Proteomics for biodefense applications: progress and opportunities

Abstract

The increasing threat of bioterrorism and continued emergence of new infectious diseases has driven a major resurgence in biomedical research efforts to develop improved treatments, diagnostics and vaccines, as well as increase the fundamental understanding of the host immune response to infectious agents. The availability of multiple mass spectrometry platforms combined with multidimensional separation technologies and microbial genomic databases provides an unprecedented opportunity to develop these much needed resources. An overview of current proteomic strategies applied to microbes and viruses considered potential bioterrorism agents is presented. The emerging area of immunoproteomics as applied to the development of new vaccine targets is also summarized. These powerful research approaches can generate a multitude of potential new protein targets; however, translating these proteomic discoveries to useful counter-bioterrorism products will require large collaborative research efforts across multiple basic science and clinical disciplines. A translational proteomic research paradigm illustrating this approach using influenza virus as an example is discussed.

IMAC-Cu: Copper-coated immobilized metal ion affinity chromatograph; SAX: Strong anion exchange; SELDI: Surface-enhanced laser desorption/ionization; WCX: Weak cation exchange.

Figure 1. Comparison of three different ProteinChip surfaces for FluMist™ serum day 0 versus day 4. A representative gel view of SELDI spectra is shown for a serum specimen (1 µl) from one FluMist vaccinee comparing day 0 and day 4 samples on three chip surfaces: SAX, WCX and IMAC-Cu. The protein chips were analyzed on a PBS-II SELDI mass spectrometer (Ciphergen Biosystems).

IMAC-Cu: Copper-coated immobilized metal ion affinity chromatograph; MW: Molecular weight; PLUNC: Palate, lung and nasal epithelial clone; SELDI: Surface-enhanced laser desorption/ionization.

Figure 2. Representative SELDI spectra of nasal swab extract proteins post from a FluMist vaccine, day 0 versus day 1. Nasal swab extracts were applied to IMAC-Cu chip surfaces and processed for SELDI analysis. Proteins attached to the nylon nasal swabs were solubilized in a 8M urea/1% CHAPS solution and 1 µl of fluid was loaded. The indicated arrows highlight the differences in the scale of the peak intensities. The box indicates a different peak pattern between the two samples. Tandem mass spectrometry sequence identification of the three main peaks from day 1 following sodium dodecyl sulfate polyacrylamide gel electrophoresis indicated two unknown membrane proteins and PLUNC.

• •2D polyacrylamide gel electrophoresis (PAGE) with mass spectrometry (MS) identification is still the most accessible approach for characterizing the proteomes of bacterial pathogens, and improved 2D sodium dodecyl sulfate PAGE methodologies, such as differential gel electrophoresis, will further refine and extend the information gathered from the approach.
• •Increased applications of isotope-coded affinity tag methods and Fourier transform ion cyclotron resonance MS for the characterization of bacterial proteomes will further expand the dynamic range of bacterial proteomes and greatly facilitate the discovery of new therapeutic and vaccine targets.
• •Immunoproteomics applications represent a potentially powerful convergence of clinical, proteomic and genomic resources to develop improved vaccines for different pathogens.
• •Proteomic profiling of serum/plasma from vaccinated or pathogen-exposed individuals using matrix-assisted laser desorption/ionization time-of-flight MS and surface-enhanced laser desorption/ionization time-of-flight MS is a largely untapped biomarker discovery and diagnostic approach that should be aggressively pursued, particularly for viruses.
• •Collaborative multi-institution, multidiscipline and multiple technology efforts are needed for effective clinical study design, sample collection and sample analysis for diagnostic biodefense pathogen assay development.

While the potential large-scale use of biologic weapons has existed since the end of World War II, the anthrax letter attacks of 2001 in the USA ushered in a new and immediate need for improved countermeasures against bioterrorism agents. For the purposes of this review and the research descriptions herein, the term bioterrorism will be used as defined in the National Institute of Allergy and Infectious Diseases (NIAID) Strategic Plan for Biodefense Research as ‘the use of microorganisms that cause human disease, or of toxins derived from them, to harm people or to elicit widespread fear or intimidation of society for political or ideologic goals. From a scientific and medical perspective, this form of terrorism is best seen as a variant of the general problem of emerging infectious diseases, the only difference being that increased virulence or spread into a susceptible population is a deliberate act of man rather than a consequence of natural evolution’ [101]. The key to effectively counter these bioterrorism agents lies in the development of new rapid diagnostic tests, new vaccines and immuno-therapies for prevention, and new drugs and biologics for treatments. As illustrated by the large numbers of potential bioterrorism agents included on the NIAID Category A–C Priority Pathogen list [102], a substantial investment in biomedical research on the properties of these pathogens, and the immune response to them, is required. In the USA alone, the 2004 NIAID biodefense research budget for biomedical research exceeded US$1.5 billion. The allocation of these types of resources to biomedical research offers the potential to further develop and utilize novel technologies. In this regard, few emerging technologies offer as much promise as those encompassed by the term proteomics, a biomedical research area that will increasingly provide new solutions and treatments against bioterrorism agents.

