Flow Cytometry

Flow cytometry allows for the rapid measurement of multiple optical properties of individual cells, particles, or compartmentalized droplets. Fluorescence-activated cell sorting (FACS) involves the flow cytometric separation of members of the cell population with desired fluorescent properties (and is not limited only to cells). Modern FACS instruments are capable of sorting up to 109 cells in one or a few days. Note that these screening capabilities approach the limits of library cloning and transformation using bacterial expression hosts (Georgiou 2000). The distinction between screening and selection is somewhat blurred at the level of flow cytometric sorting. In one sense this method is a screen, since improved variants are isolated only by measuring the phenotype of every library member. However, this may also be considered a selection method since the entire pool of variants is acted on at one time, and the improved variants are subsequently pooled together, as opposed to spatially separated, and typically enriched through multiple rounds. The high-throughput capabilities combined with the increasing availability of FACS instrumentation have led to the use of flow cytometry as one of the most enabling tools in enzyme engineering. The use of FACS requires the generation of a fluorescence signal that reflects the phenotype of interest, and a major challenge in implementing FACS for enzyme evolution is therefore correlating enzyme activity with cell or particle fluorescence. This has been accomplished in a variety of ways, which are summarized in many reviews (Becker et al. 2004; Farinas 2006; Georgiou 2000; Link et al. 2007). Here we describe some of the important applications of FACS for enzyme engineering.

Intracellular Reactions

In the simplest case, the enzyme of interest is expressed and maintained intracellu-larly, and action on a substrate triggers an intracellular fluorescent signal. An example of this kind is an assay developed by Withers and coworkers for FACS-based sorting of E. coli cells expressing a library of sialyltransferase (ST; belonging to the glycosyltransferase superfamily) (Aharoni et al. 2006). In this case, fluorescently labeled lactose-containing acceptor sugars are freely transported into and out of the cell. Upon intracellular ST-catalyzed addition of sialic acid to the fluorescent sugar, the larger and charged fluorescent sialylated lactoside product is trapped inside the cell. Unreacted fluorescent sugar is washed away and cells retaining fluorescence are sorted by FACS, with increased cellular fluorescence correlating with increased intracellular enzyme activity. Using this approach, a library of >106 ST mutants was screened and yielded a variant with up to 400-fold improved catalytic efficiency (Aharoni et al. 2006).

In a similar example, Shim et al. reported the development of a FACS-based method for isolation of aldolase antibodies (Shim et al. 2004). They used a fluores-cein-linked aldol derivative as a substrate, which upon reaction with an intracellular aldolase releases chloromethylfluorescein, which in turn reacts with nucleophiles causing cellular retention of fluorescence (Shim et al. 2004). The authors demonstrated the ability to apply this screen to isolate catalytic antibodies from a plasmid-encoded library introduced to and expressed in mammalian cells.

Glutathione transferases (GSTs) represent a family of detoxification isoenzymes that catalyze the conjugation of reactive electrophilic compounds to glutathione (GSH). Combinatorial libraries of GST were expressed in E. coli and screening was accomplished by monitoring conjugation of GSH with the fluorogenic substrate monochlorobimane (MCB), which gives rise to a unique fluorescent signal (Eklund et al. 2002). Using a similar FACS-based assay, Georgiou and coworkers evolved highly active GSTs using homology-independent recombination methods for library creation (Griswold et al. 2005). Mutant activity was screened using the substrate 7-amino-4-chloromethyl-coumarin (CMAC), which fluoresces upon conjugation with GSH (Griswold et al. 2005). Another example of the use of FACS to isolate enzymes with improved intracellular activity was reported by Kwon et al. (Kwon et al. 2004), who developed a ferrochelatase screen to improve metalloporphyrin production by E. coli. A library of 2.4 x 106 mutants was expressed in protoporphyrin-overproducing recombinant E. coli cells. The screen was based on the decrease in fluorescence of protoporphyrin IX upon ferrochelatase-catalyzed insertion of Fe(II), resulting in production of heme. In a final example, pH-sensitive mutants of GFP (pHluorins) can act as fluorescent sensors for pH shifts and offer great potential for reporting intracellular activity of hydrolytic enzymes (Schuster et al. 2005).

