A. Chemistry in Living Systems – New Tools for Glycobiology
Glycans decorate eukaryotic cell surfaces, where they are poised to mediate a variety of cell surface recognition events such as bacterial and viral binding to host cells and leukocyte adhesion during an inflammatory response. In addition to their cell surface roles, glycans can regulate many intracellular processes, including trafficking of proteins to the lysosome and transcription and translation. Despite the involvement of glycans in so many biological systems, progress toward delineating the molecular basis of their function has been slow relative to comparable studies of proteins and nucleic acids. This is due to the fact that the biosynthesis of glycans, unlike the other biopolymers, is not template driven and under direct genetic control. As a consequence, glycans can be heterogeneous and their structures difficult to perturb using conventional genetic techniques. A major focus of our research is the development of chemical approaches for probing the functions of glycans in cell-based systems, and the application of these tools to studies of glycobiology.
1. Metabolic labeling of glycans with bioorthogonal chemical reporters
Metabolic oligosaccharide engineering is the process of modulating cellular glycan structures by biosynthetic introduction of unnatural sugars containing unique functional groups termed "bioorthogonal chemical reporters" (Fig. 1). We first described the concept in 1997 and have since expanded and applied the technique to several problems in glycobiology, including proteomic analysis of protein glycosylation and chemical remodeling of cell surfaces in living animals. More broadly, the bioorthogonal chemical reporter concept has now been popularized within the chemical biology community by virtue of its use in protein labeling, profiling cellular targets of synthetic inhibitors, and identifying posttranslational modifications beyond glycosylation.
Figure 1. Metabolic oligosaccharide engineering.
A critical element of this research was the development of highly selective chemical reactions that can be performed in living systems. We identified the azide as an ideal chemical reporter for metabolic labeling of glycans due to its minimal size, biocompatibility and selective reactivity profile. We introduced the Staudinger ligation with phosphine analogs as a means to selectively tag azide-labeled glycans with probes for visualization or affinity capture (Fig. 2). Our toolkit of "azido sugars" for metabolically labeling different classes of glycans currently includes four sugars for labeling sialic acids (ManNAz), mucin-type O-linked glycoproteins (GalNAz), cytosolic O-GlcNAcylated proteins (GlcNAz), and fucosylated glycans (6-AzFuc).
Figure 2. Conversion of ManNAz to SiaNAz and subsequent Staudinger ligation.
Despite these successes, we anticipated that certain applications would be restricted by intrinsic properties of the Staudinger ligation, such as slow kinetics and nonspecific phosphine oxidation. Thus, we developed a superior bioorthogonal reaction based on the Huisgen [3+2] dipolar cycloaddition with alkynes. This typically sluggish reaction can be accelerated, as demonstrated by Sharpless and coworkers, by use of a copper (I) catalyst. Unfortunately, such catalysts are cytotoxic. We therefore adopted an alternative means of alkyne activation that would not require a toxic catalyst: ring strain. Insertion of alkynes into an 8-membered ring incurs 18 kcal/mol of ring strain, much of which is released in the transition state for cycloaddition with azides. Thus, cyclooctyne probes 1-4 were synthesized (Fig. 3) and their reactions with azides on live cell surfaces evaluated. The difluorinated cyclooctyne (4) reacts with azides approximately 60-fold faster than the phosphines used in the Staudinger ligation, and with no toxic effects on cells. We are presently evaluating this reagent for proteomic profiling of glycosylation and in vivo imaging applications.
Figure 3. Cyclooctynes for strain-promoted cycloadditions with azides in living systems.
Future directions
Imaging glycosylation
Despite numerous reports of changes in cell-surface glycosylation associated with disease, there are virtually no reports of glycan-specific imaging. The ability to chemically modify cell surface glycans in living animals provides a means to monitor changes in glycosylation in a physiologically authentic context. For example, changes in glycan profiles that occur during tumor growth and metastasis could be visualized in real time. We plan to explore applications of azidosugar metabolism/covalent ligation to tumor targeting with probes for noninvasive imaging. Toward this end, we are synthesizing probes for multiple imaging modalities and establishing murine tumor models.