The goal of this review is to summarize the methodologies and experimental rationale of successful proteomic approaches that have already been accomplished in the context of bioterrorism issues, primarily using anthrax and other bacteria as examples. The emerging area of immunoproteomics will also be addressed. Despite the enormous potential application of proteomics to clinical issues for biodefense and infectious disease research, in general, there are relatively few publications in this area in relation to other diseases such as cancer. The latter sections of the review address this issue, including discussion of applying proteomics to samples related to influenza virus infections and vaccinations. The nature of natural influenza infections, and its potential use as a bioterror weapon, make influenza a model paradigm system for biodefense proteomic applications.

Multidimensional separation & MS approaches to pathogen proteomics

The monitoring of proteomic differences in bacterial and viral pathogenesis can allow direct comparisons of strain variability, severity of infection, environmental influences and the effects of genetic manipulation [1–4]. This has been primarily carried out with pathogenic bacteria using 2D gel electrophoresis, usually involving strains exhibiting diverse phenotypes including anti-viral drug resistance, altered degrees of infectivity and pathogenicity, different growth conditions and differential genotypes. Low mass range protein display methods such as matrix-assisted laser desorption/ionization (MALDI) time-of-flight (TOF) mass spectrometry (MS) and surface-enhanced laser desorption/ion-ization (SELDI)-TOF-MS are also increasingly being applied to microbial systems [5–7]. Comparative proteomic techniques such as isotope-coded affinity tags (ICATs) [8,9], as well as different multidimensional chromatography systems [10,11], are being utilized as front-end steps prior to MS. More recent MS platforms such as Fourier transform ion cyclotron resonance (FTICR)-MS have also been applied to microbial systems [12]. In practice, some type of lysate or fraction derived from Escherichia coli is frequently utilized in the characterization of different MS platforms, thus it is not surprising that most proteomic publications for biodefense pathogens involve the differential characterization of multiple bacterial proteomes.

In Table 1, a summary and representative list of different NIAID Priority Pathogens analyzed by some type of multi-dimension separation or comparative display method followed by MS sequencing of differentially expressed proteins is presented. This list is not intended to be exhaustive in scope, just reflective of more recent proteome characterization studies for the indicated Priority Pathogens included in Table 1. One of the most common approaches used for the studies listed in Table 1 involves a 2D separation step. Differentially expressed protein targets are identified, excised and eluted (if from gels), digested with trypsin, and the amino acid sequences of the tryptic peptides determined using different MALDI instrument configurations, electrospray ion-trap mass spectrometers or some type of hybrid instrumentation. In contrast, more sophisticated affinity-based technologies such as ICAT are underrepresented in Table 1, but will likely be increasingly applied to microbial systems. Overall, the types of studies listed in Table 1 illustrate a range of examples of what can be achieved with different front-end separation and comparison methods, and no single method is currently superior. Within the current framework of the rapidly evolving area of proteomic technologies, the choice of method to apply is largely dependent on budgetary and proteomic resources available to individual investigators at their given institutions.

As the organisms in Table 1 are priority pathogens, these become candidate organisms for genomic sequencing efforts. As illustrated most effectively with the human genome and its linkage with proteomic tryptic database search engines that facilitate direct peptide sequence identities, having a bacterial genome database for each organism being characterized for differential protein expression greatly enhances the success of these efforts. In this regard, there are a multitude of genomic database resources for microbes in existence, and these will continue to rapidly evolve as other bacterial genomes are added. These databases include those available from The Institute for Genomic Research (MD, USA) [13], the Max Planck Institute (Germany) [14], and other sites accessible via the internet, all of which are also summarized in a separate publication [15]. Another critical evolving resource is searchable 2D polyacrylamide gel electrophoresis (PAGE) image databases for different pathogens and different host cell types, typified by SWISS-2D [16]. These include the identities and migration position of already characterized proteins in the reference gels. Similar, but less prevalent, ICAT reference databases for different organisms are also being established. These types of database resources will greatly decrease the redundancy of sequencing every protein on a 2D gel or ICAT analysis. Lastly, these comparative bacterial proteome studies could be analyzed using another emerging technology, 2D differential in-gel fluor-escence electrophoresis (DIGE) [17,18]. This involves differential labeling of two related protein samples with different colored dyes, a separate 1:1 sample mixture labeled with a third reference dye, followed by separation of the mixture on standard 2D gels. This has already been reported for analysis of E. coli [18], and will likely play an increasing role in these types of bacterial proteome characterization studies.

