utomatic Biodetecting and Monitoring Instruments Open New Doors for Environmental Understanding
14 ii q’Lowgh great
strides have been made in automating
hioiietection and monitoring equipment, the cost of
R&D in this field has tended to restrain development and limit application
Velvl W. Greene, College of Medical Sciences, University of Minnesota, Minneapolis, Minn. 55455
This article is based on a paper that Dr. Greene presented as part ol d symposium on Recent Developments in Research Methods and Instncmentation, National Institutes of Health, Bethesda, Md., late last year
104 Environmental Science and Technology
M
icrobiology laboratories have witnessed, in recent years, a dramatic increase in the number and variety of instruments and devices designed to carry out technological operations automatically. The field of microbial detection and monitoring is no exception. It would be useful just to catalog the ideas and instruments thus far suggested and employed, and to evaluate briefly their respective capabilities and limitations. This approach could provide an insight into the current stateof-the-art (at least with regard to nonclassified research and development), and could suggest areas requiring more intensified efforts. On the other hand, most work in automated biodetection and monitoring is part of larger programs-programs in themselves at least as interesting and challenging as the hardware they generate. A comprehensive review should consider the technological challenges as part of the larger framev.ork of economic, philosophical, and political challenges. However, in this short discussion, I can review only briefly the background of current research. Most of the emphasis in the article will be on an analysis of the scientific problem areas involved. and an overview of the general approaches employed to sol\e these problems. In addition, I will try to decribe and discuss several detection and nionitoring instruments. However. let me add a note of caution. The field of automated instrumenration requires sophisticated electronic and nitihanicd engineeringareas in Nhich my knowledge and experience .ire 1im:ied. Consequently. the
FEATURE
emphasis I place in this article on microbiological aspects of the research should not be considered as an accurate and unbiased reflection of the general problem. Rather, it should be accepted only as a point of departureand a subjective one at that. Moving from m a n to machine
Most developments in laboratory automation result from desires to save labor and money and to prevent tedium and human errors. This generalization holds for bacteriology labs, as well as others. It is certainly feasible to analyze the various steps and manipulations of a bacteriology technician and to provide electromechanical devices which can do many of the repetitive tasks automatically, and often more accurately. Usually the transition from technician to machine is gradual: a stepwise advance from simple labor-saving devices to more sophisticated decisionmaking machines to ultrasophisticated, completely automatic analyzers-recorders-computers. The degree of automation and the speed of transition are usually directly related to work load and man-hour cost and inversely related to availability of hands. But even the most imaginative bacteriological planners today still visualize-at least for a few years-a place for the human technician. A human will start the equipment and care for it; choose the material for analysis and interpret data: gather samples and bring them to the machines. neatly labeled and identified: interact with the automatic devices, as it were, and thus serve as a bridge be-
tween questions, answers, operations, and meanings. True automated biodetection and monitoring devices, however, as they are thought about today, are based on a different rationale and are being designed to meet different needs from those just described. The two national agencies which sponsor, and thereby stimulate, most of the research and development in this field are quite unequivocal in their interpretation of the word automatic. NASA's exobiologists want an automated biological laboratory which does everything that machines and technicians do. They want devices to take samples and feed analyzers, analyzers which can decide by themselves which experiments to repeat and which to carry further. They want devices to read out results and interpret the data and then transmit a condensed interpretation through considerable distances. And, most remarkably, they want all this equipment to work unattended on the surface of a planet 50 million miles from the nearest technician where no one has ever been before and after an eight-month journey in the nose cone of a rocket. Similarly, the military laboratory entrusted with our defense against a biological warfare attack is looking foi devices which can be stationed, unattended, around the periphery of strategic areas. These devices must operate continuously, sampling the environment and analyzing it for the presence of potentially lethal microorganisms. And the complete process of sampling. analysis, readout and in-
