantitatively interpretable stability data in vivo. The method is based on incorporation of structurally nonperturbing, specific binding motifs for a bis-arsenical fluorescein derivative in sites that result in dye fluorescence differences between the folded and unfolded states of the protein under study. This fluorescence labeling approach makes possible the determination of thermodynamic stability by direct urea titration in Escherichia coli cells. The specific case study we describe was carried out on the predominantly β-sheet intracellular lipid-binding protein, cellular retinoic acid-binding protein (CRABP), expressed in E. coli.
Optimal thermodynamic stability of proteins is crucial to their physiological functions and activities. Reduced protein stability can be detrimental, leading to misfolding pathologies such as the neurodegenerative diseases – Alzheimer's, Parkinson's, and other amyloid diseases (1, 2). Conversely, overly stable proteins may lose the ability to respond to allosteric modulators. Therefore, a molecular understanding of a protein requires analysis of its thermodynamic stability. A wide array of informative and sophisticated in vitro approaches yields quantitative descriptions of stability under a given set of experimental conditions. For many studies, these stabilities have been assumed to be applicable also in vivo. The crowded cellular environment, however, will modulate conformational flexibility of a protein and adds complexity to folding and unfolding pathways (3, 4). The thermodynamic stability of a protein is very likely to be altered by macromolecular crowding, which in turn will influence both folding and aggregation reactions. Additionally, molecular chaperones interact with a large fraction of the cellular proteome, as do many small ligands that are present in a cell (5); these interactions, along with changes in the oxidative potential due to responses to oxidative stress (6), can influence the thermodynamic stability of a protein in the cellular environment. Therefore, it is highly desirable to develop methods to measure protein stability directly in cells.
Technical challenges make the goal of measuring protein stability in cells extremely difficult, and only a few groups have reported in vivo stabilities. Oas and colleagues applied amide hydrogen exchange detected by MALDI mass spectrometry in a pioneering study to provide the first direct measurements of in vivo protein stability in the Escherichia coli cytoplasm (7). Their results showed the thermodynamic stability of the small monomeric δ repressor in the cell to be the same as in the test tube. However, in this method cells are lysed prior to mass spectrometry measurement. Consequently, this approach cannot readily be used to explore directly how different physiological states alter thermodynamic stabilities in the cell. Similarly, the pulse proteolysis approach recently introduced by Marqusee and coworkers (8), which is based on selective digestion of unfolded proteins in equilibrium mixtures of folded and unfolded proteins, requires conversion of intact cells into lysates. For maximum versatility, a method to measure protein stability directly in the cell, during different physiological situations, would be advantageous.
A variety very of qualitative in vivo approaches have been designed that monitor folding and solubility (which is generally reliant on proper folding) of expressed proteins. For example, one strategy is based on the necessity for correct folding for successful structural complementation of a reporter protein (9). Here, the reporter protein is split into two parts that must recombine to function, providing an efficient way of screening for folding-competent and soluble mutants. Another approach is based on reading out the efficiency of fluorescence resonance energy transfer (FRET) between an N-terminal blue fluorescent protein and a C-terminal green fluorescent protein (10). FRET efficiency will be enhanced when the fusion protein folds to the compact native state. Fusion of a protein of interest to chloramphenicol acetyltransferase was used to identify well-folded mutants on the assumption that only these would be soluble and confer resistance to chloramphenicol (11). However, all of these approaches are limited in their ability to provide a quantitative measure of protein stability.