MALDI-TOF-MS applied to biodefense

Using SELDI-TOF protein-capture chip surfaces [19,20], or MALDI-TOF with direct application of sample to a spot plate, simultaneous analysis of the population of proteins present in complex biologic materials can yield a profile unique to that specific sample. In contrast to 2D gel strategies, the SELDI and MALDI approaches are more rapid, have high-throughput capabilities for automated assay development, require orders of magnitude lower amounts of the protein sample, and can effectively resolve low-mass proteins (2000–20,000 Da). This restricted mass range can be a disadvantage when comprehensive proteomic analysis is required, an approach much better accomplished with 2D gel or FTICR-MS methods. While the mass values of multiple potential biomarkers can be identified with the SELDI approach, a major limitation of SELDI is that it cannot be used efficiently for direct amino acid sequence

Table 1. Summary table of 2D protein separations of National Institute of Allergy and Infectious Diseases Priority Pathogen agents coupled with mass spectrometry sequencing.

Species	Purpose	Method	Ref.
Bacillusanthracis	Membrane antigen identificationsEndospore proteome Membrane antigens Vaccine lot comparisons	2D-PAGE2D-LC2D-PAGE2D-PAGE	[11,42,43,61]
Brucellamelitensis	Vaccine vs. wild type strain	2D-PAGE	[62]
Burkholderiacepacia	Quorum sensing mutant vs. parent	2D-PAGE	[63]
Campylobacterjejuni	Planktonic vs. biofilm growth	2D-PAGE	[64]
Coxiellaburnetii	Proteome of lysates	2D-PAGE	[65]
Escherichia coli O157:H7	Virulent vs. nonvirulent strains	2D-PAGE	[66]
Francisellatularensis	Subspecies strain comparisonsVaccine strain vs. wild type	2D-PAGE2D-PAGE	[67,68]
Listeriamonocytogenes	Exponential vs. stationary phasesCell wall subproteome	2D-PAGE differential salt extracts	[69,70]
Mycobacterium tuberculosis	Aerobic vs. anaerobic growthStrain and method comparisons	2D-PAGE2D-PAGE/ICAT	[71,72]
Rickettsiaprowazekii	Tryptic proteome	2D-LC	[73]
Salmonellatyphimurium	Antibiotic resistancePathogenecity mutantRecombinant protein comparisons	2D-HPLC2D-PAGE2D-PAGE	[10,74,75]
Severe acute respiratory syndrome (coronavirus)	Virions and infected cell lysates	2D-LC	[76]
Toxoplasmagondii	Tachyzoite stage expression	2D-PAGE	[77]
Vibriocholerae	Aerobic vs. anaerobic growthStrain comparisons	2D-PAGE2D-PAGE	[78,79]
Yersiniapestis	Human macrophage response comparison with other Yersiniaspp.	2D-PAGE	[80]

HPLC: High-performance liquid chromatography; ICAT: Isotope-coded affinity tag; LC: Liquid chromatography; PAGE: Polyacrylamide gel electrophoresis.

determinations of the biomarker candidates, necessitating the use of other strategies for this purpose. New generations of MALDI-TOF/TOF instrumentation are emerging that facilitate identification of prevalent peptide fragments less than 4000 mass-to-charge ratio [21,22], thus increasing the types of proteomic profiling strategies that can be applied to biodefense pathogens. An example of applying MALDI to biodefense pathogen characterization is summarized in the following paragraphs, and an example of SELDI applications is presented in the next section.

Multiple MS-based studies have been reported for character-ization of the unique sporulation and vegetative properties of Bacillus spp., particularly for Bacillus anthracis [1,4,5,11,23–26]. B. anthracis strains are found throughout the world; however, this wide geographic distribution is not reflective of great genetic diversity except for documented variable number tandem repeated sequences and single nucleotide polymorphisms used in phylogenetic relationships [27,28]. While these genetic differences are important to further understand the pathogenesis of B. anthracis, proteomic methods can be applied to the identification of proteins that are differentially expressed under various culture conditions and during the course of infection. From a forensics and biodefense perspective, proteomic profiling approaches may be able to identify unique protein signatures that are specifically related to spore culture conditions as well as differences in virulence between strains of B. anthracis. In B. anthracis, the spore coats are surrounded by a hydrophobic, balloon-like glycoprotein shell termed the exosporium [29]. Multiple studies have described different protein components of the exosporium specifically, and these proteins include a collagen-like structural protein termed BclA, other intergral membrane glycoproteins, and multiple embedded soluble proteins such as racemase and superoxide dismutase [26,30,31]. In the most comprehensive analysis thus far, over 750 different proteins in the endospores of B. anthracis Sterne were identified by multidimensional chromatographies and tandem MS sequencing methods [11]. Given the complexity and growth variability of the spore proteome and the many strains of B. anthracis, higher throughput MALDI profiling strategies could provide a broader and complementary information base for developing countermeasures against these diverse anthrax strains.