formation transmittal must be accomplished quickly at literally the speed of wind, if they are to be true warning devices. It is entirely possible that future automated biodetection and monitoring devices might not have such strenuous requirements. One can easily visualize the use of such devices in critical hospital areas or in the food industry where they might be operated by humans as technician-machine, semiautomatic systems. The accusation has been made, with some justification, that NASA and the military are asking us to run before we have learned to walk. In any event, it certainly is easier to proceed gradually from semiautomatic to automatic than to jump directly into a Martian laboratory. But the fact remains that those imposing the rigorous demands in automated biodetection are paying for the research and development. Any discussion of this field would be unrealistic if it did not consider the various criteria of success that NASA or the Department of Defense consider important: automation, remote operation, speed, accuracy, specificity, reliability, and sensitivity. An operations analysis
This discussion intimates that automatic biodetection and monitoring instruments might have a variety of objectives and applications. This is true. Even though NASA and DOD share mutual interests in this field, their basic goals are different. Thus. the direction of research carried on under their respective sponsorship is different. The
Volume 2, Number 2, February 1968 105
involved in generalizing about the subject become even more pronounced when one realms the many other potential applications of automated biodetection systems and their inevitable diversification. Nevertheless, it should be possible to analyze the overall problem and to describe those general tasks common to all such systems. Perhaps the most logical way to approach this analysis is to visualize the operations which a human technician would perform if instructed by a supervisor to “run some bacteriological tests on a sample of ___ .” It could be soil, air, milk, stools, snow from Mount Everest, o r dust from Mars. Similarly, the “tests” to be run are of no concern for the moment. (They are actually of critical concern and are discussed in detail a bit later.) All we want the technician to do is to detect the presence of viable microorganisms in a sample and to tell us something about their quantity and quality. The only condition we will impose,in order to relate this hypothetical technician to a hypothetical automated instrument, is to specify that the material tested is away from the immediate confines of the laboratory. From these instructions, the following general events should take place in approximately the specified sequence. 1. The technician will go (or send someone) to the location in question. The technician or his substitute will be equipped with some type of sterile sampling device. Step one is getting there or being there. 2. The technician or delegate will sample the material to be aaalyzed. The sample will be large enough to be
dL&x&m
meaningful, small enough to be manipulatable, and representative enough to be credible. Step two is sampling. 3. The raw sample will then be transported to the laboratory (or enough of the laboratory must be transported to the sample) for processing and analysis. Subsequently, the processed sample will be transported to the analytical module. Ultimately, the analyzed sample will be discarded or, in the jargon of the system, “transported out.” Sometimes the same sample will be transported to a second and third analytical module. Step three is sample transport. 4. The raw sample must be processed, making it amenable to analysis. Processing might involve pulverizing or subdividing, diluting or concentrating, eluting. mixing, purifying, and the like. It is difficult to visualize microbiological analysis without some type of processing. Step four is sample processing. 5. The sample will now be analyzed for bacteriological content. This step is probably the key one in the process and will be elaborated upon subsequently. Step five is analysis. 6. The technician will, after a given time. read out results of the analysis, answering for himself, at least, the supervisor’s questions about presence of viable organisms. their quantity and quality. Step six is readout of the analytical results. 7. The technician can either convey the raw data to the original questioner for his interpretation. or can summarize and interpret the data himself. conveying the final answers to the uuestioner. Step seven is transmission, or interpretation and tranrniission.
