Measuring Infectivity and Virulence PDF

3 Measuring Infectivity and Virulence Because virulence is a microbial attribute that is exhibited only in the context of a susceptible host, it is critical that when designing experiments a suitable infection model system be used. First, the animal host must be susceptible t...

3 Measuring Infectivity and Virulence Because virulence is a microbial attribute that is exhibited only in the context of a susceptible host, it is critical that when designing experiments a suitable infection model system be used. First, the animal host must be susceptible to the pathogen. Ideally, infection using this animal model should mirror all of the disease symptoms that occur during natural infection. Alas, this is not always possible, and alternative infection models must then be found. Moreover, to identify factors that contribute to virulence (virulence factors) and to understand their roles in the infection process that leads to disease, one must also have some way to accurately and reproducibly measure virulence using the infection model. In this chapter, we describe various animal models that have been developed and that are being used for studying infectious diseases caused by bacteria. We also describe various methods that can be used to measure virulence and infectivity (i.e., the degree to which a host has been infected). 3.1 Animal Models of Infection 3.1.1 Human Volunteers We start with the best possible model for studying human disease: humans. Human volunteers can be, and have been, used in many infectious disease studies. Ethical considerations make this approach problematic unless the disease is not life threatening, is easily treatable, or can be prevented with a vaccine. Human volunteers have played an important role in testing of vaccines against cholera, chancroid (genital ulcers), and gonorrhea, all diseases for which there is no suitable animal model. Also, antibiotic intervention trials, such as those to assess the efficacy of antibiotic treatments for ulcers or atherosclerosis, are also ethically defensible, because the treatment presumably does no harm to the participants and may even benefit them. A clinical-trial study involving the inoculation of humans with a disease agent or the treatment of infected humans is called a (google) "prospective" study because it starts in the present and moves into the future. Another type of human clinical study is called a "retrospective" study. This type of study is done on infectious disease outbreaks that have already occurred accidentally or naturally and that are studied in Page 1 of 26 retrospect (looking back at past events) to obtain information about disease transmission or progression in humans. One example is the case of a school bus driver with tuberculosis who managed to infect a number of school children before his disease was diagnosed. This study, in retrospect, provided valuable information about the factors affecting transmission of tuberculosis. A retrospective study of various aspects of this case revealed, for example, that the likelihood of infection was directly linked to the amount of time a child spent on the bus. The fact that 40 minutes a day versus 10 minutes a day made a discernible difference gave scientists a new appreciation of how infectious tuberculosis actually is. Similarly, retrospective studies of food-borne disease outbreaks have helped to determine what types of people are most likely to develop a life-threatening disease in such cases. 3.1.2 Nonhuman Animal Models For most studies of infectious diseases, however, nonhuman animals are the models of choice, when possible. Since the time of Koch and Pasteur, laboratory rodents have been the most widely used models for infectious disease research because they are small and are thus more easily (and cheaply) housed and cared for than larger animals, such as pigs and baboons. 3.1.2.1 Rodents as Animal Models Although rodents are closely related to humans on the evolutionary scale, there are a number of important differences. 1. Anatomically, they are similar to humans, except that they have fur and tails. 2. They also have a more prominent cecum, whereas humans have a vestigial appendix, and a very different microbiota. 3. Rats do not have a gall bladder. 4. A fact that is frequently overlooked but that could be a factor in the use of rodents to study intestinal disease is that rodents practice coprophagy (routine ingestion of one's own feces), whereas humans do not (except unintentionally via unwashed hands). 5. There are undoubtedly many other differences, as is evident from the very different course some human diseases have in mice. For example, Salmonella enterica serovar Typhimurium, which causes diarrhea in humans, causes a systemic disease in mice that closely resembles the human disease typhoid fever that is caused by S. enterica serovar Typhi. However, S. enterica serovar Page 2 of 26  Typhi, which can be deadly in humans, does not infect mice. There are a number of other examples of this type of host specificity. 3.1.2.2 More Exotic Animal Models Because of these differences, for certain diseases, scientists have developed more exotic animal models that may not mimic the human anatomy in many ways and may not mimic the disease in every respect but still provide unique insight into certain aspects of the disease. Examples are ferrets as a model for gastric ulcers caused by Helicobacter, guinea pigs as models for tuberculosis, armadillos for leprosy (skin lesions caused by Mycobacterium leprae), chinchillas for otitis media (ear infections caused by Haemophilus influenzae or Streptococcus pneumoniae), and zebra fish for necrotizing fasciitis (“flesh-eating” disease caused by Gram-positive streptococci) or infections by fish-specific Mycobacterium species related to Mycobacterium tuberculosis, which causes tuberculosis in humans. Recently, the nematode Caenorhabditis elegans has been used as an animal model for certain pathogenic bacteria, such as Pseudomonas aeruginosa, that infect many different hosts. 3.1.2.3 Criteria to be Met by an Animal Model Rather than establishing rigid criteria for whether an animal model is "good" or "bad," a variety of factors need to be considered when choosing such a model for human disease. 