For example, MALDI-TOF analysis of tryptic fragments of small acid-soluble spore proteins of Bacillus spp. has proven to be diagnostic for differences within this species [1]. Another MALDI-TOF study reported that differential profiles of low-mass peptides/proteins could be determined when spore proteins from different B. anthracis strains were compared [5]. Two recent reports evaluated the ability of a conventional MALDI-TOF approach and a new hybrid ion-trap MALDI-TOF instrument to rapidly separate and identify mixtures of peptides derived from limited tryptic proteolysis of mixed spores from five Bacillus spp. [24,25]. Following on-probe digestion with immobilized trypsin, cleavage products of a limited set of bacterial proteins with molecular masses of approximately 4–125 kDa were obtained within 20 min, and bacterial peptides suitable for isolation and high-energy fragmentaion analysis were generated within 5 min. These sequenced peptides allowed rapid identification of the most abundant proteins present and their bacterial sources using standard database searches. Species-specific tryptic peptides could be generated from each of the Bacillus spp. studied [24]. In a related study, a novel quadrupole ion-trap TOF-MS was used to analyze the peptide sequences generated from the proteo-lyzed spore mixtures [25]. It was reported that using the method of on-probe solubilization and in situ proteolytic digestion of small, acid-soluble spore proteins, the different species present in the mixture could be identified in less than 20 min. This hybrid instrument resulted in a mass resolving power of 6200 on the MALDI, and a mass accuracy of up to 10 parts per million using an ion-trap TOF tandem configuration. Sequence-specific information on isolated protonated peptides stored in the ion trap was gained via tandem MS experiments with an average mass resolving power of 4450 for product ion analysis [25]. These cumulative MALDI-TOF studies illustrate the potential of applying mass spectrometers to potential field applications to quickly resolve and identify complex mixtures of microbes reflective of a given environment. As more genomic information becomes available for different pathogenic bacteria, as well as bacteria presenting normally in a given system, this approach could be critical for multiple biodefense applications and emergency first responder scenarios.

SELDI-TOF-MS applied to biodefense

SELDI-TOF-MS technology has recently been developed to facilitate protein profiling of complex biologic mixtures [19,20]. This modification of MALDI-TOF technology uses ProteinChip arrays coated with a chemical surface (e.g., ionic, hydrophobic or metal) to affinity capture protein molecules from complex mixtures. Retained proteins are subsequently analyzed by TOF-MS. With the aid of SELDI software, a retentate map is generated depicting the mass-to-charge ratio, which corresponds to the molecular weight. When this process is expanded to many hundreds of samples, population-specific protein expression profiles can be deduced that are characteristic of the assayed group. The result is a fingerprint pattern unique for the designated group.

In 2003, a new strain of coronavirus (CoV) was identified as the cause of severe acute respiratory syndrome (SARS), which infected over 8000 individuals and led to over 750 deaths worldwide. Five recent studies have applied proteomic profiling methods for analyzing serum or plasma cohorts collected from a subset of SARS infected patients in an effort to identify early detection and prognostic biomarkers [32–36]. In three of these studies, SELDI-TOF protein chip profiling was used with distinct sera or plasma cohorts [33–35]. In the largest reported study, serum samples were separated into acute SARS (n = 74; <7 days after onset of fever) and non-SARS (n = 1067) cohorts [35]. The large non-SARS cohort included samples indicative of fever and influenza A (n = 203), pneumonia (n = 176), lung cancer (n = 29) and healthy controls (n = 659). Each sample was incubated with weak cation ProteinChips (Ciphergen Biosystems) followed by SELDI-TOF spectra generation. No peak identities were determined, but a panel of four biomarker peaks could detect 36 of 37 (sensitivity 97.3%) acute SARS and 987 of 993 (specificity 99.4%) non-SARS samples. These same four peaks could also be used to distinguish acute SARS from fever and influenza cohorts with 100% specificity (187 of 187). It was concluded that this approach could form the basis for a serum proteomic profiling assay for the early detection of SARS infections [35]. In a separate SELDI study, the profiles of 89 longitudinal sera samples collected from 28 SARS patients were compared with 72 sera from control patients without SARS [33]. A total of 12 distinct protein peaks were identified as being differentially diagnostic for SARS, one of which was serum amyloid A (SAA). Subsequent SAA concentration determinations in 45 longitudinal serum samples found a good correlation of SAA concentration with the extent of pneumonia in a small subset of severe SARS cases [33].