We have developed a fluorescence-based approach to determine protein stability in vivo (12, 13) using a well-behaved model system – the 136-amino-acid cellular retinoic acid-binding protein (CRABP) (14). CRABP is visualized in the context of all macromolecules present in the cell using the membrane-permeable bis-arsenical fluorescein-based dye ‘FlAsH’ (15). This fluorescent dye ligates to a genetically engineered tetracysteine motif (Cys-Cys-Xxx-Yyy-Cys-Cys); the extremely rare occurrence of this motif in the cellular proteome ensures high specificity of labeling (15). By engineering the specific tetracysteine sequence (here Cys-Cys-Gly-Pro-Cys-Cys) into the internal Ω-loop of CRABP (incorporating the native Gly-Pro present in this loop), we created a tetra-Cys CRABP variant that binds FlAsH and yields a fluorescent emission intensity sensitive to the conformational state of the protein, with the denatured ensemble hyperfluorescent compared to the native state (13). FlAsH fluorescence can therefore be used to follow the transition from native to unfolded CRABP during unfolding by chemical denaturant. This approach enables determination of the free energy of unfolding in vivo; FlAsH fluorescence can be used as a direct read-out to monitor the urea-induced unfolding of tetra-Cys CRABP directly in the cell. The complexity of the cellular environment demands cautious interpretation of the thermodynamic data obtained, as many cellular components may be perturbed by the urea treatment. Nonetheless, direct observations in cells will provide new insights. Fulfilling a requirement for its use in these measurements, tetra-Cys CRABP is soluble and indistinguishable in structure and function from its native counterpart whether FlAsH-labeled or unlabeled (13). The FlAsH labeling approach has also been applied to mutants of CRABP in order to explore the effects of specific residue substitutions: for example, mutation of the helix-terminating residue Pro39 to Ala was known to retard the folding and unfolding of CRABP (16), and P39A tetra-Cys CRABP shows a high tendency to form aggregates in vitro (17). By incorporating the tetra-Cys motif into P39A CRABP, we could follow formation of aggregates in real time in vivo (13).
The sensitivity of the FlAsH quantum yield to the conformational state of the protein is the premise for the application of this approach to directly measure in vivo stability. Application of this strategy to other proteins necessitates a careful structure-directed choice of a sequence from the target protein, so that incorporation of the tetra-Cys motifis tolerated without structural perturbation, and also that the FlAsH quantum yieldis sensitive to the folding of the protein host. In addition to the overall structural constraints each protein provides for incorporation of the tetra-Cys sequence, the geometric properties of the binding sites are crucial to the FlAsH fluorescence characteristics (B. Krishnan and L. Gierasch, manuscript in preparation). Although the successful design of a FlAsH-binding tetracysteine tag into a given protein might require multiple trials, a clear advantage provided by this system is the direct read-out of stability in intact cells.
This article describes an approach that we recendy developed to determine in vivo protein stability using a high-copy-number plasmid for the expression of the protein of interest in E. coli BL21(DE3) cells. Our results show that other E. coli mutant strains WG710 and WG708 (18) give comparable and reproducible results to the E. coli BL21(DE3) cells (19,20). This provides evidence for the general applicability of the procedure to any E. coli strain. To obtain reliable and reproducible results it is important to start the urea titrations at a time point after induction when a sufficient protein amount is already synthesized in the cell. The use of high-copynumber plasmids yields adequate protein as soon as one hour after induction. One might consider using low-copy-plasmids as well; ultimately, the starting point of the urea titration should be established in any particular case depending on the rate and yield of protein biosynthesis for the plasmid used.
Cell viability and changes in the cell number during the incubation times in urea (Note 7) were qualitatively assessed by measuring the optical density at 600 nm (Fig. 7.2), using LB medium as a blank. This method alone, while rapid, is not sufficient to observe a potential deleterious effect of urea on cell viability, since dead cells can also contribute to the bulk absorbance at 600 nm. For more careful assessment of viability, the fraction of viable cells should be determined on solid nutrient medium by taking one μL samples of the cell suspension, diluting them 10,000-fold in fresh sterile LB medium, plating them on LB-agar (LB medium solidified by addition of 15 g/L agar), and incubating the plates at 37°C overnight. Colonies are then counted, and the number of the viable cells is calculated per mL (Fig. 7.2).
It is essential for the determination of in vivo stability that the protein under study remain soluble throughout the measurement. Hence, we recommend a control to insure that the measured FlAsH signal arises from soluble protein; separation of soluble and insoluble protein by fractionation gives direct information on how the protein under study behaves in the expression system used. In comparing the behavior of different proteins and mutants of any given protein under study, it may be of interest to assess the extent to which a given protein partitions to the insoluble (pellet) fraction, and this same protocol can be used for this goal.
During the time course of growth or urea incubation of the labeled cells expressing either tetra-Cys CRABP or P39A tetra-Cys CRABP (see Section 3.1), 20 μL of cell suspension is withdrawn and concentrated twice by centrifugation at 7,297 X g for 3 min and subsequent resuspension in 50 mM Hepes buffer, pH 7.5. A volume of 2 μL of this concentrated suspension is immobilized in 1% agarose in LB and imaged with a fluorescent microscope (Nikon Eclipse E600, Melville, NY) with excitation at 485 nm and a 510-nm emission cut-on filter. The images are processed with the Openlabs software (Improvision, Lexington, MA). Examples of micrographs are shown in Fig. 7.3.