1. Ideally, an animal model for a human infection should contract a disease whose symptoms and distribution of bacteria in the body mimic the human form of the infection. 2. Similarly, animals should acquire the disease by the same route as humans. 3. It should be easy and inexpensive to maintain the animal. For some diseases, nonhuman primates are the only suitable animal models, yet these animals are difficult to obtain, particularly in large numbers, and are expensive to house and maintain. 4. An animal model must have produced insights that are consistent with observations of the disease in humans. 5. Research on the animal model should led to effective interventions in human disease. There are also ethical considerations. 6. The model should not involve causing significant pain to the animal, which might be considered unwarranted or unethical. Page 3 of 26 7. More recently, another criterion has surfaced for choosing an animal model: the ease with which the animal model can be manipulated genetically. 3.1.2.4 The Usefulness of Genetically Manipulated Animal Model One of the primary reasons /or putting forth the nematode Caenorhabditis elegans as an animal model, despite the fact that primitive worms are a long way evolutionarily from humans, is that there are many mutant strains of C. elegans, making it possible to investigate the effects of host traits in the development of a disease. There are few scientists who would fail to agree that an imperfect animal model is better than no animal model at all. On the other hand, the course of disease can be very different in distant models, such as C. elegans, that have innate immunity but lack adaptive immunity. Recently, scientists have focused on modifying conventional animal models, such as mice, to gain unique insights into particular infectious processes. 1. Infant mice have immature immune systems compared to adult mice, so they are often more susceptible to infection. 2. Likewise, irradiated mice are immunocompromised because X rays have destroyed their immune cells. 3. Nude mice are genetically defective in their ability to produce T cells. 4. Neutropenic mice are defective in their ability to produce neutrophils. 5. SCID (severe-combined-immunodeficient) mice are genetically defective in their ability to produce B cells and T cells, and hence all of these types of mice are more susceptible to infection. 3.1.2.5 Transgenic Mice as Animal Models A good reason for remaining attached to the laboratory-rodent-as-model issue is the increasing number of available mouse strains with specific genetic alterations (i.e., transgenic mice). These include not only the so-called "knockout mice," which have disruptions in specific genes, but what we might call "knock-in mice," mice that have had human genes introduced into their genomes. Both knockout and knock-in mice have been used to study the attachment of Helicobacter pylori to the gastric epithelium. Knock-out mice that lack decay-accelerating factor (DAF) have been used to demonstrate the binding of H. pylori to DAF, thus leading to inflammation of the gastric mucosa. Alternatively, a knock-in mouse, the Leb mouse, which carries the human Lewis b antigen (one of the blood group carbohydrate Page 4 of 26 antigens on the surfaces of human cells), has been used to study H. pylori infections because the Lewis b antigen appears to be another receptor for attachment of H. pylori to gastric cells and gastric mucin. Such mutant animals offer vast new possibilities for research on infectious diseases, possibilities that are only beginning to be explored. 3.1.2.6 Germ-Free Animal Models Two other useful animal models have recently emerged as important tools for studying infection processes and the host immune response. 1. Gnotobiotic (germ-free) animals are raised in sterile environments and have no bacteria in or on them. As a result, they have severely underdeveloped mucosa-associated-lymphoid tissue and no partial immunity due to prior exposure. Unfortunately, they are very expensive to buy and to maintain. 2. Specific-pathogen-free animals are animals that are raised in an environment free of a particular pathogen but are otherwise exposed naturally to other microbes. These models eliminate preexisting immunity due to prior exposure to a microbe that might complicate the interpretation of host response to the pathogen of interest. Gnotobiotic animals have been particularly useful for studying the development of immunity to a pathogen in a "truly naive" (unimmunized) animal. Such animals have provided invaluable insight into the nature of commensalism and those characteristics that determine whether a microbe will become a commensal or a pathogen. 3.2 Measuring Bacterial Infection in Animal Models 3.2.1 Ethical Consideration 3.2.1.1 Justifications for Carrying Out an Experiment using an Animal Model Experiments involving animals also involve numerous ethical issues. Disease models invariably require infection of animals, most often rodents, with bacteria that cause disease symptoms, distress, and sometimes death. Therefore, there must be truly compelling reasons for carrying out these experiments. Page 5 of 26 1. The genetic and physiological properties of the bacterial strains to be tested should be fully characterized before animal experiments are considered. 2. The animal experiments should be designed to test critical hypotheses that can provide useful information for understanding the bacterial disease being studied, even if negative results are obtained. 3. Most important, all procedures used in animal experiments must be approved in advance by a duly appointed institutional committee whose members have experience performing animal experiments, including veterinarians. 3.2.1.2 Requirements for an Approved Animal Protocol Approved animal protocols contain a variety of important details about the experiments and conform to stringent standards of experimental design. 1. It needs to be argue persuasively that no alternative model can provide information equivalent to that provided by the proposed animal model. 