What are the likely identities and functional properties of the different serum protein markers that are being identified by SELDI? Are these peaks only representative of acute-phase reactants, as has been a consistent criticism of the proteomic profiling of serum approach [37], or do the peaks reflect innate immune responses or pathogen-derived proteins? There have not been sufficient studies to definitely determine the answer to these questions, and it is possible that the peaks are representative of each possibility. Without fractionation and removal of major serum and plasma proteins prior to SELDI analysis, it is most likely that the differential markers reflect acute-phase responses; however, this does not preclude them as being useful for diagnostics and/or distinct for a particular type of viral infection. For example, in a 2D gel study of plasma samples from four SARS patients, the majority of differentially expressed proteins were identified as acute-phase proteins, including a novel marker, peroxiredoxin-II secreted by T-cells [36]. The authors hypothesize that these types of T-cell-derived markers could reflect the innate immune signaling cascades resulting from the SARS-CoV infection. Much work remains to be done in the identification of these low-mass serum biomarkers, and application of complex body fluid-derived mixtures to hybrid ion-trap MALDI instruments as described for the spore protein mixtures could facilitate these efforts [25].

Immunoproteomics

It follows from the different protein display and identification studies mentioned previously that these methods could be funneled toward, or directly adapted to, development of improved vaccines by identifying the antigenic components of different pathogens. The immunoproteome for a given pathogen consists of all identified antigens present in an infected host [38–40,82]. Recognition of every potential epitope derived from the pathogen’s genome does not appear to be required for an effective immune response, as this occurs against a subset of antigens and epitopes that provide the necessary protection/neutralization [38]. 2D electrophoresis and blotting of whole-cell lysates (or membrane-enriched fractions) provides a display method to identify clinically relevant subsets of antigens following incubation of the blots with pathogen-exposed sera samples. Most in vivo antigens for that particular pathogen can thus be identified following MS sequencing. High-resolution 2D electrophoresis and unambiguous identification are prerequisites for reliable results. After statistical analysis, the resulting antigens are candidates for diagnostic assay or vaccine development and/or targets for therapy [38–40,82].

Specific to the priority pathogen list, different immuno-proteomic studies have been reported for Francisella tularensis [41], anthrax [42,43], Shigella [44] and Mycobacterium tuberculosis [45]. For example, the attenuated live vaccine strain of F. tularensis was used to generate whole-cell lysates, integral membrane protein fractions and basic protein fractions that were separated on 1- and 2D gels, then transferred to nitrocellulose [41]. Sera collected from patients suffering from tularemia was used to probe the immunoblots, and compared with control sera from healthy donors and sera from patients with Lyme disease. From this approach, 80 potential antigenic spots were identified, and a smaller subset was selected for MS sequencing analysis. In the sera from patients with tularemia it was found that the predominant antigenic species were different variants of 60- and 10-kDa chaperonins isolated from the integral membrane and whole-cell lysates [41]. Another example has been recently described for anthrax (B. anthracis) using sera from infected animals as the antibody sources [42]. This study was unique in that it described a predictive computational screen of the anthrax genome to identify vaccine candidates, then compared these results with the functional immunoproteomic assay. Six out of eight proteins in this in vivo screen had not been previously identified as antigenic, and five of the eight proteins had been predicted in the computational screen. This study illustrates how combining all resources available for a particular antigen (e.g., genomic, proteomic, immunologic or in vivo infection model) can generate novel vaccine candidates [42]. For each priority pathogen, developing the assays and systems to provide the capability to perform this type of comprehensive experimental approach should be emphasized.

Another immunoproteomic approach focuses on characterizing the pathogen-derived peptides bound to major histocompatibility complexes (MHC) on antigen-presenting cells that elicit effector T-cell responses to the pathogens. Following immunoaffinity purification and dissociation of bound peptides from the MHC complexes, the peptides are sequenced by tandem MS. The identified peptides thus represent potential vaccine candidates for that pathogen. This approach has recently been comprehensively reviewed [46,47], and has the potential to be highly effective when coupled with other comprehensive analysis strategies as described in the preceding paragraph.

Influenza as a clinical biodefense paradigm system for proteomics

The authors’ own proteomic efforts in biodefense research within the next 5 years will center on developing diagnostic assays, improving influenza vaccine strategies and comprehensively characterizing the immune response to influenza virus infection. The authors believe that the human influenza virus is an ideal model for the comprehensive proteomic characterization of a virus important to biodefense/infectious disease threats. Why influenza virus as a paradigm? This is based on multiple considerations:

• •Bioengineering of the influenza virus to generate viral strains never previously seen in the human population remains a looming bioterrorism threat [48]. Additionally, influenza strains could be engineered to be drug resistant to current anti-influenza drugs. Introduction of a strain such as this could have devastating consequences, essentially creating supercarriers of infection that would spread rapidly through the immune-naive human population. This is not a realistic scenario at present, as only a few laboratories possess the requisite tools to generate recombinant virus stocks. However, this is likely to change within the next 5 years, and no guarantees can be made that the technology will not end up in the possession of bioterrorists.
• •Containing natural influenza infections still remains a daunting challenge. Influenza is a leading cause of catastrophic disability, greatly affecting the quality of life of elderly persons [49,50]. In the USA alone, an estimated US$10 billion is spent annually due to the impact of influenza [51], and this cost will rise as the population of senior citizens rapidly expands [52].
• •Influenza morbidity and mortality is realized primarily in older adults and caused by the immune response to influenza virus. Specifically, elevated levels of cytokines are associated with influenza symptoms, including fever and headache [53,54]. The host immune response and viral pathogenicity are quite variable between pathogens. Current influenza vaccines are cost effective, but far from perfect; up to 61% of vaccinated elderly people still acquire influenza infection [55]. An antibody response to vaccine declines with age, and the mechanism responsible for this decline remains elusive [55,56]. A better vaccine, as well as early diagnosis and novel treatment targeting the harmful immune response, will bring huge advances in disease prevention and containment beneficial for both biodefense issues and managing natural disease outbreak. Proteomic strategies are thus key to making these needs a reality.

Attaining these goals should be readily feasible, as a whole repertoire of reagents and models are available for influenza research, including well-defined viral stock preparation methodologies, cell line and animal infection models. Clinically, millions of individuals are vaccinated against influenza and millions more are infected naturally each year. Challenge strains, defined antibody detection assays and clinically useful antiviral agents are also available. Cumulatively, obtaining a statistically significant number of research samples related to influenza infections and/or vaccinations will be straightforward if their collection is incorporated into study protocols. For example, the authors’ preliminary experiments indicate that elderly adults have a lower TH1 T-cell response to influenza vaccine than young adults, and the reduced TH1 response is proportional to the reduced antibody response [57]. The mechanism of the age-associated decline in the TH1 response, and the precise cause of the TH1 senescence leading to the reduced antibody response in elderly people, is an area for future research. Using different proteomic analyses applied to serum and immune cell isolates, the authors hope to identify surrogate markers associated with either the antibody and/or T-cell response, and then characterize these surrogate markers and investigate their roles in immune senescence. Additionally, in the event of a bioterrorist attack or natural outbreak of infectious disease (such as SARS), early diagnosis is pivotal for treatment and containment of outbreak. The authors believe that characterizing and identifying host immune responses to infections can be used for early diagnosis as well as new treatment strategies targeting any harmful aspects of the host immune response. Influenza is an ideal system for applying current and emerging proteomic technologies to accomplish these goals. Two examples from a recent pilot study of proteomic profiling strategies applied to vaccinated subjects are presented in the next section to illustrate how effective this approach could be for clinical biodefense studies.

Pilot study: proteomic applications to clinical samples from FluMist vaccinees

At the Glennan Center (VA, USA), six healthy young volunteers (21–30 years of age) were recruited, and received the live virus FluMist vaccine intranasally. Serum and nasal swabs were obtained from each subject immediately before (day 0) and on days 1, 2, 4, 7 and 14 post vaccination. For serum, dramatic differences in the SELDI profiles were observed, particularly at day 4 compared with day 0. On all three chip surfaces, over 25 distinct proteins were significantly (p < 0.05) over- or underexpressed in day 4 sera from all six Flumist-vaccinated subjects. In Figure 1, the 3–12 kDa gel view comparison of a day 0 and 4 FluMist recipient is presented for each of the three chip surfaces. These peaks reflect transient increases and decreases on day 4 that rebound to near day 0 values by day 7. Besides further highlighting the changes at day 4, this figure also illustrates how using multiple chip surfaces increases the available number of potential biomarkers that could be targeted for further identification and sequencing.

For the nasal swab samples obtained at the same time as the sera samples, not surprisingly only samples from day 1 or 2 post-FluMist administration indicated any differences in the protein profiles relative to day 0 baseline profiles. A representative profile from a day 1 versus day 0 FluMist recipient is shown in Figure 2. Note the large difference in intensity scale between the two samples, as there was significant upregulation of proteins in the 11–16 kDa range and in the lower mass region within the box (5–8 kDa). Four of six FluMist recipients had similar responses at day 1, and these proteins returned to baseline after 2 days (data not shown). An 8–16% sodium dodecyl sulfate-gel was used to separate the swab fluids from a day 1 FluMist recipient. Three bands of approximately 10, 14 and 16 kDa were excised from the gel, protein eluted and trypsinized, and applied to a LCQ DECAXP ESI mass spectrometer (ThermoFinnigan). For the 10 kDa band, three nonredundant peptides matching human palate, lung and nasal epithelial clone (PLUNC) were found. PLUNC is from a newly discovered gene family similar to human bactericidal/permeability-increasing protein and other mammalian lipopolysaccharide-binding and lipid transport proteins [58,59]. For the 14 and 16 kDa proteins, the sequence identifications were more hypothetical, identifying two putative membrane proteins of unknown function.