This oversimplSed operation analysis might suggest the magnitude of the task involved in a truly automated detection and monitoring system. Every step must be carried out by scrupulously integrated electromechanical machines, and the efficiency of the entire system will be no better than that of its weakest link. For example, the most ingenious analytical device will fail if it is mated with an inadequate sampler. Similarly, the best computer will have nothing to interpret if the readout device keeps mistaking noise for signal. Research and development activities are currently under way in all phases of this program, although all are not receiving equal attention. Obviously, the different tasks are so closely interwoven and interdependent that it is unrealistic to evaluate the state-of-theart for each task independently. A technological advance in any step will profoundly influence the type of performance required in the other steps. Thus, the development of high volume air samplers which collect thousands of viable bacteria per minute (instead of one or two per minute) could significantly improve the reliability of those bacteriological warfare detection devices which require micrograms of protoplasm for analysis. Similarly, an analyzer u hich can handle either gravel or dust would preclude the need for a processor which has to reduce everything to dust. In general. however. the mechanical and electronic aspects of automated detection and monitoring are at a more advanced level than the microbiological piece< of the problem. The fifth step. analysis, remains the most critical component of the challenge. Un-
fortunately, it s t i l l awaits its own breakthrough. (This statement is conditioned only in part by my amazement at the daily miracles of electronics; in part it results from years of waiting for bacteria to grow a little faster.) Questions instmments ask
The ultimate job of any automated analytical instrument is to answer questions. When biodetection and monitoring devices are divorced from their mechanical attachments for sampling, transporting, processing, and the like, the devices become essentially biophysical answer machines. The type of answers provided will dif€er, depending on the design of the instrument, and can be, for example, specific or general answers or perhaps, quantitative or qualitative answers. Answers which might satisfy NASA’s exobiologists might not be sufficient for the bacteriological warfare defense monitors. Still, all a p proaches to automated microbiological analysis confront this axiom: The quality of the answer depends entirely on the type of question asked. Indeed, transition from science fiction black boxes to useful hardware hinges as much on our ability to rephrase a problem into a series of simple answerable questions as it does on our ability to choose electromechanical tools which might answer the question. One might argue that the foregoing is so obvious that it hardly merits emphasis. Yet, serious frustrations in automated biodetection and monitoring research derive directly from the key words “rephrase a problem into a series of answerable questions.” In dloquial terms, this is the name of the
game and is worth a further explanation. Our analytical instruments should answer five basic questions:
*Is there a viable microorganism present? Are you sure? What kind of microorganism is it? * H o w many are there? *Where did it come from? It is not dil3icult to put these questions in words. However, the lack of suitable biological definitions and the primitive state of a comprehensive biological theory make it extremely difficult to restate some of these questions to machines which, after all, can answer yes or no to only one question at a time. Perhaps this is a natural result of dealing with the unique phenomenon called life. Perhaps it is only a temporary impasse which can be resolved by programming a computer with a course in sophomore biology. But it appears, at this juncture, that a real advance will be made when we learn the right questions to ask, and how to ask them. For example, the first question implies that there is some agreement about what a viable microorganism is. (It is sometimes easier to d e h e love and justice than to define viable.) To be sure, it is possible to describe many criteria and characteristics of living microorganisms. But no single criterion, or even group of criteria, is sufficiently inclusive to include bacteria, fungi, viruses, and protozoa, and, at the same time, sdiciently exclusive to exclude pollen grains, dust, wool fibers, and insect scales. Biologists mature rapidly as scientists when they start to realize that even such time
ride size and reproduction f d as objective criteria when used on unknown samples in the time frame alloted for analysis. Nonetheless, this problem can be approached by breaking the original question into a sequence of logical and related subquestions which can each be posed independently and answered yes and m.Thus, the question about the presence of a viable microorganism might become:
Is there a particle present? *If yes, is it between 0 5 and 10 p in diameter? If yes, does it contain carbon? Hydrogen? Nitrogen? Some other appropriate substance? If yes, are these elements present in certain linkages? Does this particle become two particles in time t under conditions c? *Does this particle become four particles in time 2t under conditions c? Does one particle change substrate s to product p? * D o four particles produce four times as much p? Does formation of product p stop when inhibitor z is introduced?