2. The protocol must contain extensive information about the (i) choice of the animal model to be used, (ii) ways to minimize the number of animals required that will still give statistically significant results, (iii) precedents from the scientific literature, and (iv) documentation of appropriate training of laboratory personnel. Besides detailed experimental methods, these documents must contain (v) extensive information about care of the animals before and during the experiments, anesthesia, minimization of pain and discomfort, and euthanasia. 3. In general, death of the animals should not be used as an end point, and complete monitoring rubrics to judge pain and discomfort are compiled, including how to judge whether the animals are sufficiently sick or moribund to require euthanasia. 4. Finally, these protocols must be reviewed and renewed annually. 3.2.2 Animal Model Basics Because virulence factors are defined as factors that allow a bacterium to infect, to cause symptoms, and sometimes to cause death, measuring properties such as infectivity and lethality is an important part of virulence studies. There are several guiding principles in the design and execution of animal models of infectious disease. Page 6 of 26 1. Selection of Appropriate Animal: First, an animal species that reflects the disease process in humans or that will answer the questions being asked should be chosen. There must be a truly strong rationale to use species higher than mice, rats, or zebra fish. 2. Route of Infection: A route of infection must be chosen. Bacterial broth or plate cultures are diluted to defined starting doses based on numbers of bacteria, usually reported as colony-forming units (CFU). (i) In the simplest models of systemic infection, the bacterial dose is simply injected with a syringe into the peritoneal cavity of a mouse. (ii) In simple models of invasive respiratory infection, a small drop containing the bacteria is placed on the nose of an anesthetized mouse whose mouth is gently held closed for a moment. The mouse will then inhale the drop into its nasopharynx and lungs. (iii) In other simple models of infection, the dose of bacteria is delivered orally or directly to the lung or stomach of anesthetized animals through fine tubes. (iv) Other, more complicated modes of infection are used for specific models. 3. Monitoring the Infection: Once the species and route of infection are chosen, methods for monitoring the infection must be chosen and planned. (i) In some cases, very tiny amounts of blood are removed from tail veins at different intervals, and the numbers of CFU in the blood are determined by spreading dilutions onto plates. Bacteremia can lead to the accumulation of well over 10 to 10 CFU per ml of blood. 8 9 (ii) In many cases, the number of CFU per organ needs to be determined, and this paradigm involves euthanizing the animal, removing the organ, grinding it in physiological saline solution, and spreading a dilution onto plates to determine the number of CFU. (iii) To determine colonization in the nasopharynx, saline is injected from the dissected trachea of the mouse and collected at the nose. In all of these models, several animals need to be sacrificed for each dose and time point following infection to obtain statistically significant results, and the number of animals can become quite high. New methods to image infections in live mice (biophotonic imaging), which substantially reduce the number of animals needed for some of these experiments, are described below. Page 7 of 26 3.2.3 Survival-Curve Analysis and Biophotonic Imaging 3.2.3.1 Survival-Curve Analysis Survival-curve analysis is a commonly used approach that determines the median survival time of animals following infection with a wild-type bacterial pathogen and mutants that are being tested for their virulence properties. Advantages of this approach are that it can provide statistically robust results using a relatively small number of animals and that it can be combined with biophotonic imaging. In the example shown in Figure 3.1, a wild-type strain of S. pneumoniae and a relSpn mutant containing a deletion of the gene whose protein product catalyzes the synthesis of the signaling molecule, guanosine-pentaphosphate and guanosine- tetraphosphate [(p)ppGpp], were inoculated intranasally at a dose of about 107 CFU into a specific strain of outbred mouse. A control was included in which the relSpn mutation was complemented by a copy of the wild-type rel+Spn gene located elsewhere in the bacterial chromosome. Ten mice were inoculated per strain tested. The infected mice were examined for moribundity every few hours following inoculation. As noted above, death was not used as an end point, but rather, mice that were moribund (i.e., showing obvious signs of acute illness) were euthanized. The survival curve was generated by counting the moribund and nonmoribund mice at each time point. A special kind of statistics, Kaplan-Meir analysis with log rank tests, was used to generate the median survival times and the P values, which are an indication of statistical difference. The survival curve showed that complemented and wild-type bacteria caused statistically similar median survival times of about 60 hours at this dosage in this mouse strain. In contrast, the median survival time of mice inoculated with the relSpn mutant was about 140 hours, which means that the virulence of the relASpn mutant was significantly attenuated. Another important lesson from this study is that the median survival time depends on the bacterial dose and the type of mouse used. Note: In-bred: Bred of parents closely related (Homogeneity) Out-bred: Bred of parents not closely related (Heterogeneity) Page 8 of 26 Figure 3.1: Disease progression and survival of mice infected with luxABCDE rel+Spn, luxABCDE relSpn, and complemented luxABCDE relSpn bgaA::relp-rel+Spn strains of serotype 2 Streptococcus pneumoniae. The mice were inoculated intranasally with 6 x 106 CFU, and disease progression was followed in real time by survival-curve analysis (A) and biophotonic imaging (B). Page 9 of 26 3.2.3.2 Monitoring the Progress of Infection Colony-Forming Unit (CFU) Count We could have drawn a tiny drop of blood or removed the lungs to determine the number of CFU at each time point of the experiment. However, the load of bacteria in the blood would probably not be that informative, because the blood CFU tracks with the severity of infection, and moribund animals are bacteremic. To determine the number of CFU per