In summary, the intent of presenting these pilot SELDI studies from FluMist vaccinees was to illustrate that there are clearly distinct and detectable biologic differences present in serum and nasal swab protein extracts. Whatever proteomic platform is available to a particular investigator, incorporating longitudinal collection of body fluids during vaccination (or treatment) trials for any pathogen should be considered as these fluids represent a largely uncharacterized reservoir of potential biomarkers for vaccine efficacy, treatment response, disease progression and other applications.

Expert opinion & five-year view

In the context of utilizing proteomics and related resources for biodefense applications, there is reason for great optimism, as well as reasons for great concern. There has been excellent progress in applying and developing the most innovative and advanced proteomic resources for application to bacterial pathogens, particularly anthrax. A clear convergence of proteomic, genomic and immunologic information is evident that holds great promise for the design of improved vaccines and identification of new treatment targets. The mechanistic and functional information gained from these studies will positively impact many other areas of human health research, and the eventual collateral benefits of the proteomic methodologies developed for biodefense applications could be enormous when applied to other, less pathogenic bacterial infections in humans, animal and plant diseases, and different environmental systems. These statements are based on currently proteomic technologies, and since this is one of the most rapidly evolving areas in biomedical research, there is no reason not to expect that even better methods and instrumentation will quickly emerge for biodefense applications.

In contrast to the progress for bacterial pathogens is the application of proteomics to viral pathogens. There is an embarrassingly sparse body of literature in this field, even if studies related to HIV are included, and this should be a great cause for concern in the context of bioterrorism threats and public health in general. There are almost as many proteomic-related reports evaluating clinical specimens from SARS patients [32–36] as those reported for HIV, influenza and all other viruses on the priority pathogen list combined. Even inclusion of cell line- or animal model-related proteomic studies does not significantly alter this statement. Clearly, an improved strategy for applying proteomics to viral infections is needed. Hence, the discussion of influenza as a paradigm system was included in an attempt to initiate and encourage these types of studies.

In the authors’ opinion, this situation reflects a largely reactionary research viewpoint to whatever infectious disease or bioterror pathogen is currently in the news. On one hand, this can be beneficial, as there is no argument that increased understanding of anthrax infection and development of countermeasures was needed. Hopefully, the types of studies referenced herein illustrate the progress and great potential benefits derived from proteomic analyses of anthrax. On the other hand, why are there more proteomic reports for SARS, which so far has resulted in far less mortality than a typical flu season in the USA, than other viral pathogens that are more urgent biodefense threats? Do we need to wait for the avian flu to finally adapt to a more virulent human strain, or worse, wait for a bioengineered strain to be released, before initiating intensive studies? This is not an argument for forgoing critical continued research on SARS, but to encourage increased applications of proteomics to other more prevalent and/or morbid viruses. The comprehensive approaches (i.e., proteomic, genomic and immunologic) applied to anthrax research can readily be adapted to studies of influenza virus, HIV, and many other viruses on the priority pathogens list.

In the context of biodefense applications, the next 5 years should bring a wealth of emerging and rapidly expanding resources to facilitate increased proteomic applications to bacterial and viral systems. The multidimensional separation methods coupled to MS analysis of tryptic peptides for the biodefense-related bacterial proteomes listed in Table 1 highlights the utility of a proteomic approach, one which will likely continue to find increased uses as different front-end separation technologies emerge. Obviously, the more microbial/viral genomes available for different species, the easier it will be to perform functional proteomic studies. The increase in 2D-PAGE gel reference sites will also greatly facilitate these efforts, and similarly, ICAT reference sites will be equally critical. This path is clear, the application of DIGE and a host of future ICAT approaches can be readily accomplished. As more access to FTICR-MS instrumentation increases, coupled with appropriate genomic databases, there is an unprecedented opportunity to fully characterize the proteomes of individual bacterial pathogens, as well as the host response to viral and bacterial infections.

For the more clinical application of proteomics, typified by SELDI and MALDI analysis of blood fluids, there are some lessons that have already been learned from applying these technologies to developing cancer diagnostics [20,60], including a recent study evaluating viral hepatitis conditions and liver cancers [81]. Thousands of proteomic serum analyses have been performed for cancer diagnostics, and these cumulative experiences have highlighted several areas that are needed for improving these types of profiling studies. These include establishing uniform sample acquisition, processing and storage protocols, resolving MS instrumentation and peak sensitivity/resolution issues, as well as improving data analysis tools. The fact that all of these issues are readily addressable, and are actively being pursued across multiple clinical, academic, biotechnology and biopharmaceutical levels, is highly encouraging. These issues are not specific to cancer-related studies, and will be equally applicable to biodefense studies.

