Luke D. Lavis

Bright Building Blocks for Chemical Biology

Small-molecule fluorophores manifest the ability of chemistry to solve problems in biology. As we noted in a previous review (Lavis, L. D.; Raines, R. T. ACS Chem. Biol.2008, 3, 142–155), the extant collection of fluorescent probes is built on a modest set of “core” scaffolds that evolved during a century of academic and industrial research. Here, we survey traditional and modern synthetic routes to small-molecule fluorophores and highlight recent biological insights attained with customized fluorescent probes. Our intent is to inspire the design and creation of new high-precision tools that empower chemical biologists.

Carbofluoresceins and carborhodamines as scaffolds for high-contrast fluorogenic probes.

Fluorogenic molecules are important tools for advanced biochemical and biological experiments. The extant collection of fluorogenic probes is incomplete, however, leaving regions of the electromagnetic spectrum unutilized. Here, we synthesize green-excited fluorescent and fluorogenic analogues of the classic fluorescein and rhodamine 110 fluorophores by replacement of the xanthene oxygen with a quaternary carbon. These anthracenyl “carbofluorescein” and “carborhodamine 110” fluorophores exhibit excellent fluorescent properties and can be masked with enzyme- and photolabile groups to prepare high-contrast fluorogenic molecules useful for live cell imaging experiments and super-resolution microscopy. Our divergent approach to these red-shifted dye scaffolds will enable the preparation of numerous novel fluorogenic probes with high biological utility.

Ester bonds in prodrugs.

A recent study challenges the oft-held notion that ester bonds in prodrug molecules are cleaved rapidly and completely inside cells by endogenous, nonspecific esterases. Structure-activity relationship studies on acylated sugars reveal that regioisomeric compounds display disparate biological activity, suggesting that ester bonds can persist in a cellular context.

Synthesis of Rhodamines from Fluoresceins Using Pd-Catalyzed C–N Cross-Coupling

A unified, convenient, and efficient strategy for the preparation of rhodamines and N,N′-diacylated rhodamines has been developed. Fluorescein ditriflates were found to undergo palladium-catalyzed C–N cross-coupling with amines, amides, carbamates, and other nitrogen nucleophiles to provide direct access to known and novel rhodamine derivatives, including fluorescent dyes, quenchers, and latent fluorophores.

Onconase cytotoxicity relies on the distribution of its positive charge.

Onconase® (ONC) is a member of the ribonuclease A superfamily that is toxic to cancer cells in vitro and in vivo. ONC is now in Phase IIIb clinical trials for the treatment of malignant mesothelioma. Internalization of ONC to the cytosol of cancer cells is essential for its cytotoxic activity, despite the apparent absence of a cell‐surface receptor protein. Endocytosis and cytotoxicity do, however, appear to correlate with the net positive charge of ribonucleases. To dissect the contribution made by the endogenous arginine and lysine residues of ONC to its cytotoxicity, 22 variants were created in which cationic residues were replaced with alanine. Variants with the same net charge (+2 to +5) as well as equivalent catalytic activity and conformational stability were found to exhibit large (> 10‐fold) differences in toxicity for the cells of a human leukemia line. In addition, a more cationic ONC variant could be either much more or much less cytotoxic than a less cationic variant, again depending on the distribution of its cationic residues. The endocytosis of variants with widely divergent cytotoxic activity was quantified by flow cytometry using a small‐molecule fluorogenic label, and was found to vary by twofold or less. This small difference in endocytosis did not account for the large difference in cytotoxicity, implicating the distribution of cationic residues as being critical for lipid‐bilayer translocation subsequent to endocytosis. This finding has fundamental implications for understanding the interaction of ribonucleases and other proteins with mammalian cells.

