Potential perils from bioterrorism to bird flu are increasingly pushing proteomics researchers to identify molecules involved in the infection process. Often stymied in characterizing all the proteins of a single organism, investigators must now contend with the complexities inherent in characterizing two intertwined, antagonistic organisms: host and pathogen.

"There's really just not a whole lot of proteomic work published yet" on host-pathogen interactions, says Sandra L. McCutchen-Maloney, biodefense proteomics group leader at Lawrence Livermore National Laboratory in Livermore, Calif., who recently coauthored a review of the literature.1 Prospects for such research are improving, however, thanks to technical advances and increased funding. If presentations at meetings are any indication, significant papers should emerge within a year, predicts Philip C. Hanna, an associate professor of microbiology and immunology at the University of Michigan Medical School in Ann Arbor.

Hanna directs one of seven biodefense proteomics research centers (BPRCs) that the US...


Most projects to date have focused on the pathogen's proteome, which is simpler than the host's and easier to manipulate genetically in follow-up studies. Findings about bacteria, in particular, are "more easily interpreted" because microbes regulate protein expression much more strictly than hosts do, says Eustache Paramithiotis, head of the BPRC at Caprion Pharmaceuticals in Montreal. But a pathogen-centered approach requires that a bacterium, virus, or parasite be able to thrive on its own or be extricable from its host. Otherwise, the host's proteome can swamp the pathogen's.

A host-centered approach, on the other hand, benefits from this imbalance, which renders host proteins easier to identify. But the most abundant – albumin in plasma, for example – must be selectively removed, or else they can mask the rarer signaling molecules thought to be crucial to the host's response to infection. Moreover, large discrepancies seem to occur between protein sets expressed by hosts of the same species. In an unpublished study of 12 people, McCutchen-Maloney found that about 200 plasma proteins varied enough to differentiate one subject from another.

Joshua N. Adkins, who manages the BPRC at Pacific Northwest National Laboratory in Richland, Wash., is initially focusing on the bacterium Salmonella typhimurium, though he eventually plans to scrutinize the macrophages that it infects. Adkins has cultured the pathogen in LB broth, which promotes its growth, and in magnesium-minimal medium (MgM), which mimics its intracellular environment. He has already cataloged more than 2,400 proteins, 200 of which are unique to the MgM condition. In a related project, his team is seeking to purify the intracellular vacuole that contains the bacteria and to examine the pathogen and host proteins there. Adkins reports that in early experiments, vacuole isolates have been contaminated by host-cell mitochondria, which are "about the same size and similar in structure."


McCutchen-Maloney has observed the reverse situation. "When we characterize the host response, oftentimes we actually find pathogen proteins," she says. "And so, in a sense, we are characterizing the pathogen." She intends to later target both proteomes simultaneously and systematically.

BPRC groups at the University of Michigan Medical School and Caprion Pharmaceuticals are already attempting such a feat. At the university, Hanna and Nicholas H. Bergman, a research assistant professor of bioinformatics, are surveying host and pathogen proteins expressed during the six hours after mouse macrophages are infected by Bacillus anthracis, the bacterium that causes anthrax. The short time-course of infection is advantageous, notes Bergman. "If you're going to get anything useful out of this sort of study, all the cells in the culture – or in the organism, if you're looking at an in vivo model – have to be roughly synchronized," he explains. "Otherwise the noise in the system drowns out any real signal."

In collaboration with John R. Yates III, a cell biology professor at The Scripps Research Institute in La Jolla, Calif., the Michigan team is compiling lists of proteins present in various samples; ultimately they will compare lists from different stages of infection and from infected versus noninfected cells. The team has already detected several thousand proteins, but, says Bergman, "we don't have nearly as many anthracis proteins identified as we'd like." The ratio of host to pathogen proteins is typically 20 to 1. To boost the B. anthracis signal, the team will gently lyse mouse cells under conditions that don't disrupt the bacteria and then will quickly segregate the bacteria by centrifugation.