This questioning goes on as long as it takes to satisfy a predetermined set of standards. We actually do t h e s things and ask these questions of ourselves continually in the laboratory without recognizing the inherent logic processes The challenge here is to recognize and define our own logic; to assign priorities to question and subquestion sequences; to select judiciously the various parameters listed as r, c, s, p, and z; and to choose enough independent quesVdnme 2, Nmobe~2. F
m 1-
io7
tions so that the combined answers will generate as little equivocation as possible. Of course, each subquestion might mean another analysis chamber or module, and each test adds its cost in money, weight, power, and probability of failure. But once the questions are defined, it is only a matter of engineering effort to design a test or device to answer it. All five general questions can be attacked in this fashion. The question Are you sure? must be rephrased as a series of subquestions which, on the one hand, supply redundancy to the first question and, on the other hand, compare the unknown sample with known nonviable artifacts. What kind of microorganism? will become a series of subquestions dealing with morphology, biochemistry, antigen-antibody specificities, nutrition, activity, pathogenicity, and the like. How many are there? becomes deceptively easy to answer. It involves repeating the one question sequence on different quantities of sample, asking is there an organism present in 10 liters, 1 liter, 100 ml.?, and the like. The final question Where did it come from?, should resolve the significance of our findings. NASA wants to make sure that any organism it finds on Mars came from Mars and was not an accidental stowaway. The bacteriological warfare people want to distinguish between a probable enemy offensive and an accidental sneeze. Consequently, this question will be broken down into subquestions that can be related to meteorology, to persistence, to data reproducibility among replicate samples, and to internal references
108 Environmental Science and Technology
about past experiences during simulated and experimental trials. T h e practical automation of microbiological analysis might be considerably easier than I have previously desoribed. No one has yet suggested that we make a device that can answer all questions equally well. Thus, the people involved in bacteriological warfare are interested primarily in pathogens, and only those present in certain concentrations. NASA exobiologists, on the other hand, are interested in any life on Mars. Yet if they can't find perfectly reproducing bacteria, they might be satisfied with a strand of DNA or a molecule of ATP. In other words, it might be true that all approaches to automated analysis must deal with the five basic questions posed, but they do not have to deal equally with each question. Similarly, the maximum degree of accuracy, sensitivity, smed, and specificity should be ideally met by every instrument. But in practice, trade-offs and compromises, both in the quality of questions asked and the value of answers expected, have become an accepted fact of life. Approaches t o microbiological analysis
So far, I have reviewed the general features of automated biodetection and monitoring, emphasizing the basic aims of the research, the mechanical operations involved, and the underlying logic of instrumented analysis. Now it is possible to examine the actual attempts which have been and are being made to perform these tasks. This examination could quickly generate as much confusion as enlightenment. The most direct approach
would be to describe the various instruments, and to evaluate each critically. Since, however, the actual hardware is continually being altered and upgraded, and since specifk instruments are usually oriented to a specific task (for example, bacteriological warfare detection of exobiology), this approach might be unfair or misleading. Furthermore, only some of the many suggestions for automated instruments have progressed to fabricated instruments. Some suggestions are still suggestions; others are on drawing boards; still others are in that engineering-semantic limbo of breadboard models and prototypes. It is quite possible that a given analytical approach, which is not yet finally instrumented, will be mated to a suitable sampling, processing, or readout device today, and will easily become the best detector available tomorrow. Therefore, the following discussion of automated microbiological detection will deal essentially with the approaches suggested or employed to carry out the analytical tasks. These approaches (see Table 1) are classified into three general categories: Approaches based on detecting and monitoring particles. Approaches based on detecting key biochemical components. Approaches based on detecting biological activity. Since the different methods within each category share many of the same advantages and limitations, this classification permits a systematic evaluation of several techniques simultaneously. Furthermore, existing identifiable instruments also can be grouped in Ta-