Latent Blue and Red Fluorophores Based on the Trimethyl Lock

Fluorescent molecules are indispensable tools in modern biochemical and biological research, being used as labels for biomolecules, indicators for ions, stains for organelles, and substrates for enzymes.[1] The major targets of this last class are hydrolases that catalyze the removal of a masking moiety, thereby modulating fluorescence.[2] A critical property of fluorogenic hydrolase substrates is their chemical stability in aqueous solution, as spontaneous hydrolysis can compete deleteriously with enzymatic activity. New substrate classes that exhibit increased stability while maintaining enzymatic reactivity would be highly desirable. Our laboratory recently reported on the use of the “trimethyl lock” strategy in the design of a latent fluorophore.[3] This latent fluorophore consists of a trimethyl lock component inserted between a dye and enzyme-reactive group. The trimethyl lock is an o-hydroxycinnamic acid derivative in which unfavorable steric interactions between three methyl groups encourage rapid lactonization to form a hydrocoumarin and release a leaving group.[4, 5] Our initial latent fluorophore exhibited remarkable stability in aqueous solution, but released a xanthene dye (rhodamine 110) upon hydrolytic cleavage by an esterase. Here, we explore the modularity of our design, probing its applicability to unrelated dyes that absorb at short (blue) and long (red) wavelengths. Coumarin-based compounds comprise an important class of blue dyes possessing UV or near-UV excitation wavelengths.[6] Acyl esters and acyloxymethyl ethers of 7-hydroxycoumarin (i.e., umbelliferone) can be substrates for esterases.[7–9] 7-Amino-4-methylcoumarin (AMC) is used widely as the basis for protease substrates.[10–12] Upon amidation, the excitation and emission wavelengths of AMC are shifted to shorter wavelengths with concomitant reduction in quantum yield.[10] We reasoned that AMC could be subjected to our latent fluorophore strategy. Accordingly, we condensed AMC with acetylated trimethyl lock 1[13] to give pro-fluorophore 2 (Scheme 1), which displayed the expected hypsochromic shift of excitation and emission spectra relative to free AMC (Figure 1). The hydrolysis of profluorophore 2 was catalyzed by porcine liver esterase (PLE) with kcat/KM = 2.5 × 105 M−1s−1 and KM = 8.2 μM (Figure 2A). This kcat/KM value is 102-fold greater than the apparent kcat/KM value for the latent fluorophore based on rhodamine 110.[3] (We use the term “apparent” because full fluorescence manifestation requires the lactonization of two trimethyl lock moieties.[3]) Figure 1 Normalized fluorescent excitation–emission spectra of pro-fluorophore 2 and AMC. Figure 2 (a) Kinetic traces (λex 365 nm, λem 460 nm) and Michaelis–Menten plot (inset) for the hydrolysis of pro-fluorophore 2 (20 μM→9.8 nM) by PLE (2.5 μg/mL); kcat/KM = 2.5 × 105 M−1s−1 ... Scheme 1 Route for the synthesis of pro-fluorophore 2, and the chemical structure of 4-methylumbelliferyl acetate (3). To evaluate the relative stability of pro-fluorophore 2, we monitored the accretion of fluorescence in phosphate-buffered saline (PBS) containing bovine serum albumin (BSA; 1.0 mg/mL) and either pro-fluorophore 2 or 4-methylumbelliferyl acetate (3, which is a common esterase substrate[9]). Pro-fluorophore 2 proved to be highly stable compared to 4-methylumbelliferyl acetate (Figure 3). An additional advantage of pro-fluorophore 2 is that the product of its hydrolysis, AMC, shows little change in fluorescence at pH ≥ 5.[1] In contrast, the fluorescence of 4-methylumbelliferone is highly variable due to its pKa = 7.40[14] being near the physiological pH. Figure 3 Time course of the generation of fluorescence (λex 365 nm, λem 460 nm) from pro-fluorophore 2 (50 nM) or methylumbelliferyl acetate 3 (50 nM) in PBS containing BSA (1 mg/mL). To access longer wavelengths, we turned to the oxazine class of red dyes. One such dye, cresyl violet (CV) has been used for decades to stain tissues[15, 16] and as a laser dye.[17] Although CV has a maximal absorbance at 586 nm, its fused benzo ring broadens its absorption spectrum[18] and thereby allows excitation with a variety of light sources. Diamide derivatives of CV are virtually nonfluorescent and thus useful in the production of fluorogenic protease substrates.[19–21] The attributes of CV make this dye an attractive candidate for our latent fluorophore strategy. Accordingly, we condensed CV with acetylated trimethyl lock 1[13] to give pro-fluorophore 4 (Scheme 2). Although pro-fluorophore 4 was stable in PBS containing BSA (data not shown), its hydrolysis was catalyzed by PLE with apparent kinetic parameters of kcat/KM = 1.2 × 105 M−1s−1 and KM = 0.14μM (Figure 2B). The kcat/KM value for the hydrolysis of pro-fluorophore 4 is similar to that of pro-fluorophore 2; the KM value is, however, 10-fold lower than that of pro-fluorophore 2. Scheme 2 Route for the synthesis of pro-fluorophore 4. Pro-fluorophore 4 serves as a probe for esterase activity in live mammalian cells. The broad excitation peak allowed confocal microscopy experiments using excitation at 543 nm and emission at 605 nm. Pro-fluorophore 4 was hydrolyzed by esterases endogenous in the cells of a rodent (Figure 4A) or human (Figure 4B) to give a red-stained cytosol within minutes. The somewhat enhanced fluorescence observed in the human cells could be indicative of more efficient internalization or hydrolysis of the latent fluorophore. After longer incubation, the liberated CV became further localized in subcellular compartments (data not shown), which is consistent with previous reports.[21] To date, there have been few reports of long-wavelength esterase substrates that are useful in cell biology.[1] Its evident stability, optical properties, and fast intracellular eduction make pro-fluorophore 4 useful in a wide variety of biological applications, especially in assays involving fluorophores of different wavelengths. Figure 4 Unwashed mammalian cells incubated for 15 min with pro-fluorophore 4 (10 μM) at 37 °C in DMEM and counter-stained with Hoechst 33342 (5% v/v CO2(g), 100% humidity). (a) CHO K1 cells. (b) HeLa cells. In conclusion, we have established that one component of our latent fluorophores—the dye—is modular (Figure 5). Specifically, we have now prepared useful latent fluorophores from three dyes (blue, green,[3] and red), all linked by a trimethyl lock moiety to an esterase-reactive group. In future work, we shall explore the modularity of the other component—the enzyme-reactive group. We anticipate that the end result will be a broad spectrum of stable latent fluorophores with numerous applications in biochemical and biological research. Figure 5 Modules in the latent fluorophores described in this work.

