By many estimates, the end of the antibiotic age is upon us. Bacteria can divide so quickly and the prevalence of antibiotics is so great that they’ve evolved resistance to the drugs targeting their key pathways, like cell wall synthesis. But what if there was a new way to target certain pathogenic bacteria – not with broad-spectrum antibiotics but by targeting proteins in your own cells? That very well might be the distant promise of new work in protein prenylation.
Microbes are incredibly clever and have evolved many mechanisms for evading, reprogramming, and exploiting your body’s cellular responses. Certain types of immune cells, like macrophages, recognize bacteria as foreign and eat them, engulfing the microbes in specialized compartments, which are acidic and filled with proteases that destroy the invaders (this process is called phagocytosis). As they are being enveloped, some pathogenic bacteria inject proteins into the host cell, which will reprogram the host cell’s responses.
For this to work, though, some of the bacterial proteins might need to be modified by the host cell’s machinery to work. Such is the case with Legionella pneumophila, the causative agent of Legionnaire’s disease. When inhaled, these bacteria invade and multiply in macrophages found in your lungs (1 in figure). As they’re phagocytosed, the bacteria secrete hundreds of proteins into the macrophages, many of which localize to membrane compartments in host cells, and this process is quite important for proliferation of these bacteria (2). Basically, the bacteria secrete soluble proteins, which are tagged with a particular lipid group by host proteins called prenyltransferases (3). In eukaryotes, prenyl groups for proteins come in two flavors – farnesyl, a 15-carbon chain, or geranylgeranyl, a 20-carbon train. These prenyl groups can control cellular localization of the modified proteins, trafficking between cellular compartments, further post-translational modifications, and interactions with other proteins (4).
The modified bacterial proteins contain a recognition code call the CaaX sequence, which is required for prenylation. Two host proteins catalyze most of the farnesyl and geranylgeranyl transfers in eukaryotes, and these proteins have been creatively named farnesyltransferase (FTase) and geranylgeranyltransferase I (GGTaseI), respectively. The transferases add their prenyl substrate to a cysteine on the target protein (the C of CaaX). “X” represents the target protein’s C-terminal residue, which confers some discrimination between FTase and GGTaseI, although there is some overlap in recognition. The “a” residues between the target’s cysteine and C-terminal residues are quite variable. These are the same recognition requirement for prenylation of host proteins. In other words, there’s nothing particularly special about the bacterial proteins; it’s just another way that pathogens co-opt the host system.
So this process is important for one particular type of bacteria. But how widespread is host-dependent prenylation of pathogenic proteins? Elia Wright, a doctoral student in Carol Fierke’s lab at University of Michigan, set out to understand this. Wright started looking for pathogenic proteins that had a CaaX motif. She focused her search on proteins that bacteria secrete as they’re phagocytosed, because she wouldn’t expect many other bacterial proteins to encounter the host prenyltransferases. Wright’s first step was to determine whether the sequences she identified could be modified by mammalian prenyltransferases.
Generating full-length proteins for biochemical studies can be a laborious process, requiring lots of time and resources to generate a few milligrams of protein, sometimes after months of optimization. Short peptide synthesis, by comparison, is consistent and straightforward. Plus a robust biochemical assay using fluorescently labeled peptides allowed Wright to define the kinetics of peptide prenylation. She used peptides with the sequence of the last 6 residues of the putative targets (which includes the CaaX sequence) and a dansyl group attached to the N-terminus. She mixed a peptide with a purified prenyltransferase and its lipid substrate and then monitored the fluorescence of the peptide. Dansyl is sensitive to the hydrophobicity of its immediate environment, so addition of the prenyl group to the peptide increases its fluorescence. Using this approach, Wright identified many peptides derived from pathogenic bacteria that were prenylated in vitro. The catalytic efficiencies for the bacterial peptides and those of host substrates were comparable, suggesting that the pathogenic ones might be able to compete with host proteins.
This assay told Wright that human FTase and GGTase I could modify the peptides, but she wanted to know whether the results would hold in a cellular context. So she expressed short peptides (in this case the last 15 residues of candidate proteins) attached to green fluorescent protein (GFP) in HEK293 cells. GFP can be detected by fluorescence microscopy, and its localization is generally control by what’s attached to it. So it provided an easy way to track peptide localization in cells. If a peptide wasn’t prenylated, as was the case for the known negative control myc, GFP was distributed throughout the cell. The C-terminal peptide of H-ras, a host protein that's farnesylated, served as the positive control, localized to plasma membranes. In this system, Wright’s peptide candidates could also cause GFP to localize to cellular membranes and compartments consistent with prenylation – Golgi, endoplasmic reticulum, and, in one case, even nuclear membranes.
Wright’s results suggest that a wide range of secreted proteins from pathogenic bacteria can be modified by human prenyltransferases, and they can even compete with endogenous host proteins. Wright used peptide substrates, but sequence and structural elements beyond the CaaX sequence can alter prenylation. Expression of full-length proteins might provide additional insight, but most systems for manipulating mammalian cells result in overexpression that doesn’t reflect what happens in nature. Ideally researchers would like to use a proteomics approach, a method that uses mass spectrometry to provide sensitive, global analysis of both host and pathogen protein prenylation. First, they need a way to capture and enrich prenylated proteins. A group at Max Planck Institute demonstrated the proof-of-concept for this approach for a specialized prenyltransferase, Rab geranylgeranyltransferase.
The Fierke lab teamed up with the lab of Richard Gibbs at Purdue University to identify novel substrates that might work with FTase and GGTaseI. It’s proving to be a challenge, as these enzymes seem to be refractory to most substrate modifications. Wright notes, “We are making strides in understanding what part of the natural farnesyl and geranylgeranyl diphosphate [substrate] compounds are recognized and which areas of the molecule can be chemically altered without changing the specificity of the enzyme, which is something that most people disregard when using [e.g. tagged or taggable analogs].” Furthermore, Wright notes that they are testing their analogs against arrays of peptides, rather than one or two selected sequences, so they can understand how the analogs compare to natural substrates. Sadly Dr. Gibbs recently passed, but the Fierke lab continues the work they started together.
There’s a long way to go before we’ll really understand the broader relevance of pathogenic protein modification by host prenyltransferases. Wright started with L. pneumophila but is now expanding her studies to peptides from other pathogens such as organisms causing food poisoning and tuberculosis. With a lot of patience, hard work, and a little luck, one day this might provide a new strategy for tackling tough infections.