Classification of automated biodetection and monitoring approaches General category Physical particle detection
Key biochemical components
Suggested approach
Instrument
Magnification
Vidicon microscope Mechanized microscope
Light scattering
Particle ratio alarm Particle counters Aerosol photometers
Volume displacement
Coulter Counter
Antigen detection
FAST
AutoAnalyzer Dyes and staining
Biological activity
Partichrome analyzer J band detector
Biolum i nescence and fluorescence
Fluorimeters
Optical activity
UV polarimeter
Pyrolysis products detection
Pyrolizer Mass spectrometers Chromatographs
ATP detection
Luciferin-luciferinase system
Proteins, nucleic acids, or others
UV and IR spectro-
Growth (increase in cell mass or numbers)
Mi n ivator M u ltivator Wolf Trap
CO, evolution
Gulliver
Phosphatase activity
Poised enzyme Multivator
Substrate change (pH, Eh, Ozinterchange)
Wolf Trap Marbac M in ivator
Pathogenic effects
Tissue cultures
photometers Particle electrophoresis
ble I according to their underlying methodological rationale, eliminating the need for a detailed and repetitive discussion of specific potentials and drawbacks. Any evaluations will be subjective. No exobiology device has been tested on Mars. Performance data on bacteriological warfare devices are classified. Thus, all we can do is make educated guesses. Particle detection approaches
Methods based on particle detection are typified by speed and the ease with which they can be automated. They are also extremely nonspecific. 'The demands of air pollution and industrial hygiene have made available a large number of devices which can detect and quantify particles automatically. Microorganisms are really just a special type of particle, so such instruments can be used, practically without adaptation, for many microbiological purposes. Indeed, microscopy, nephelometry, volume displacement, and photometry are standard techniques in the bacteriology laboratory where samples are known to consist of microbial particles. Conversely, these instruments can not distinguish between a microbial particle and any other kind of particle. So they are severely restricted in an environment containing both signal and noise in unknown ratios. The desirable qualities of particle technology can be utilized in combinations with approaches from the other categories. For example, the partichrome analyzer automatically floods particles (collected by impaction on a continuous tape) with biological stains
Volume 2, Number 2, February 1968
109
and scans the tape with an automated
microscope looking for those particles which retain the dye. Similarly, automated nephelometry and fluorimetry are used as the readout techniques for Wolf Trap, Multivator, and Minivator -4 essentially bioactivity approaches. An automatic aexosol light-scattering instrument, the particle ratio alarm, is routinely employed in bacteriological warfare monitoring, despite nompecificity. This apparatus continually measures and records the numbers of airborne particles in various size ranges. When the ratio of 1-5, particles to 0.5-10, particles becomes significantly hi& the machine sounds an alarm, because the former size range 1-5, is classically considered most important for pulmonary retention. Normally, the ratio should be relatively constant. In the case of a bacteriological warfare attack, however, the agent probably would be disseminated as a 1-5p aerosol, and the ratio would be considerably increased.
Relative accuracy and speed of automated approaches
Particle detection
= =
-
Accuracy is the ability of an approach to determine me actual number of organisms present in a sample. is me time for the complete operation (fmm sam. pling to reporting).
Relative sensitivity and specificity of automated approaches
= =
Key biochemical component
Biological activity
Sensitivity is the lowest threshold concentration of organisms detectable. Specificity is the ability of an approach t o distinguish between viable organisms and nonviable matter. or to distinguish between a given type of organism and others
lcompomnts
Methods based on key biochemical components may employ a e r e n t analytical tools or concepts, but they are all directed toward recognizing those unique chemical structures or molecules present in biological material and absent in nonbidogical material. These approaches, therefore, are more sophisticated and meaningful than the simple physical tests for particles. On the other hand, they all have certain inherent drawbacks: diminished sensitivity, increased need for sample manipulation, and inability to distinguish between biological material that is living or dead. NASA’s exobiologists
Biological activity
sp”d
Particle detection
Kw-
Key biochemical component
Relative automatability and remote operation reliability of automated approaches Excellent Very good
Good Fair Poor
Particle detection
=
Key biochemical component
Biological activity
Automatability is me ability to carry out the complete operation independently. Remote operation reliability is the ability to perform functions without mutine sewicing or calibration.