lung per strain, we would need at least five animals per time point, which would increase the number of animals needed from 10 mice per bacterial strain tested to about 100 mice per strain. Although both of these approaches are legitimate and would yield meaningful results, new technologies have made other approaches possible. Biophotonic Imaging One new alternative approach is biophotonic imaging. In this approach, a bacterial luciferase operon (luxABCDE) is transplanted into bacterial strains to be tested at a chromosomal location that does not affect virulence. This luciferase operon from a Vibrio species contains all of the genes for bacterial luciferase and its substrate needed to make the bacteria glow in the dark. The bacteria carrying the luciferase operon are used to infect the mice as described above. However, now, at intervals of about 8 hours, the mice not only are checked for disease progression, they are lightly anesthetized in a light-tight chamber connected to a supersensitive digital camera that can detect and quantitate light produced by the bacterial infection in the mouse. The camera literally looks right through the mouse to follow the course of infection (Figure 3.1B). In the example shown, wild-type and complemented strains of S. pneumoniae caused localized pneumonia in the lungs of the mice. However, deletion of the relSpn gene changed the course of infection completely in most of the mice. The bacteria initially localized to the lower abdomen, probably in lymph nodes, instead of in the lungs. Later, the bacteria moved to the peritoneal cavity before moving to the lungs and bloodstream. Thus, the relSpn gene is not only required for full virulence, it dictates the normal course of infection of S. pneumoniae. This change in the course of infection would not have been detected by simply dissecting lungs, and the experiment required far fewer animals. Page 10 of 26 3.2.4 LD50 and ID50 Values 3.2.4.1 Determining ID50 and ID50 Values Two other common measures widely used to define virulence in animal models are the 50% lethal dose (LD50) and 50% infectious dose (ID50). The LD50 value is the dose at which 50% of the animals are moribund, i.e., the number of bacteria needed to cause terminal acute infections in 50% of the animals. LD50 values measure a much later event in the disease process, namely, moribundity of the host that will lead to eventual death. However, not all pathogens kill the host, even during a natural infection. In such cases, it is necessary to have some other means of measuring the extent of infection caused by the bacterium. The ID50 value is the infectious dose at which 50% of the animals are infected, i.e., the number of bacteria necessary to infect 50% of the animals exposed to the bacterium. ID50, values measure the ability of the bacterium to colonize the host, establish infection, and manifest disease symptoms that are measurable. The most common method for determining an ID50 value is to determine (i) the number of CFU present at a certain site in the animal (e.g., blood, liver, spleen, lung, lymph node, or brain) after a certain time of infection, but there are other disease indicators that can also be used, such as (ii) the host's temperature (fever), (iii) the size of the necrotic lesion, (iv) increased swelling from edema, or (v) loss of motor function. (vi) Biophotonic image analysis, instead of CFU, can also be used to follow the progress of infections for ID50 value determinations. Plots of the number of animals infected (or that become moribund) versus the number of bacteria in the inoculum (dose) are sigmoidal, not linear (Figure 3.2). To locate the 50% point precisely, statistical methods, such as the procedure of Reed and Muench or probit analysis, are used. These methods not only determine the 50% point mathematically, but also provide a measure of experimental error that is important for deciding whether two ID50 or LD50, values differ from one another by a statistically significant amount. The reason a 50% value is determined rather than a 100% value is that it is much easier to determine a 50% value accurately because it is in a region of the curve where maximal change occurs. Students who have trouble with the concept that the lower the LD 50, or ID50 value, the more lethal or infectious, respectively, the bacterium is, might make use of the following reminder: when it comes to LD50 or ID50, less is worse. Page 11 of 26 Figure 3.2: Typical curves obtained when determining ID50 values for bacteria with different levels of infectivity. For bacterium A, only 102 organisms per animal are required to cause disease in 50% of the animals (ID50 = 102). In comparison, for bacterium B, 104 microbes per animal are required to cause disease in 50% of the animals (ID50 = 104). Therefore, bacterium A is more infectious than bacterium B because it requires fewer cells of bacterium A to cause disease. 3.2.4.2 Limitation of LD50 and ID50 Values LD50 and ID50 values have proven to be useful measures of lethality and infectivity, but they have some important limitations. 1. These parameters reflect the cumulative effect of the many steps involved in colonization and production of symptoms. Therefore, they lack a certain level of sensitivity, and the failure of a mutation in a bacterial gene to increase the LD50 or ID50 value does not always mean that the mutation did not affect some virulence determinant. 2. Moreover, LD50 and ID50 determinations require the use of a reasonably large number of animals at each bacterial dose. 3. Another limitation is that LD50 and ID50 values at best provide relative measures of virulence when different strains of a bacterial species or different mutants of the same strain are compared, and indeed, they provide the most useful information when clearly defined isogenic strains are used. They can be misleading when misused to compare two different diseases. For example, the bacterium that causes cholera has an ID50 value of about 10,000 bacteria when ingested by humans, whereas the bacterium that causes bacterial dysentery has an ID50, value of 10 to 20. At first glance, this might appear to indicate that the bacterium that causes dysentery leads to a more serious disease than the Page 12 of 26  bacterium that causes cholera, but this is not the case. In fact, cholera is often a fatal disease whereas bacterial dysentery seldom causes death. The difference in ID50 values is due lo the relative abilities of the two species of bacteria to survive passage through the acid environment of the stomach, the first step in infection by an ingested pathogen. Thus, comparisons of LD50 or ID50 values must be made with care and are best applied to assessing the relative infectivity or lethality of closely related strains of bacteria. 