Identification of novel glycan-based tumor biomarkers
The availability of biomarkers for early detection of cancer would improve the accuracy of diagnosis and long-term patient survival. Biomarkers that can be detected in body fluids, such as serum or urine, are particularly important to identify, as they allow for non-invasive screening across a large population. At present, there are less than a handful of biomarkers in clinical use, and these suffer from a high rate of false-positives. There is extensive literature that links changes in mucin-type O-linked glycosylation with epithelial cancers. We plan to employ metabolic labeling with GalNAz as a means to identify changes in mucin-type glycosylation in murine cancer models, with the long-term objective of finding new biomarkers. The first step in this process is to develop a method for proteomic identification of GalNAz-labeled glycoproteins using mass spectrometry. The method will then be used to analyze serum and tumor tissue samples from normal and tumor-bearing mice.
2. Small molecule modulators of enzymes involved in glycan biosynthesis and regulation
The field of glycobiology could benefit from small molecule tools that affect cell-surface glycan expression in a temporally controlled fashion. One means to achieve this is by targeting the enzymes involved in the biosynthesis and processing of glycans within the secretory pathway. Glycosyltransferases are the primary biosynthetic enzymes that assemble glycans from monosaccharide building blocks. Sulfotransferases further modulate the structures and functions of glycans by installing sulfate esters, akin to phosphorylation of proteins by kinases. There are ~250 predicted glycosyltransferases in the human genome and 37 glycan-specific sulfotransferases. A major goal of our research is to develop small molecule modulators of these enzymes. Thus far we have adopted two approaches: (a) design of targeted libraries for the discovery of glycosyltransferase and sulfotransferase inhibitors, and (b) engineering Golgi enzymes for small molecule control of their subcellular localization. Representative examples are summarized below.
(a) Identification of ppGalNAcT inhibitors using a library screening approach
We have been particularly interested in small molecule inhibitors of the polypeptide N-acetylgalactosaminyltransferases (ppGalNAcTs), the enzymes that initiate mucin-type O-linked glycosylation (Fig. 4). Mucin-type O-linked glycosylation is as prevalent in higher eukaryotes as the better understood N-linked form of glycosylation, yet the functions of the posttranslational modification are largely unknown. Just as inhibitors of N-linked glycosylation (i.e., tunicamycin) have proven to be indispensable research tools, inhibitors of O-linked glycosylation would have a major impact on the field of glycobiology. None were available prior to this work.
All mucin-type O-linked glycans possess a core GalNAc residue that is appended to Ser or Thr residues by the ppGalNAcTs. We therefore targeted this enzyme family (24 isoforms predicted in humans) with the initial goal of identifying a broad-spectrum inhibitor. We designed a library of uridine analogs directed to the enzymes’ common UDP-GalNAc binding site and screened the 1300-member collection against several ppGalNAcT isoforms. Compound I-68A (Fig. 4) was identified as a general ppGalNAcT inhibitor that is active in cells and cultured murine organs. The compound has subsequently been used by several other labs for studies of O-linked glycan function.
Future directions include the development of second-generation libraries from which we hope to identify compounds with greater potency and selectivity for individual ppGalNAcT isoforms.
Figure 4. Mucin-type O-linked glycosylation is initiated by the 24 ppGalNAcTs in humans. Compound I-68A was identified from a uridine library screen as a broad-spectrum ppGalNAcT inhibitor that is active in cells and in organ culture.
(b) Small molecule modulators of Golgi enzyme localization
The Golgi-associated glycosyltransferases and sulfotransferases share a common anatomy comprising a luminal catalytic domain and a membrane-bound localization domain. The two domains can function independently but their fusion is required for the enzymes to colocalize with and then act upon glycoconjugates within the secretory pathway. We engineered the enzymes so that their separate domains can undergo an induced association in the presence of a small molecule drug (Fig. 5). The two domains were fused to the proteins FRB and FKBP, which bind simultaneously to the natural product rapamycin forming a high-affinity ternary complex. In the presence of rapamycin, the enzymes’ catalytic domains are colocalized with their substrates in the Golgi cisternae, whereas in the absence of the drug the catalytic domains are simply secreted from the cell. The drug thereby activates an otherwise inactive enzyme in a reversible and temporally controlled fashion. We used this approach to control the activities of glycosyltransferases and sulfotransferases in cell-based systems. Future directions include studies of tumor-associated glycans in metastasis models and extension to transgenic mouse systems.