In order to achieve incorporation of unnatural amino acids at specific sites in a protein in vivo, Schultz and coworkers combined selection with FACS-based screening to evolve aminoacyl-tRNA synthetase specificity toward desired tyrosine analogs (Santoro et al. 2002). To accomplish this, a plasmid was constructed that carries the GFPuv gene under control of a T7 promoter, along with the gene encoding T7 RNA polymerase containing amber codons inserted at positions previously determined to "tolerate" amino-acid substitutions. The T7 polymerase should therefore be functional (and able to express GFPuv) as long as amino acids are incorporated at the amber codons during translation. The same plasmid additionally carries the gene encoding a chloramphenicol resistance reporter (CmR), also containing an amber codon, and the M;YtRNACUA gene encoding an orthogonal amber suppressor tRNATyr derived from Methanococcus jannaschii. This tRNA enables amino-acid incorporation at amber codons and can be aminoacylated by the M. jannaschii tyrosyl-tRNA synthetase (MjYRS), and not by an E. coli tRNA synthetase. A second plasmid expresses a mutant library of 109 variants of MjYRS, which was to be evolved to accept selectively chemical groups other than the phenolic side chain of tyrosine. Using E. coli, a round of selection on chloramphenicol in the presence of the desired tyrosine analog isolated clones expressing MjYRS mutants that allow for amber codon incorporation leading to functional expression of CmR. This "positive" selection, however, does not eliminate clones expressing nonspecific MjYRS mutants that load the MjtRNATyr with amino acids other than the desired tyrosine analog. Amber suppression during T7 RNA polymerase expression and in the absence of the desired Tyr analog is the result of nonspecific amino-acid incorporation and enables GFPuv expression from the T7 promoter. "Negative" sorting by FACS was therefore next employed to eliminate fluorescent clones in the absence of the specific Tyr analog used in the positive selection. Recovery of nonfluorescent clones therefore yields MjYRS mutants specific to only the unnatural amino acid that was used in the positive selection. The authors isolated variants that selectively accommodated one of a variety of unnatural amino acids (e.g., p-isopropyl-phenylalanine) and no natural amino acid.

The Schultz lab applied another FACS-based evolution approach to engineer Cre recombinase from bacteriophage P1 to mediate recombination at new DNA recognition sites (Santoro and Schultz 2002). In this system, a reporter plasmid contained the genes for two GFP variants having different excitation and fluorescence emission properties, EYFP and GFPuv. Only EYFP was placed downstream of a promoter. The plasmid also contained two natural or mutated Cre recognition sequences (loxP). A second plasmid contained a library of Cre variants (~108 library clones). Without Cre-mediated recombination, E. coli cells harboring the reporter plasmid express only EYFP. Cells expressing Cre mutants capable of recombination at the mutated recognition sequence reversibly rearrange the reporter plasmid so that GFPuv and EYFP exchange positions. This in turn causes those cells to express GFPuv from some plasmids, and EYFP from others. Positive sorting by FACS for GFPuv expression using mutated loxP sequences yielded Cre variants with activity toward the new recognition sequence. Subsequent negative sorting for clones only having EYFP fluorescence (and no GFPuv) using native loxP sequences in the reporter plasmid allowed for isolation of Cre mutants with selectivity toward the new loxP sequence and not the native one.

Cell Surface Display

One of the key limitations to intracellular expression and screening of enzyme libraries is the requirement that substrate be transported across cell membranes. Other potential limitations include the possibility that the intracellular environment (e.g., pH, reduction potential) may not be suitable for supporting a properly folded and functional enzyme of interest (e.g., bacterial cytoplasms do not readily support protein disulfide bonds). Displaying the enzyme on the cell surface enables retention of a genotype-phenotype linkage and provides a potential solution to these intracellular screening problems. Surface display technologies coupled to binding-based assays are superb for selecting protein variants with improved ligand binding properties, but have proven difficult to implement for enzyme evolution since reaction products often diffuse away from the enzyme. Therefore, correlating a screenable cellular phenotype (typically fluorescence) with catalytic properties of the displayed enzyme poses a challenge.