3.2.5 Competition Assays 3.2.5.1 Principle of Competition Assays A way to make infection experiments more sensitive is to use a competition assay and to determine the competitive index (CI). The CI is defined as follows: CI = [Output ratio (CFUmutan/CFUwild type)]/[Input ratio (CFUmutant/CFUwild type)]. In this assay, the animal is infected with a mixture of mutant and wild-type bacteria (input ratio = CFUmutant/CFUwild type). After the bacteria are given time to establish an infection, samples are taken from various parts of the animal, and the ratio of the number of mutant to the number of wild-type bacteria (output ratio = CFUmutant/CFUwild type) is determined. (1) If the ratio of mutant to wild-type bacteria is the same as in the infecting dose (CI = 1.0), the mutation had no detectable effect, but (2) if the wild-type outcompetes the mutant (CI 1.0), the mutation caused an increase in virulence. 3.2.5.2 Advantages of Competition Assays 1. One reason this type of assay is more sensitive than using LD50 or ID50 values is that the mutant and wild-type bacteria are competing for the same turf, so we are comparing the fitness of the mutant to compete with the wild-type bacterium for survival in the host environment. 2. A second reason is that it is often easier to quantify the ratio of the wild type to the mutant accurately than it is to determine the LD50 or ID50, value accurately. 3. Also, it allows the investigator to use fewer animals, since it is not necessary to use multiple animals for each dose to generate a dose curve. Page 13 of 26 3.2.5.3 Disadvantages of Competition Assays However, there are some serious issues to consider when interpreting competition experiments. 1. Mutants that grow more slowly in broth cultures will likely lose in competition assays in an animal. 2. It can also be difficult to interpret what subtle differences obtained in competition experiments actually mean in terms of the mechanism of virulence. 3. The competition experiment paradigm precludes the use of some detection approaches, such as biophotonic imaging, because the light levels produced from the parent and the mutant are indistinguishable. 4. A very serious concern is trans effects that may allow mutants defective in virulence to grow in the presence of the wild-type parent strain. For example, suppose a mutant cannot produce a secreted toxin that is required for disease symptoms. Using single-strain approaches, such as survival-curve analysis, the parent strain would appear virulent, whereas the toxin mutant would be strongly attenuated. However, in the competition experiment, the parent strain may secrete sufficient toxin for the mutant to grow fully. In this case, approximately equal numbers of CFU of the parent and mutant would be recovered from the infection, and the conclusion from the competition experiment would be that the mutant was fully virulent. This is clearly the wrong conclusion. Usually a combination of single-strain and competition infection experiments is needed to begin to understand processes as complicated as bacterial virulence in animal models. 3.3 Tissue Culture and Organ Culture Models 3.3.1 Tissue Culture Models 3.3.1.1 Advantages of Tissue Culture Models Although animal models are the gold standard of research on bacterial virulence, animals present a complex system in which many variables cannot be controlled. 1. Cultured mammalian cells are commonly used to provide a more easily controlled system for investigating host-bacterium interaction. Page 14 of 26 2. Tissue culture cells can be grown in defined medium under reproducible conditions with only one or a limited number of cell types represented, which makes measurements and interpretations easier and more reproducible. 3. It is also easier to perform experiments involving radioactive compounds and to introduce foreign DNA into tissue culture cells. 4. Cultured cells can be readily visualized by microscopic techniques, and cells expressing certain fluorescently labeled marker proteins can be sorted using high-speed devices. 5. They also cost less per day to house than laboratory rodents, do not fight with each other, and have seldom been known to escape from their cages. 3.3.1.2 Disadvantages of Tissue Culture Models Because of the very important role tissue culture cells have played in molecular investigations of bacterium-host cell interactions, it is important to understand their limitations, which must be kept in mind when interpreting the results of experiments. 1. Primary cultures of mammalian cells that are not derived from tumors can be obtained from animal tissue, biopsy material, or the blood of human volunteers (e.g., macrophages), but these primary cell types usually undergo a limited number of divisions in culture and can only be maintained for a couple of weeks. 2. "Immortalized" cells that continue to divide in culture are often used in routine experiments. These immortalized cells are derived directly from a tumor, by fusing a primary cell type with a tumor cell, or by continuous selection of primary cell lines for unregulated growth. Propagation of mammalian cells that are unregulated for growth in culture leads to the accumulation of numerous mutations, gene rearrangements, and gene duplications. These mutations are not only uncharacterized because they are so numerous and complex, they are also not easily reproducible. Two separate sets of primary cells from the same source that are treated in exactly the same way to produce immortalized cell lines will have different combinations of mutations and rearrangements. 