Figure 5. A strategy for conditional activation of Golgi enzymes by small molecule-mediated assembly of catalytic and localization domains.
B. Mycobacterial Sulfation Pathways
Sulfated glycans are known to play important roles in cell-cell communication in higher eukaryotes. The associated sulfotransferases are considered to be regulatory enzymes in these processes. By contrast, the functions of sulfated metabolites and sulfotransferases in prokaryotes are less well understood. About six years ago we embarked on a project aimed at elucidating the roles of sulfation pathways in mycobacteria. The project was initiated by our discovery of four putative sulfotransferase genes in the M. tuberculosis genome, which we termed the stf family. Our immediate goals were then (1) to define the repertoire of sulfated metabolites in M. tuberculosis, (2) to identify the relevant sulfotransferases, and (3) to define their biological roles as they relate to pathogenesis.
1. Genetic and biochemical studies of Sulfolipid-1 biosynthesis
Prior to our work, only one sulfated glycoconjugate had been reported in M. tuberculosis, the putative virulence factor sulfolipid-1 (SL-1, Fig. 6). Its biosynthetic machinery was largely unknown, with the exception of a polyketide synthase (Pks2) that assembles the three methyl-branched fatty acyl side chains. In the past three years we have elucidated a majority of the SL-1 biosynthetic enzymes using primarily genetic approaches. We determined that Stf0 is the sulfotransferase responsible for sulfation of SL-1, catalyzing the first committed step in the biosynthetic pathway. This step is followed by two acyl transfer reactions that we recently discovered are mediated by the enzymes annotated as PapA2 and PapA1, respectively. A lipid transporter termed MmpL8, identified by Jeff Cox’s group at UCSF, flips the intermediate “SL1278” (Fig. 6) to the extracellular environment, where two as yet unidentified acyl transferases complete the assembly of SL-1. Further, we solved the X-ray crystal structure of Stf0.
Figure 6. Biosynthesis of SL-1. Trehalose is first sulfated by Stf0 to form trehalose-2-sulfate (T2S), which is then acylated by PapA2 to form SL687. This metabolite is then elaborated by PapA1 and Pks2 to form SL1278, which resides on the interior of the cell membrane. Transport of SL1278 to the cell’s exterior by MmpL8 is required for further elaboration to SL-1.
The abundance of stf genes suggested that M. tuberculosis produces several sulfated metabolites in addition to SL-1. In order to identify these sulfated compounds we formed a collaboration with Prof. Julie Leary (UC Davis) and developed a mass spectrometry-based approach to metabolome-wide profiling of bacterialextracts. We identified several new sulfated compounds including one with m/z 881.56, which we term S881. Recently, we obtained genetic evidence that S881 is the product of the sulfotransferase Stf3. Deletion of stf3 (and S881) caused a hastened progression of disease in a murine infection model, suggesting that S881 is a negative regulator of virulence. We speculate that the metabolite activates either the innate or adaptive immune response, thereby decelerating infection.
Future directions
We plan to probe the effects of sulfated metabolites on pathways relevant to immunogical activation. This will be accomplished using mutant M. tuberculosis strains lacking specific sulfated metabolites, with an initial focus on SL-1. Extracts from these mutants will be tested in cell-based assays of TLR function and CD1 signaling. In addition, the phenotypes of these mutants in murine infection models will be tested using transgenic mice that lack specific immunological pathways. In addition, we plan to expand our structural and mechanistic work to include the acyl transferases that participate in SL-1 assembly. Finally, we plan to elucidate the molecular structure of S881 and other sulfated metabolites that were identified in our metabolome screen.
2. Structural and mechanistic studies of sulfur assimilation enzymes
M. tuberculosis achieves a state of latency during which it can survive within host macrophages in a non-replicative state for decades. To persist in this environment, the bacteria must possess significant protection against macrophage-generated oxidants such as NO, superoxide and peroxide. Reduced sulfur-containing metabolites (i.e., thiols and sulfides) are known to fulfill this function in other organisms, thus we speculated that the enzymes involved in sulfur assimilation and reduction might be vital for survival of M. tuberculosis during latency.