are interested in both, but bacteriological warfare defense must know the difference between a pathogen and powdered sawdust even though each contains DNA, protein, and ATP. Most approaches are analytical techniques familiar to biochemists and biophysicists. The problems mainly concern miniaturization and reliability during remote operation and are not discussed in detail in this article. Certain approaches, however, may not be as familiar. These are the military's Fluorescent Antibody Staining Technique (FAST) and the pyrolizer device, as well as NASA's ATP and J band approaches. The FAST system might be considered a sophisticated descendent of the partichrome analyzer. It employs a high volume sampler, incubates the sample with fluorescent tagged antibodies, deposits the reacted mixture on a transparent tape, washes away the unadsorbed antibodies and dye, scans the tape with an ultraviolet microscope, and records the numbers of fluorescent particles on the tape. By preselecting the appropriate antiserum or mixture of antisera, it is possible to ascertain in a very short time the presence of specific types of organisms in the environment without too much interference from nonbiological and nonspecific biological noise. The sensitivity (threshold value below which there is no detection) of this device is classified, but published preliminary work suggests the system is sensitive to levels of less than one organism per liter of air. The pyrolizer device heats a sample of collected aerosol, converting the protein into NH4+, which is measured in an ion detection chamber. Although
this device can detect the presence of as little as 0.1 pg. of nebulized albumin per liter of air, it is of questionable use microbiologically, because of the nonmicrobiological proteinaceous noise commonly present in most environments. The I bands are characteristic absorption spectra of biological chemicals (proteins, peptides, RNA, DNA, and carbohydrates, for example) after reaction with a dye, 3,3diethyl-9methyl 4,5,4,5, dibenzothiacarbocyanine bromide. An unknown sample can be added to a solution of this dye, and the spectral shift of absorption that occurs if the biochemicals are present will be measurable by conventional spectrometry. ATP detection is based on the phenomenon of light emission which OCcurs when the firefly enzyme luciferinase reacts with its substrate luciferin. In the absence of the ATP molecule, no reaction-and thus no light emiss i o n d c c u r s . An ATP free enzymesubstrate mixture is prepared, and an unknown sample is added. If microorganisms are present, their ATP will trigger the reaction and the light generated can be measured photometrically. However, many nonviable biologicals contain ATP which will act as noise in this system, also. In general, this category of a p proaches has great promise. Most present work involves engineering development. Most of the devices, such as polarimeters, chromatographs, spectrophotometers, and fluorimeters, lend themselves to automation and even miniaturization. However, questions about whether they can be depended on for remote operation without frequent attention, and the inherent prob-
and specificity, have yet to be resolved. Biological activity approaches
Approaches using biological activity are the most sophisticated. They not only determine that something is present and that it contains biochemicals, they actually ascertain that something is doing something. If one can detect an activity characteristic of living things, the inference is fairly strong that living things are present. Furthermore, if the activity can be measured quantitatively, it might be possible to extrapolate back to the quantity of living material which is responsible for the activity. Nevertheless, these approaches are not completely unequivocal. In almost every case, nonliving substrates can mimic a characteristic viable activity. Thus, crystals can grow, shattered glass can simulate an increase io numbers, rocks can absorb and desorb gases, substrates can deteriorate spontaneously, and the like. However, automated biodetection is never intended to do more than provide strong inferences and cannot be designed for absolute accuracies, so we will have to tolerate these potential confusions-or at least use some redundant tests to spot them. Biological activity approaches have two significant advantages and two serious drawbacks when compared with other methods. The advantages are specificity and sensitivity. The disadvantages are relatively slow detection times and problems of specific microbial requirements. Given enough time, one living cell can produce enough cells or chemicals Vdume 2, Number 2, February 1968 111
Velvl W . Greene is associate professor of public health and of microbiology in the College o f Medical Sciences, University of Minnesota (Minneapolis), a position he has held since 1965. Dr. Greene previously held the positions o f research fellow (1953-56) and assis tant professor (19S9-61) at the University of Minnesota, with positions as assistant professor at Southwestern Louisiana Institute (1956.59) and manager of life science research at Litton Industries (Beverly Hills, Calif., 196164) in the interims. He received his B.S.A. (1949) at the University o f Manitoba (Winnipeg, Manitoba, Canada), and his M.S. (1951) and Ph.D. (1956) at the University of Minnesota. Dr. Green har also studied at North Carolina State College (1952-53). Dr. Green has published in the fields of environmental microbiology, aerobiology, sanitation and contamination control, and food microbiology, as well as related fields. He is a member of the American Society for Microbiology, the Society o f Industrial Microbiology, American Public Health Association, and American Association for the Advancement o f Science.