3. Cells in culture are no longer in the same environment as in the organ of origin, and many genes that were expressed by cells in an intact organ may not be expressed in cultured cells. Culture medium does not represent the natural conditions and biochemical composition found in the body, where fluids and cellular secretions may significantly influence cellular processes in ways that are not reproducible in a culture system. Consequently, tissue culture cell lines lose Page 15 of 26  many traits of the original tissue from which they were derived and may not display the same surface markers in culture. 4. A related problem is that most cultured cells lose their normal shape and distribution of surface antigens. Cells in an intact animal are usually polarized. That is, different regions of the cell surface membrane are exposed to different environments, e.g., lumen, adjacent cells, underlying blood, and other tissue types. 5. Still another problem with using cultured cells as, for example, representatives of human mucosal surfaces is that real mucosal surfaces are covered with mucus and bathed in solutions that are difficult to mimic in an in vitro system. For example, the fluid bathing the small-intestinal and colonic mucosal cells is anaerobic and contains bile salts. The fluids bathing the vaginal mucosal cells and the bladder mucosal cells also have a low oxygen content and high concentrations of compounds, such as urea or lactic acid, that could have an effect on mucosal-cell physiology. 6. Finally, real tissues consist of multiple cell types, not of a single cell type. Today coculturing more than one cell type can be done for a number of systems to mimic the natural interactions that might occur. For example, adding activated T cells or B cells, which serve as cellular sources of cytokines and chemokines, to a culture of dendritic cells can induce them to undergo maturation. 3.3.1.3 Differences between cells of an actual mucosal membrane and tissue culture cells Membranes on different sides of polarized cells contain different sets of proteins, a feature that is presumably important for their function. This is illustrated in Figure 3.3 for a layer of mucosal cells, but the same considerations apply to cells in other parts of the body. In a layer of normal mucosal cells, the apical surface is exposed to the external environment (e.g., the lumen contents in the gastrointestinal tract), whereas the basal and lateral surfaces are in contact with the extracellular matrix (a combination of proteins and polysaccharidcs that "glues" the cells together). Mucosal cells in the gastrointestinal tract and some other tissues are also connected to each other by tight junctions, which are made of specialized tightly binding protein complexes that make an impermeable connection between adjacent cells. In contrast, tissue culture cells that are grown as nonconfluent monolayers do not have differentiated surfaces, and proteins that are found only on the apical or on the Page 16 of 26 basal-lateral surface of cells in vivo may be distributed over the entire surface of such cells, assuming they are produced at all. Allowing the cells to grow to confluence does not necessarily solve this problem. It is usually necessary to provide an extracellular-matrix substitute and to provide hormones to obtain a polarized monolayer in culture. Production of a polarized monolayer in culture has been achieved for some types of cell lines, but the expression and distribution of relevant surface molecules should still be checked before concluding that a polarized cell monolayer in culture is the same as tissue in the intact animal. In practice, this is seldom done. Figure 3.3: Differences between cells of an actual mucosal membrane and tissue culture cells. (A) Actual membrane in vivo. (B) Nonconfluent, nonpolarized tissue culture cells. (C) Polarized monolayer of tissue culture cells attached to a semipermeable membrane. Page 17 of 26 3.3.1.4 Some Important Considerations Using Tissue Culture Models The fact that there are problems with existing tissue culture cell lines does not mean that such cell lines are not extremely useful. If their limitations are kept in mind, cultured cell lines can be marvelous tools for discover. Once a new phenomenon has been discovered in cultured cells (e.g., attachment of bacteria to a mammalian cell receptor or reorganization of the host cell cytoskeleton), experiments can be designed that use organ cultures or animals to test the importance of the phenomenon in vivo. It is not uncommon for a mutation that affects the ability of a bacterium to infect a tissue culture cell to have no effect when tested in animals, so taking the study from tissue culture cells to the animal is an important step. In other words, tissue culture cells can be important for generating hypotheses, which can then be tested in the intact animal. It is a mistake, however, to take the results obtained from studies of tissue culture cells and extrapolate directly from them to the disease in humans. 3.3.2 Gentamicin Protection Assay to Study Adherence Properties of Bacteria 3.3.2.1 The Steps in a Gentamicin Protection Assay for Mammalian Cells Tissue culture cells have been widely used to study the adherence properties of bacteria and are particularly useful for identifying virulence factors involved in binding and invasion by intracellular pathogens. One such assay that researchers frequently use to distinguish between mutants that are defective in attachment and those that are defective in invasion is the gentamicin protection assay (Figure 3.4). In this assay, a monolayer of mammalian cells is incubated with bacterial cells at a certain multiplicity of infection (MOI), which is defined as the number of input bacteria per mammalian cell, in the well of a culture plate. Following incubation for a period of time to allow binding and invasion to occur, which may vary depending on the particular organism being examined, the samples are divided into three sets. First Set (Total CFU) For the first set, which will provide the total number of bacteria (CFU) that bound to the mammalian cells, (1) the medium containing unattached bacterial cells is removed and placed in a separate tube. (2) The mammalian cells are then lysed using mild detergent or gentle scraping (disruption that lyses the mammalian cells Page 18 of 26 but does not lyse the bacteria), and (3) the mixture is added back to the tube containing the medium with unattached bacteria. (4) This suspension is plated on agar