In a pioneering application of these methods, Georgiou and coworkers displayed a library of the serine protease OmpT on the E. coli surface and used FACS to isolate variants with improved activity and substrate specificity (Olsen et al. 2000b). Their strategy is depicted in Figure 6.1. A cleavable fluorescence resonance energy transfer (FRET) peptide substrate containing a fluorophore and quenching partner separated by the target scissile bond was embedded on the cell surface via electrostatic interaction (Olsen et al. 2000b). Cleavage at the target peptide sequence results in fluorescence due to disruption of the FRET interaction, allowing for a correlation between activity and fluorescence. Using this approach protease activity on the sequence Arg-Val was improved 60-fold.

Applying directed evolution to improve enzyme activity typically results in mutants with relaxed substrate specificity, unless specificity is included in the screening criteria during evolution. Georgiou and coworkers improved their protease display system by incorporating a means of screening for substrate specificity (Varadarajan et al. 2005). In this system, the substrate with the desired scissile bond sequence was essentially the same as the FRET substrate described previously, in which cleavage results in cell fluorescence. Also included in the screen was a fluorescent "counter-selection" substrate containing an undesired protease cleavage site. The charge on the counter-selection substrate is such that it will not adhere to the cell surface unless cleaved. Positive selection and counter-selection substrates are different colors, allowing simultaneous activity and specificity screening using multicolor hv hv

FIGURE 6.1 FACS-based screen for protease mutants (Olsen et al. 2000b; Varadarajan et al. 2005). A FRET substrate containing fluorophore (Fl) and quenching compound (Q) binds to the surface of E. coli through electrostatic attraction. The quenching compound is released through hydrolysis of the scissile bond (S) by a surface-displayed OmpT mutant, resulting in a fluorescent signal on the cell surface.

FIGURE 6.1 FACS-based screen for protease mutants (Olsen et al. 2000b; Varadarajan et al. 2005). A FRET substrate containing fluorophore (Fl) and quenching compound (Q) binds to the surface of E. coli through electrostatic attraction. The quenching compound is released through hydrolysis of the scissile bond (S) by a surface-displayed OmpT mutant, resulting in a fluorescent signal on the cell surface.

flow cytometry to sort cells with the desired fluorescent properties (cells having fluorescence indicative of FRET substrate cleavage and not counter-selection substrate fluorescence). This approach resulted in an OmpT variant with both high activity on a non-native substrate as well as exceptional substrate specificity (Varadarajan et al. 2005). This approach has the potential to be generally applicable, but it requires substrates derivatized with FRET reagents and will be significantly more difficult to implement for enzyme reactions where substrates are not cleaved (e.g., oxidations or reductions). Similar reaction-specific methods of anchoring fluorescently labeled enzyme reaction products to the cell surface as a means of monitoring catalysis are now being developed for FACS screening (Wilhelm et al. 2007).

The use of bacteria for displaying and engineering eukaryotic proteins requiring disulfide bonds or posttranslational modifications has inherent limitations. Yeast surface display of proteins is an established technology and has recently been implemented for isolating variants of horseradish peroxidase (HRP) with enhanced enantioselectivity (Lipovsek et al. 2007). HRP contains four disulfide bonds and a heme prosthetic group, and is not expressed in a soluble form in bacteria. Two different mutant libraries of the HRP gene (one random mutation library generated by error-prone PCR and one library generated by active-site-directed saturation mutagenesis) were transformed into an S. cerevisiae strain engineered to display the expressed mutant proteins on the cell surface. HRP-catalyzed radical polymerization of cell surface tyrosine residues with fluorescently labeled L-tyrosinol or D-tyrosinol results in cell surface incorporation of the fluorescent label. To select for mutants with enantioselectivity for D-tyrosinol over L-tyrosinol, FACS was used in alternating rounds of positive and negative selection, by isolating clone populations with high fluorescence following exposure to labeled D-tyrosinol (positive selection) and low fluorescence following exposure to labeled L-tyrosinol (negative selection). The opposite selection strategy instead enriches mutants selective for L-tyrosinol. Screening the saturation mutagenesis library using this strategy yielded mutants with 3.8-fold and 7.5-fold improved selectivity for D- over L-tyrosinol and L- over D-tyrosinol, respectively (compared to wild-type HRP). In contrast, the most highly represented clones selected from the random mutagenesis library were not improved relative to wild-type.