A general method to fine-tune fluorophores for live-cell and in vivo imaging

Pushing the frontier of fluorescence microscopy requires the design of enhanced fluorophores with finely tuned properties. We recently discovered that incorporation of four-membered azetidine rings into classic fluorophore structures elicits substantial increases in brightness and photostability, resulting in the Janelia Fluor (JF) series of dyes. We refined and extended this strategy, finding that incorporation of 3-substituted azetidine groups allows rational tuning of the spectral and chemical properties of rhodamine dyes with unprecedented precision. This strategy allowed us to establish principles for fine-tuning the properties of fluorophores and to develop a palette of new fluorescent and fluorogenic labels with excitation ranging from blue to the far-red. Our results demonstrate the versatility of these new dyes in cells, tissues and animals.

Fluorogenic affinity label for the facile, rapid imaging of proteins in live cells

Haloalkane dehalogenase (HD) catalyzes the hydrolysis of haloalkanes via a covalent enzyme-substrate intermediate. Fusing a target protein to an HD variant that cannot hydrolyze the intermediate enables labeling of the target protein with a haloalkane in cellulo. The utility of extant probes is hampered, however, by background fluorescence as well as limited membrane permeability. Here, we report on the synthesis and use of a fluorogenic affinity label that, after unmasking by an intracellular esterase, labels an HD variant in cellulo. Labeling is rapid and specific, as expected from the reliance upon enzymic catalysts and the high membrane permeance of the probe both before and after unmasking. Most notably, even high concentrations of the fluorogenic affinity label cause minimal background fluorescence without a need to wash the cells. We envision that such fluorogenic affinity labels, which enlist catalysis by two cellular enzymes, will find utility in pulse-chase experiments, high-content screening, and numerous other protocols.

Virginia Orange: A Versatile, Red-Shifted Fluorescein Scaffold for Single- and Dual-Input Fluorogenic Probes

Fluorogenic molecules are important tools for biological and biochemical research. The majority of fluorogenic compounds have a simple input-output relationship, where a single chemical input yields a fluorescent output. Development of new systems where multiple inputs converge to yield an optical signal could refine and extend fluorogenic compounds by allowing greater spatiotemporal control over the fluorescent signal. Here, we introduce a new red-shifted fluorescein derivative, Virginia Orange, as an exceptional scaffold for single- and dual-input fluorogenic molecules. Unlike fluorescein, installation of a single masking group on Virginia Orange is sufficient to fully suppress fluorescence, allowing preparation of fluorogenic enzyme substrates with rapid, single-hit kinetics. Virginia Orange can also be masked with two independent moieties; both of these masking groups must be removed to induce fluorescence. This allows facile construction of multi-input fluorogenic probes for sophisticated sensing regimes and genetic targeting of latent fluorophores to specific cellular populations.

Distinct substrate selectivity of a metabolic hydrolase from Mycobacterium tuberculosis.

The transition between dormant and active Mycobacterium tuberculosis infection requires reorganization of its lipid metabolism and activation of a battery of serine hydrolase enzymes. Among these serine hydrolases, Rv0045c is a mycobacterial-specific serine hydrolase with limited sequence homology outside mycobacteria but structural homology to divergent bacterial hydrolase families. Herein, we determined the global substrate specificity of Rv0045c against a library of fluorogenic hydrolase substrates, constructed a combined experimental and computational model for its binding pocket, and performed comprehensive substitutional analysis to develop a structural map of its binding pocket. Rv0045c showed strong substrate selectivity toward short, straight chain alkyl esters with the highest activity toward four atom chains. This strong substrate preference was maintained through the combined action of residues in a flexible loop connecting the cap and α/β hydrolase domains and in residues close to the catalytic triad. Two residues bracketing the substrate-binding pocket (Gly90 and His187) were essential to maintaining the narrow substrate selectivity of Rv0045c toward various acyl ester substituents, as independent conversion of these residues significantly increased its catalytic activity and broadened its substrate specificity. Focused saturation mutagenesis of position 187 implicated this residue, as the differentiation point between the substrate specificity of Rv0045c and the structurally homologous ybfF hydrolase family. Insertion of the analogous tyrosine residue from ybfF hydrolases into Rv0045c increased the catalytic activity of Rv0045 by over 20-fold toward diverse ester substrates. The unique binding pocket structure and selectivity of Rv0045c provide molecular indications of its biological role and evidence for expanded substrate diversity in serine hydrolases from M. tuberculosis.