plates in serial dilutions, and the bacteria are allowed to grow to form colonies, which are counted. This represents the total number of CFU in each well at the end of the experiment. Page 19 of 26 Figure 3.4: Gentamicin protection assay. Shown are the steps in a gentamicin protection assay for mammalian cells that are grown in suspension. Second Set (Attachment) For the second set, which will provide the number of adherent bacteria (CFU), (1) the infected mammalian cells are first washed several times with a buffered solution to remove any nonadherent bacteria, (2) the mammalian cells with adherent bacteria are lysed, and (3) serial dilutions of the suspension are plated to determine the number of cell-associated bacteria (in CFU). (4) The ratio of cell-associated CFU to total CFU at the end of the experiment is defined as the adhesion frequency. Third Set (Invasion) For invasion assays, (1) a third set of wells is washed as described above to remove nonadherent bacteria, and (2) fresh culture medium containing the antibiotic gentamicin is added to each well. Gentamicin cannot enter the mammalian cells, so it kills only extracellular bacteria and does not kill bacteria that have already entered the mammalian cells (i.e., internalized bacteria are protected from the antibiotic and therefore survive and can be counted). (3) The cells are incubated with the gentamicin-containing medium for anywhere from 30 minutes to 2 hours, at which time the medium is removed, and (4) the cells are washed and lysed as described above. (5) Serial dilutions are then plated to determine the number of intracellular bacteria (as gentamicin-resistant CFU). (6) The ratio of gentamicin-resistant CFU to cell-associated CFU at the end of the experiment is defined as the invasion frequency. This allows invasion to be measured as an event separate from adherence. It is also possible to report the ratio of the number of gentamicin- resistant CFU to the total number of CFU in the well, which may vary greatly depending on the frequency of adherence of a given bacterial strain for the given Page 20 of 26 mammalian cells used. There may be some variation in the details of the experiment. For example, if the mammalian cells do not form adherent monolayers and are instead cultured in suspension, then centrifugation steps must be added during the wash steps (as shown in Figure 3.4). 3.3.2.2 Interpretation Gentamicin Protection Assay 1. If a bacterial mutant is defective in an adhesion factor that prevents it from attaching to the mammalian cell, then no colonies will form on the agar plates for the second and third sets of wells. 2. Mutants that can still adhere but are defective in invasion factors, so that they cannot enter cells or survive once inside, will produce colonies on plates from the second set of wells but will have no colonies on plates from the third set of wells. 3. By comparing the adherence frequencies and the invasion frequencies for wild- type bacteria versus mutants, it is possible to determine the nature of the virulence factor that was defective in the mutant bacteria. 3.3.2.3 Invasion Success Curve Studying pathogenic mechanisms of Campylobacter jejuni, a food-borne pathogen that causes diarrhea, is hampered by the lack of simple animal models that mimic human disease, so cell culture assays have provided useful alternative ways to investigate C. jejuni interaction with host epithelial cells. Use of the gentamicin assay has shown that invasiveness varies considerably depending on (1) the C. jejuni strain, (2) the human cell lines used, (3) the number of bacteria in the inoculum (multiplicity of infection, or MOI), and (4) other assay conditions. Researchers in the field have consequently sought to standardize assay conditions and to set up tests for comparisons between strains. One such test is to generate an invasion success curve (Figure 3.5), which plots the log (MOI/cell) versus the log (invaded bacteria/cell) to give the minimum MOI (minMOI) required to obtain the maximum number of internalized bacteria per cell (BImax). In comparing strains, those with lower minMOIs are more invasive. Page 21 of 26 Figure 3.5: Invasion success curve. Shown is an invasion success curve for an invasive Campylobacter strain. The average number of invaded bacteria per mammalian cell was calculated for each well in a plate over a wide range of MOIs, starting with a low MOI and relatively small numbers of internalized bacteria, until a maximal invasion plateau (BImax) was reached, where increasing the MOI no longer increased the number of internalized bactcria. Each data point represents the results obtained for one plate well. The BImax and minMOI (lowest MOI required to reach the BImax) values are indicated by arrows. 3.3.3 Plaque Assay for Cell-to-Cell Spread Some intracellular pathogens, such as Legionella and Chlamydia, are able to invade eukaryotic cells, multiply within the eukaryotic cell until they burst from the host cell, and then spread from one host cell to the adjacent host cell. During this process (called cell-to-cell spread), the bacteria replicate within each invaded host cell, killing the host cell in the process and creating a cleared zone of killed cells (called a plaque) around the initial cell that was invaded. Some pathogens, such as Listeria and Shigella, are also able to spread laterally in a monolayer by propelling themselves into the adjoining cell without being released into the medium. Researchers have developed a modified gentamicin tissue culture assay that allows the assessment of an intracellular pathogen's ability to spread from cell to cell. In this plaque assay, (1) tissue culture cells are grown on plates to form an even confluent monolayer, and (2) bacteria are added to the medium and (3) incubated Page 22 of 26 with the mammalian cells for a short time to allow invasion to occur (Figure 3.6). (4) The medium with unattached bacteria is removed, and (5) the monolayer is incubated with gentamicin to kill any remaining extracellular bacteria. (6) The monolayer is then gently covered with another layer of agar containing gentamicin (to prevent diffusion of the bacteria through the medium), and the cells are further incubated. (7) After a while, the living cells are stained, and plaques (cleared areas) in the