or mass to be detected very easily. In contrast, the concentration of key biochemical components contained in a single cell is extremely small-usually not enough to be detected by even the most sensitive biophysical instrument. The words “given enough time” also point out one of the disadvantages of activity detection. The generation time of the fastest organism known is not less than 15 minutes. Thus, detection time for low microbial concentrations has to be considered at least as hours, a period wbicb does not really meet the specification of rapid. The second disadvantage deals with microbial diversity and enviroumental preference. Whereas all microbes are particles and most microorganisms share a common biochemistry, the conditions necessary to support their growth and activity differ remarkably from species to species. Thus, a given temperature, medium, pH, and the like will permit certain selected organisms to grow and metabolize hut will inhibit others. Unless the organism being detected is well known, it is quite possible to investigate an environment abounding with life, and miss finding it by a bioactivity approach. This is obviously the great frustration of exobiologists: What medium do Martian microorganisms prefer? How many different media should we try? The specific methods and instruments in this category are really ingenious attempts to automate elementary bacteriology laboratory exercises. Wolf Trap is a miniature growth chamber where turbidity changes are measured by a continuous recording nephelometer and pH changes are monitored by a recording pH electrode. Marbac measures changes in Eh of a growth medium and plots the results against time. Minivator and Multivator are multiple chambers where a given sample is simultaneously tested for growth, fluorescence, enzyme activity, and the like. Readouts are obtained by integrated and miniaturized fluorimeters, nephelometers, colorimeters, and potentiometers. Both Multivator and the Poised Enzyme device use substrates in which a phosphate radical is linked to some fluorescent or chromogenic moiety. The complete substrate is quenched, hut if phosphatase, which apparently is
a universal enzyme, attacks the molecule to liberate PO,, the remaining fraction is also liberated and can be detected fluorimetrically or colorimetrically. Gulliver continues to excite the imagination. This device was originally intended as a routine laboratory instrument for evaluation of water quality. It has been redeveloped into a completely automatic, remotely operable, and qnite reliable apparatus. It inoculates a sample with a medium containing CE4labeled sugars. As the microorganisms grow, they attack the sugar and liberate radioactive COP which is detected by strategically placed Geiger-Mueller counters. Of all the instruments described in this section, Gulliver and Wolf Trap come closest to meeting the requirements of general life detection, though they should by no means he considered as ideal. Table 2 is a summary evaluation of the three general categories of automated approaches. The tabulated merits and limitations constitute a subjective critique. Also, within each category there are specific approaches which disrupt the general pattern. However, this table does provide some idea about the trade-offs involved in the field. Speed and automatability are usually gained at a sacrifice of specificity. Specificity and sensitivity are usually gained at a sacrifice of speed and reliability, and so forth. No one approach and no one group of approaches will solve all of our problems or meet all of our needs. This is the underlying reason for NASA’s concern with a planetary lahoratory rather than a given single instrument. Similarly. the bacteriological warfare people are thinking of redundant monitoring systems, rather than devices. A combination of approaches from all three categories, And perhaps several different techniques from each category working concurrently or in sequence, would augment significantly the information generated from anysingle experiment. But this type of development and integration is expensive and time consuming. It is fair to say that automated biodetection and monitoring still has the bulk of its history ahead of it.