We appreciate critical reading of the manuscript by Joanna Swain, Beena Krishnan, and Qinghua Wang. The authors gratefully acknowledge support from the National Institutes of Health (grants GM027616 and a 2006 NIH Director's Pioneer Award to LMG),and DFG-project IG73/4-1 and the Heisenberg award IG73 1-1 (to ZI).
1This protocol can be adapted to any protein provided that a successful design of a FlAsH-binding motif that is sensitive to global conformational changes can be achieved. In addition, the engineered tetra-Cys sequence should not perturb the structural integrity and thermodynamic stability of the protein host. Moreover, the system can be used to compare the destabilizing effect of mutations before a time-costly and potentially low yield purification is undertaken.
2LB absorbs at 600 nm (OD600 is between 0.07 and 0.09), and fresh LB is used as a blank in the OD measurements.
3The EDT stock solution (10 mM in DMSO) is best stored at –20°C and is freshly diluted prior to each labeling experiment. EDT generates an extremely unpleasant thiol smell, and the initial aliquoting and addition to the culture needs to be handled exclusively under the hood. Containers with tight-fitting lids (i.e., falcon tubes, Eppendorf tubes, or Schott bottles) should be used instead of common flasks or culture tubes for cultivation of the cells. The used pipette tips or other plastic materials should be collected in a tightly closed container (preferably stored under the hood) and then discarded according to safety regulations.
4Different fluorescence patterns are observed upon induction of soluble and aggregation-prone proteins; the signal of a soluble protein reports on the steady increase of its amount during protein synthesis, while that for an aggregation-prone variant may show an altered signal reporting on aggregate formation. In die case of FlAsH-labeled tetra-Cys P39A CRABP, aggregates are hyperfluorescent, leading to a pronounced upswing in fluorescence of bulk cell samples upon initiation of aggregation (13). Fluorescence microscopy images confirm these results: Whereas the fluorescence is spread uniformly throughout the cytoplasm in cells expressing the soluble tetra-Cys CRABP, in cells expressing aggregation-prone P39A tetra-Cys CRABP hyperfluorescent aggregates are observed near the poles (Fig. 7.3) (13). In parallel, cell fractionation studies reveal the partitioning of the protein between the soluble and insoluble fractions.
5We observed that the incubation time required for establishment of equilibrium between folded and unfolded populations of CRABP in vivo is significantly shorter than in vitro (Figs. 7.1 and 7.4). Based on our preliminary measurements, we believe that acceleration of the unfolding rate is the most likely factor leading to faster equilibration in vivo (12), but studies are underway to dissect all possible factors. Our initial approach to measuring stability in vivo used a short incubation time of the cells in different urea concentrations (30 min) with a goal of minimizing the negative impact on viability of the cells at higher urea concentrations (13). However, follow-up studies on the incubation time dependence of quantitative in vivo stability experiments clearly indicate that the minimum time required for equilibration for either tetra-Cys CRABP or P39A tetra-Cys CRABP in vivo is 75 min, as indicated by the absence of further change in the urea melt as incubation time is increased (Fig. 7.1A and B) (12). The time dependence of establishing an equilibrium between folded and unfolded populations in vivo can vary significantly (i.e., fast-folding proteins might require shorter incubation times), and the optimal incubation time needs to be determined for each protein by monitoring in parallel the cell viability and the approach of the observed fluorescence to a constant value at any given urea concentration.
6To extract any thermodynamic data from the stability curves, the urea melts need to be reversible. In our case, the melts are reversible only for the completely soluble tetra-Cys CRABP protein (Fig. 7.5). To measure the stability of aggregation-prone proteins, which form detergent-resistant aggregates (i.e., amyloid aggregates), one might consider shorter induction times or lower expression levels in order to carry out the urea melt before insoluble structures are formed.
7Longer incubation times are accompanied by losses in cell viability (Fig. 7.2). To offset the impact of cell loss, prior to FlAsH-fluorescence measurements the optical density (OD600) is recorded, and the OD600 of all the samples is normalized to the optical density of the 3-M urea sample by addition of fresh LB medium at the appropriate urea concentration.
8The lysozyme treatment converts the cell pellet into a very viscous suspension due to the release of DNA. The subsequent hydrolysis with DNase reduces the viscosity of the solution.