WITHIN the genus of drugs capable of rendering a patient sufficiently unaware to tolerate the anguish of surgery, only a few are compatible with full reintegration of the personality at the conclusion of the procedure. We have become so adept in the delivery of this narrow subset of agents that the marvel of anesthetic reversal is considered commonplace, and is taken for granted by caregivers and researchers alike. Still, can it be that such a dramatic intrusion on the normal function of so complex a system leaves no echo or aftershock? Work reported in this issue of the Journal suggests otherwise. 1Fütterer et al. exposed rats to 3 h of a single concentration (5.7%) of desflurane in air. Cytosolic proteins isolated from whole brain homogenates at the immediate conclusion of the exposure interval, and at 24 and 48 h thereafter, were separated by two-dimensional polyacrylamide gel electrophoresis (2D PAGE), stained prior to identification of the excised spots by mass spectrometry (MS), and quantified by comparison of spot volumes with those derived from unexposed control animals. Analysis of spots taking up the stain revealed a handful of proteins with either increased or decreased relative abundance persisting 72 h after anesthetic inhalation. The authors’ contribution represents the first, albeit preliminary, report of a change in the profile of expressed protein content in the brain after administration of an anesthetic drug in widespread clinical use, and merits consideration in its broader context.
This Editorial View accompanies the following article: Fütterer CD, Maurer MH, Schmitt A, Feldmann RE Jr, Kuschinsky W, Waschke KF: Alterations in rat brain proteins after desflurane anesthesia. Anesthesiology 2004; 100:302–8.
To tackle their novel research question, Fütterer et al. have taken a well-traveled proteomic approach. The proteome is generally defined as the complement of proteins expressed by a genome at a particular point in time. Proteomics refers to the qualitative and quantitative comparison of proteomes used to elucidate the differences between two states of a cell, tissue, or organism, i.e. , awake and anesthetized. Proteomic research aims to identify and quantify all proteins, protein isoforms and modifications, protein-protein interactions, structural and functional correlates, and higher-order complexities in a specific context. The endeavor is enabled by the advent of high-throughput methodologies permitting the parallel analysis of hundreds to thousands of proteins and peptides. Sorting by 2D PAGE, coupled with detection by MS, is the senior and most widely used method.
Compared to genes, proteins are structurally, functionally, and temporally much more complex. Because the important factor about a gene is its linear sequence, DNA analysis is a relatively straightforward problem of scalability. We know that the number of human genes (transcriptional units) is finite, falling within the range of 30,000–40,000, and well below most estimates made before completion of the Human Genome Project. Conversely, the number of distinct proteins, which function by virtue of their shifting three-dimensional shapes, is thought to exceed 1,000,000. Many factors account for the difference between the number of gene and protein species. Two that predominate are alternate splicing of the transcriptional unit, and posttranslational modifications of the nascent protein (e.g. , phosphorylation, glycosylation, methylation, or acetylation). Interestingly, a single gene (e.g. neurexin) may encode up to 1,000 different proteins. 2Thus, the DNA sequence provides a template allowing investigators to compare predicted amino acid sequences from completed genomes with the constellation of measured proteomic data. The hurdle for proteomic research is that whereas each fragment of DNA behaves biochemically much like every other, each protein possesses unique properties, imparting differences in solubility, mass, isoelectric point, presence or absence of cofactors, and folding optima, among others. To complicate matters, the dynamic range of abundance in protein mixtures from biologic sources may span 10 orders of magnitude, with low abundance entities nevertheless subserving essential physiologic functions.
To confront these challenges, Fütterer et al. engage in quantitative expression profiling, wherein a biologic sample is characterized by separating, identifying, and quantifying as many proteins as possible, with a focus on those altered in relative abundance with reference to a control sample. In this version of discovery-directed research, investigators often have no idea what will be observed at the conclusion of their efforts. The objective is to generate fresh testable hypotheses and acquire original information about previously recognized proteins, rather than to validate suspected functions and interactions of differentially expressed proteins. As a corollary, results reported from such experimental designs must not be regarded as comprehensive. Failure of a specific protein to make the list does not mean that it is not present in the sample.
Although 263 spots embodying distinct proteins met criteria for analysis in the present investigation, it is reasonable to estimate that the brain as a whole expresses many hundreds of thousands of proteins within any given time frame. Where have the rest of the proteins gone? They have most probably fallen beneath the radar of the methods chosen by Fütterer et al. in this inaugural investigation, and their presence and relative abundance remain to be discerned by proteomic techniques and technologies capable of higher resolution of complex and mixed-abundance samples. Solubilization of the protein content of a heterogenous cellular population at the whole organ level, separation of intact proteins, and visualization by silver staining, as done in this study, permit only a limited display of polypeptides that are relatively plentiful in the composite. In particular, such a crude approach precludes detection of lipophilic membrane constituents of great interest to anesthesiologists (e.g. , ion channels, alkaline proteins, and multimeric protein complexes). Fortunately, a variety of strategies are available for use preceding the 2D PAGE separation step to reduce complexity, increase sensitivity, and enrich the sample. These include microdissection, ultracentrifugation, sequential extractions with reagents of increasing solubilizing power, pH purifications, isoelectric fractionation, subproteome digestion to signature peptides, and protein tagging.