In order for large-scale proteomic profiling studies to be accomplished, or initiated, on readily available clinical specimens associated with biodefense pathogens such as influenza, there are additional considerations to those mentioned above. Whenever clinical trials are being designed, prospective sample collection should be included in the study design, particularly if a blood draw is already a likely component. Depending on the study, other available fluids should also be collected (e.g., urine, nasal fluids and saliva). Another approach would be to use archived samples from previous clinical trials. Either way, what is necessary to accomplish these types of studies is to create an integrated collaborative framework of protein chemists and mass spectroscopists, sample acquisition and biorepository staff, biostatisticians and epidemiologists, clinicians/pathologists and patient co-operation and consent. A deficiency in any of these individual categories will compromise the outcome of the entire project. At the assay level, there is a need to continue to develop high-throughput and reproducible protein fractionation procedures to identify potential low-concentration protein biomarkers. The key to achieveing this is to develop inter- and intrainstitutional translational research groups to bring together the necessary resources to capitalize on the immense promise proteomics technologies have in the application of biodefense related research.

Table 1. Summary table of 2D protein separations of National Institute of Allergy and Infectious Diseases Priority Pathogen agents coupled with mass spectrometry sequencing.

Species	Purpose	Method	Ref.
Bacillusanthracis	Membrane antigen identificationsEndospore proteome Membrane antigens Vaccine lot comparisons	2D-PAGE2D-LC2D-PAGE2D-PAGE	[11,42,43,61]
Brucellamelitensis	Vaccine vs. wild type strain	2D-PAGE	[62]
Burkholderiacepacia	Quorum sensing mutant vs. parent	2D-PAGE	[63]
Campylobacterjejuni	Planktonic vs. biofilm growth	2D-PAGE	[64]
Coxiellaburnetii	Proteome of lysates	2D-PAGE	[65]
Escherichia coli O157:H7	Virulent vs. nonvirulent strains	2D-PAGE	[66]
Francisellatularensis	Subspecies strain comparisonsVaccine strain vs. wild type	2D-PAGE2D-PAGE	[67,68]
Listeriamonocytogenes	Exponential vs. stationary phasesCell wall subproteome	2D-PAGE differential salt extracts	[69,70]
Mycobacterium tuberculosis	Aerobic vs. anaerobic growthStrain and method comparisons	2D-PAGE2D-PAGE/ICAT	[71,72]
Rickettsiaprowazekii	Tryptic proteome	2D-LC	[73]
Salmonellatyphimurium	Antibiotic resistancePathogenecity mutantRecombinant protein comparisons	2D-HPLC2D-PAGE2D-PAGE	[10,74,75]
Severe acute respiratory syndrome (coronavirus)	Virions and infected cell lysates	2D-LC	[76]
Toxoplasmagondii	Tachyzoite stage expression	2D-PAGE	[77]
Vibriocholerae	Aerobic vs. anaerobic growthStrain comparisons	2D-PAGE2D-PAGE	[78,79]
Yersiniapestis	Human macrophage response comparison with other Yersiniaspp.	2D-PAGE	[80]

HPLC: High-performance liquid chromatography; ICAT: Isotope-coded affinity tag; LC: Liquid chromatography; PAGE: Polyacrylamide gel electrophoresis.

• •2D polyacrylamide gel electrophoresis (PAGE) with mass spectrometry (MS) identification is still the most accessible approach for characterizing the proteomes of bacterial pathogens, and improved 2D sodium dodecyl sulfate PAGE methodologies, such as differential gel electrophoresis, will further refine and extend the information gathered from the approach.
• •Increased applications of isotope-coded affinity tag methods and Fourier transform ion cyclotron resonance MS for the characterization of bacterial proteomes will further expand the dynamic range of bacterial proteomes and greatly facilitate the discovery of new therapeutic and vaccine targets.
• •Immunoproteomics applications represent a potentially powerful convergence of clinical, proteomic and genomic resources to develop improved vaccines for different pathogens.
• •Proteomic profiling of serum/plasma from vaccinated or pathogen-exposed individuals using matrix-assisted laser desorption/ionization time-of-flight MS and surface-enhanced laser desorption/ionization time-of-flight MS is a largely untapped biomarker discovery and diagnostic approach that should be aggressively pursued, particularly for viruses.
• •Collaborative multi-institution, multidiscipline and multiple technology efforts are needed for effective clinical study design, sample collection and sample analysis for diagnostic biodefense pathogen assay development.