© 2005 WILEY-VCH Verlag GmbH & Co.

After infecting human monocytes with the plague bacterium Yersinia pestis, researchers purified cytoplasmic (C), membrane (M), and nuclear (N) fractions from host cells. Proteins from each fraction were then separated by two-dimensional gel electrophoresis. The circled spots represent 16 proteins whose abundance increased or decreased more than 1.5-fold in infected versus uninfected cells. (From C.G. Zhang et al., Proteomics, 5:1877–88, 2005.)

Bergman and Hanna are also using cDNA microarrays to measure mRNA expression levels in host cells and bacteria. The reasoning behind the dual approach, Bergman explains, is that proteomics can overlook proteins that are present in minute quantities, and genomics doesn't indicate the extent to which transcripts are translated into proteins. "The thinking is that if we can combine the two, we'll miss a lot less," he says.

BPRC investigators at Scripps, however, view genomics less as an adjunct to proteomics than as an attention-focusing prelude. Benjamin W. Neuman, an assistant professor in the molecular and integrative neurosciences department, plans to infect human colorectal cancer and kidney cells with the SARS coronavirus to determine how its 26 gene products interact with each other and with host proteins. According to Michael J. Buchmeier, a neurosciences professor at Scripps, once genomic profiling is completed, "the second round would be to pick out what you considered to be critical genes – this is where experience comes into play – and to look at the expression of those in the cell, using some kind of assay," such as mass spectrometry. Earlier this year, a Scripps team, including Buchmeier, reported applying such an approach (with Western blotting and ELISAs substituting for spectrometry) to a mouse hepatitis coronavirus that infects neuronal and glial cells.4


At Caprion Pharmaceuticals, Paramithiotis has adopted another strategy, which he calls "organellar isolation," to focus his analysis of the vast raw data generated by proteomics technology. In collaboration with Michel Desjardins, of the Montréal University, his group studies Brucella abortus, a bacterium that causes the chronic human and animal disease brucellosis. After infecting mouse macrophages with this pathogen, researchers will withdraw samples periodically before the cells lyse two days later. Organelles and various cellular compartments will be stringently purified from each sample, and quantitative mass spectrometry applied.

Organellar isolation has two rationales. First, it lets investigators localize the pathogen and host proteins that mass spectrometry has identified. Thus, many B. abortus proteins should turn up in phagosomes; the bacterium replicates in these organelles, which sequester material that macrophages internalize from their environment. Second, organellar isolation allows for the detection of rarer proteins present in just a part of a cell. In contrast, "if you process an entire cell, you'll get the top 100 or 200 most-abundant proteins," notes Paramithiotis.

Despite his cell-fractionation protocol, he worries that the host's proteome will swamp the bacterium's. His group consequently plans on isolating bacteria from the host cell before processing them for analysis. And to make that analysis more computationally manageable, he says, "we won't catalog all the proteins found in an organelle; we'll only go after the ones that change" from sample to sample. Once these are identified, the group will explore how B. abortus infection is affected by knockouts or knockdowns of the genes encoding these proteins.

Paramithiotis hopes this work will uncover general mechanisms of infection that might apply to many dangerous bacteria with intracellular life cycles. McCutchen-Maloney conversely wants proteomics to implicate specific pathogens at early stages of infection, when symptoms are often nondescript and flu-like. At a recent meeting, she and colleagues reported that protein profiling could distinguish between samples of human blood incubated with any of seven pathogens, including B. anthracis and Y. pestis.5

Protein profiling, McCutchen-Maloney says, might soon be applied to exclude certain diagnoses and confirm others, if supplemented with throat and blood cultures and with PCR. But she notes that clinical trials are hampered because the United States lacks enough people exposed to these pathogens, and security and safety issues can prevent shipments of tissue samples from other countries. While expressing a wish for further studies, McCutchen-Maloney asserts, "I think that we are actually prepared to handle clinical samples today."

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?