We elucidated the pathways for sulfur assimilation in mycobacteria, as shown in Figure 7. We then identified APS reductase as an attractive target, as this enzyme represents the first committed step in the biosynthesis of cysteine, methionine, and various cofactors including the abundant mycobacterial reductant mycothiol. Indeed, we generated a mutant strain of M. tuberculosis lacking APS reductase observed attenuated virulence in a murine model of infection, validating APS reductase as a drug target. We have recently performed a detailed mechanistic analysis of this enzyme and solved its three-dimensional structure.
Figure 7. The sulfate assimilation pathway. Free sulfate is taken up by specific transporters, then activated by ATP sulfurylase (CysD) to form APS. A GTPase (CysN) supplies the energy for the reaction. APS kinase (CysC) phosphorylates APS to form PAPS, the substrate for sulfotransferases. APS is also reduced by APS reductase (CysH) using thioredoxin (Trx) as a shoichiometric reductant. The sulfite (SO32–) formed is further reduced to sulfide, and the sulfur is incorporated into cysteine. This amino acid is then converted to numerous metabolites, including methionine, cofactors, and the abundant antioxidant mycothiol.
Simultaneously, we focused considerable effort toward understanding the molecular details of the first step in sulfate activation, catalyzed by ATP sulfurylase. This enzyme is another attractive drug target, as it lies upstream of both the sulfur reduction and sulfation branches of the pathway in Figure 7. An unusual feature of ATP sulfurylase is its reliance on a G protein to supply the energy for an otherwise uphill reaction—formation of the phosphsulfate anhydride in APS from ATP. In order to better understand how GTP hydrolysis is coupled to ATP sulfurylation, we solved the X-ray crystal structure of the complex and performed kinetic experiments with enzyme mutants.
Future directions
We plan to develop small molecule inhibitors of ATP sulfurylase as potential drug leads. Our approach will combine library screening with structure-based optimization, followed by biological testing in cell-based assays and murine infection models.
C. Biocompatible Coatings for Carbon Nanotubes and Development of the Nanoinjector
Carbon nanotubes (CNTs) are molecular wires that are endowed with remarkable structural, electrical, and mechanical properties. Their potential applications in biology include sensing, imaging and scaffolding for cell growth, but are presently limited by chemical incompatibility of the CNT surface with biological components and their aqueous milieu. Indeed, unmodified CNTs are highly toxic to cultured mammalian cells, limiting their applications in living systems. To address this problem, we developed a biomimetic surface modification of CNTs using glycosylated polymers designed to mimic natural cell surface mucins (Fig. 8). The polymers were end-functionalized with lipid tails for self-assembly on the CNT surface through hydrophobic interactions. Mucin mimic-coated CNTs were soluble in water, resisted non-specific protein binding and bound specifically to biomolecules via receptor-ligand interactions. Importantly, we found that CNTs coated with the mucin mimics were non-toxic to cultured cells, opening new avenues for applications in biosensor and tissue engineering platforms.
Figure 8. Biocompatible coating for carbon nanotubes (CNTs). A lipid-terminated glycopolymer self-assembles on the CNT, forming a biomimetic coating that passivates the CNT against non-specific protein binding and reduces cytotoxicity.
In parallel, we are developing new technologies that integrate biocompatible CNTs. A major effort in this area focuses on the nanoinjector, a device that delivers small numbers of molecules to the cell’s interior without physiological damage. The nanoinjector comprises a single CNT attached to an AFM tip. The CNT is coated with a linker molecule that possesses a pyrene moiety on one end and cargo on the other, separated by a disulfide bond. Using the AFM, the modified CNT is positioned above a target cell and lowered until the CNT pierces the plasma membrane. Within the reducing interior of the cell, the disulfide bond cleaves thereby releasing the cargo. The CNT is finally retracted and the cell monitored. Unlike classic microinjection techniques that can damage cells, the nanoinjector causes no visible harm. We demonstrated proof-of-concept by nanoinjection of single fluorescent quantum dots into mammalian cells.
Future directions include nanoinjection studies with biomolecules such as proteins and plasmid DNA. In addition, we are exploring the use of mucin mimics as synthetic modulators of cell surface behavior.
Return to top
|