monolayer can be observed where bacteria have invaded and spread from cell to cell, killing the mammalian cells in the process. Mutants that are defective in efficient cell-to-cell motility form small plaques, while mutants that are defective in factors necessary for intracellular survival or replication do not form any plaques. Figure 3.6: Plaque assay for assessing cell-to-cell spread by an intracellular pathogen. Page 23 of 26 3.3.4 Fluorescence Microscopy Techniques for Assessing Effects of Pathogens on Host Cells 3.3.4.2 General Consideration The development of new reagents for probing the interior of a mammalian cell or changing its chemical environment has generated opportunities for sophisticated new approaches to studying the bacterium-host cell interaction. For example, fluorescent dyes attached to specific reagents (chemicals or proteins) or monoclonal antibodies that bind specifically to host cell cytoskeletal components, such as actin, have been used to follow by fluorescence microscopy the cytoskeletal rearrangements caused by bacteria attaching to and invading host cells. These methods have also been used to observe the changes in cellular morphology caused by toxins or effector proteins produced by the bacteria. 3.3.4.2 Intracellular Responses due to Bacterial Invasion 1. One particularly useful reagent for visualization of cytoskeletal rearrangements caused by the formation of actin filaments in cells is the marine metabolite phalloidin, which is a compound that binds very tightly to polymerized F actin (which forms the actin filaments) but not to monomeric, free G actin. When phalloidin is linked to a fluorescent dye, such as rhodamine (red) or fluorescein isothiocyanate (green), fluorescence microscopy can be used to monitor actin cytoskeletal changes and actin stress fiber formation. 2. Calcium release from intracellular stores (called calcium mobilization) is another intracellular response often triggered by interactions of a cell with bacteria or bacterial toxins. A number of fluorescent dye reagents (such as Fura-2) that can detect calcium levels inside tissue culture cells are available. 3.3.4.3 Markers to Visualize the Location of Pathogenic bacterium or Its Protein A number of host cellular markers have been developed that allow researchers to visualize where bacteria or bacterial proteins are located inside host cells. These cellular markers are mammalian proteins that are known to localize to particular compartments within the host cell. They are introduced into the mammalian cells by transfection with mammalian expression vectors (usually plasmids or retroviruses) carrying genes for these cellular marker proteins that have been modified with tags, such as green or red fluorescent proteins or epitope tags that are recognized by antibodies conjugated to fluorescent dyes. The tags can be used to visualize the Page 24 of 26 proteins inside the host cells by fluorescence microscopy. These are only a few examples of reagents and approaches that are currently being generated by cell biologists to study intracellular processes but which are proving equally applicable to studies of bacterium-host cell interactions. 3.3.5 Organ Culture Models 3.3.5.1 General Consideration Organs or portions of them can now be kept viable in vitro for longer and longer periods of time. Such tissues are called organ cultures. In the case of organ cultures, techniques for in situ quantitative detection of proteins and mRNA make it possible to detect elevated expression of bacterial genes in a particular tissue. An advantage of using organ cultures is that there are usually multiple cell types present, including some from the immune system, which might allow a better approximation of what is happening during a natural infection. Organ cultures provide a much better model than tissue culture cells of what transpires in an animal or human host, but they may be more difficult to obtain and to maintain. An organ culture can begin to deteriorate within hours or days, making it difficult to do long- term experiments, yet there have been successes. Ex vivo organ culture models have been particularly helpful in cases where research has been hampered by lack of a suitable infection model. An example is the case of the important food-borne pathogen C. jejuni, in which researchers developed an organ culture model using human gastrointestinal tissue, obtained from endoscopic biopsies of patients, to highlight the surprising propensity of C. jejuni to adhere to mucosal tissue via its flagellum (not pili). 3.3.5.2 Sources of Tissues and Organs Scientists who study the adherence of bacteria to skin cells have benefited enormously from the popularity of cosmetic surgery, which generates large amounts of skin tissue. Similarly, hysterectomies make fallopian tube and uterine tissues available, although the availability of these tissues has dropped considerably in recent years due to a trend away from performing complete hysterectomies. However, the number of people lining up to donate portions of their liver or heart is rather limited. Page 25 of 26 3.3.4.3 Artificial Organ Cultures Because of the limited supply of organ cultures from donors, scientists have begun to develop artificial organ cultures. One example of this is artificial skin equivalents, which are made by culturing a suspension of human foreskin fibroblasts in a matrix of native acid-soluble collagen- and serum-supplemented medium at 37°C, where the collagen polymerizes and traps the cells, which then elongate and spread for several days. Then, a freshly isolated suspension of human skin-derived keratinocytes is seeded onto the surface of the collagen-fibroblast matrix, and the keratinocytes are allowed to grow to cover the surface, depositing basement membrane beneath them, differentiating into epidermal cells, and leading to formation of skin layers. Other connective-tissue and bone, heart, liver, and neuronal-tissue equivalents are also being developed, and one day, hopefully, we will see these systems utilized for host-bacterium interaction studies. Reference Bacterial Pathogenesis: A Molecular Approach, Third Edition by Brenda A Wilson, Abigail A Salyers, Dixie D Whitt and Malcolm E Winkler. 2011. ASM Press, American Society of Microbiology, Washington DC. Page 26 of 26