2D PAGE separates proteins based on their electrical charge in the first dimension and their molecular mass in the second dimension, as reflected by divergent protein mobility in a polyacrylamide gel matrix. The technique enjoys wide popularity because high-affinity detection probes and previous knowledge about specific protein properties are not required. However, 2D PAGE is hampered by many constraints: substantial amounts of sample must be loaded, proteins manifest differential staining sensitivities, manual image analysis is a bottleneck to high-throughput data acquisition, and comigrating proteins confound analysis. Also, 2D PAGE is labor-intensive and exhibits significant experimental variation, as the spot selection protocol devised by Fütterer et al. attests. Alternatives to gel-based methods have recently been introduced, including liquid chromatography and protein-detecting microarrays. 3,4These and related technologies are much more amenable to large-scale, “shotgun” determinations of complex sample admixtures, with advantages over 2D PAGE in sample size, scalability, flexibility, control of ambient conditions, and capacity for automation.
MS as used by Fütterer et al. is the detection method of choice in the preponderance of recent proteomic investigations. 3The mass spectrometer is able to resolve many tens of thousands of protein and peptide species by measuring the mass to charge ratio (m/z ) of ions. Matrix-assisted laser desorption/ionization is the process by which proteins refractory to ionization without destruction are converted first to peptides by trypsin digestion, and then to ions by short laser pulses prior to MS. Matrix-assisted laser desorption/ionization–time of flight analysis provides the simplicity, accuracy, and sensitivity necessary for peptide mass mapping, in which peptides are identified by matching a list of observed masses with the archived menu of all masses of each entry in a database. Although matrix-assisted laser desorption/ionization-time of flight is highly efficient in the identification of gel-separated proteins, the measured signal intensity does not correlate with the amount of analyte present in the sample because MS is not an inherently quantitative technique. To draw quantitative conclusions, other methods must be appended, such as the relatively coarse spot-volume estimates used by Fütterer et al. Even so, results are not reportable in absolute unit amounts, comprising a major limitation of 2D PAGE-MS methods.
High-throughput proteomics are currently restricted by requisite comparison to incomplete protein sequence databases. Decades may elapse before closure of the human (or, for that matter, any mammalian) proteome is approximated. Moreover, matching observational data to archived data are not failsafe. Rates of false identification, that is, the probability that the candidate peptide has produced the observed spectrum by chance, are not known with precision, underscoring the choice of Fütterer et al. to use the Mascot score for this purpose. Because protocols are based on successive iteration between experimental and archived data, a framework to estimate statistical power and appropriate sample size for up to tens of thousands of comparisons has yet to be determined. Statistical methods to estimate the significance of associations between protein expression patterns and sample groups remain close to the drawing board, although tools such as cluster analysis, in which proteins of unknown function clustering consistently with those of defined function become candidates for further validation, hold great promise. In any case, as discovery of protein expression patterns becomes increasingly high-throughput, functional validation at the bench will continue to be painstaking, and low-throughput, for years to come. The huge amounts of data generated by proteomic investigations have led to calls for the standardization of protein identification and quantification, and for the organization of the Proteomics Standards Initiative and Human Brain Proteome Project of the Human Proteome Project. *
Turning to the substance of the research question raised by Fütterer et al. ,several additional precautions must be borne in mind. Any mapping exercise risks recapitulation of the debacle of phrenology, whether it be correlating traits to DNA sequence, or cellular perturbations to proteomic expression profiles. Great care must be taken in experimental design to minimize and, whenever possible, eliminate systemic, epiphenomenal associations unrelated to primary effects of anesthetic drugs on the nervous system. In this respect, Fütterer et al. must be commended for care taken to control the possible confounding influence of hypotension, hypoxia, hypoventilation, gender, and other background variables. Inevitably, recalcitrant variables (e.g. , the confounding consequences of immobility and loss of sensorimotor input to the nervous system during anesthetic exposure) will resist the design of even the most sophisticated trials. In the present context, few would argue that 5.7% desflurane (1 minimum alveolar concentration in the rat) represents a full-fledged model of surgical anesthesia. Single doses for single durations of single agents do not support firm or generalizable conclusions; they mandate more sophisticated experimental designs, replication of the authors’ observations in other laboratories, and functional validation of the reported protein express fluctuations. Confirmation with corollary methods are also awaited, using, for example, two-dye fluorescent labeling of pooled proteins from anesthetized and awake sources, separated on the same 2D-DIGE (2-D Differential In-Gel Electrophoresis) gel to quantify differential expression on a single platform, and imaging MS of whole brain sections. 5,6
A small but growing body of literature indicates that anesthetics in clinically relevant concentrations and durations have profoundly detrimental neuronal consequences in those predisposed by environment, age, and genotype. 7–10That the consequences of drug-induced coma fade without repercussion in the otherwise normal brain may be as much a product of wishful thinking as of collective unwillingness to test the axiom. Indeed, 2D PAGE-MS has been available for several decades, but until Fütterer et al. , no one has thought to perform the relevant (and, in retrospect, compelling) investigation. In turn, companion investigations of protein expression profiles after anesthetic exposure in heart, liver, lung, vascular smooth tissue, and other tissues are readily envisioned. Besides confirmation and elaboration of observations made by Fütterer et al. , the profession’s burden going forward will be to perform the problem-oriented research necessary to transfigure broad, but ultimately shallow, proteomic insights into deeper biological understanding, thereby devising safer, more effective, and possibly more evanescent anesthetic interventions.