Compartmentalization of Whole Cells

Often a FACS-based whole-cell screen for enzyme activity is not possible, even when enzymatic activity can be linked to fluorescence. For example, the fluorescent molecule may rapidly diffuse out of the cell, or if the enzyme is displayed on the cell surface, the fluorescent product may not remain associated with the cell. In these situations compartmentalization of the cells in water-in-oil (w/o) emulsions can be used to keep the phenotype of interest (fluorescence production) associated with the genotype (the cell that produced the fluorescence) within a microdroplet (Taly et al. 2007). The w/o emulsion can be further converted into a water-in-oil-in-water (w/o/w) emulsion, which can be readily analyzed or sorted via flow cytometry. This was demonstrated by Tawfik and coworkers (Aharoni et al. 2005). A library of mutants of the mammalian enzyme serum paraoxonase (PON1) was expressed in E.

coli, and single cells were encapsulated in w/o/w droplets (<10 fL per drop). Droplets containing mutants with improved thiolactonase activity were isolated by FACS, using thiobutryolactone as substrate combined with a fluorogenic thiol-detecting dye that is retained in the droplets. More than 107 mutants were screened, resulting in isolation of clones with 100-fold improved catalytic efficiency (Aharoni et al. 2005).

In Vitro Compartmeimtalization

The compartmentalization techniques described in the previous section were first developed for in vitro genotype-phenotype linkage and screening applications, in a process termed in vitro compartmentalization (IVC) (Aharoni et al. 2005b; Boersma et al. 2007b; Farinas 2006; Taly et al. 2007). In this system, aqueous w/o or w/o/w emulsion microcompartments or microdroplets can act as artificial cells in which enzyme variants are expressed by in vitro transcription and translation. Following in vitro protein expression (IVE), the microdroplets are subject to a fluorescence-based enzyme activity assay. Microdroplets can be the size of a small bacterium, and >1010 individual microdroplets per mL can be readily produced and screened by FACS. The IVC technique offers several potential benefits over the use of cells to express enzyme libraries. Substrates, products, and reaction conditions that are incompatible with in vivo systems are more likely to be compatible with emulsions. In addition, problems related to substrate transport across cell membranes are also eliminated, and the simplified environment of microdroplets as compared to cells reduces enzyme inhibition and cross-reactivity during screening. As an example, IVC was used to evolve the P-galactosidase activity of the E. coli protein Ebg, whose natural function is unknown (Mastrobattista et al. 2005). In this example the aqueous compartments of w/o/w emulsions contained a library of mutated ebg genes (on average one mutant gene per droplet), all necessary IVE reagents, and a fluorogenic substrate for screening p-galactosidase activity (fluorescein-di-P-D-galactopyranoside) by FACS. Following IVE the microdroplets were sorted by FACS and variants with more than 300-fold enhanced catalytic efficiency toward the fluorogenic substrate were isolated.

Another benefit of IVC for enzyme engineering is that the microdroplet environment during screening can vary significantly from that required for gene expression. If conditions are to be altered following IVE, the initial emulsions containing the genetic information and corresponding mutant enzyme may need to be broken so that new emulsions can be formed under the desired screening conditions. In this case it is necessary to physically link the genetic information to the expressed enzyme. This has been accomplished via microbead display, whereby a streptavidin-coated bead acts as scaffold to support a single copy of a biotinylated mutant gene. Following compartmentalized IVE, multiple copies of the expressed enzyme variant are captured onto the bead via affinity tags. Griffith and Tawfik used this method to screen for phosphotriesterase (PTE) mutants with improved activity (Griffiths and Tawfik 2003). Following IVE and protein capture in one emulsion, a second emulsion containing phosphortriester substrate was used to initiate the PTE reaction. In this example, product was also coupled on the beads, and after the PTE reaction the emulsion was broken and the beads were treated with fluorescently labeled antiproduct antibodies and screened by FACS. This method resulted in identification of a PTE mutant with 63-fold higher activity than wild-type enzyme (Griffiths and Tawfik 2003).

Was this article helpful?

0 0

Responses

  • chris
    How to use flow cytometer for protein engineering?
    2 years ago

Post a comment