DNA Microarrays Tackle Infectious Disease

Faced with the duty of informing the public on the never-ending barrage of emerging and reemerging diseases, as well as on the threat of bioterrorism, infectious disease researchers must be equipped with an armament of laboratory tools designed to deal with complex biological problems.

Fortunately, however, genomics has lightened their load. For the first time in the history of infectious disease research, the genetic sequence of many infectious agents makes it possible to design biological assays unimaginable a decade ago. It is no longer necessary to study the expression of one gene at a time, because genomics has made it possible to study multiple genes simultaneously.



By enabling researchers to perform simultaneous, genome-wide gene expression analysis, DNA microarrays have been used to tackle previously insurmountable biological problems. Even more exciting is the fact that DNA microarrays have been used in infectious disease research in other ways as well. Many researchers are using DNA microarrays to study infectious disease, with a major focus on host-pathogen interactions and technical problems associated with current DNA microarray analysis. They are also developing many new ways to resolve such problems.


Host-pathogen interactions
Frank DeLeo, PhD, an investigator at the laboratory of bacterial pathogenesis at the National Institute of Allergy and Infectious Disease, Bethesda, Md., uses genomics to understand host-pathogen interactions that occur during the innate immune response. While investigating the response of neutrophils to bacterial pathogens such as methicillin-resistant Staphylococcus aureus and Streptococcus pyogenes, DeLeo has used human microarrays extensively to understand, at the transcriptional level, neutrophil activity during the inflammatory response.

DeLeo mainly uses oligonucleotide-based human DNA microarrays from Affymetrix Inc., Santa Clara, Calif., to look at neutrophil gene expression. But he has also used custom designed Affymetrix DNA microarrays that contain the complete genomes of eight strains each of S. aureus and S. pyogenes. He has also used pathogen-specific microarrays to examine, at the transcriptional level, the response of S. aureus and S. pyogenes to attack by neutrophils.

“Our approach has been to characterize how the pathogen interacts with human neutrophils and how it behaves in an animal model, and then use the microarray approach to try to identify from sort of a global gene expression perspective what phenotypes confer with what we see in the animal model,” DeLeo says.

As a result of the microarray analysis, he was able to pull out a number of pathogen genes with no known function that he later targeted for knockout; the effect of knockout on pathogen survival was then analyzed in ex vivo neutrophil assays and in an animal model.

While working with DNA microarrays in his experiments, DeLeo noticed one major shortcoming: the conundrum of how to standardize the microarray data for analysis. He says there are many ways to do this, including normalization of signal across the entire microarray, normalization based on an internal control composed of a set of housekeeping genes, or sampling nearby tissues or resting cells. Because systems are so different among different investigators, it is sort of hard to figure out a standardized approach of normalizing microarray data.”

There is also the issue of choosing the relevant control to use as a reference for future experiments. DeLeo says researchers should always remember to ask themselves how relevant their control sample is.

It is especially difficult to find relevant control samples for studies using animal models of human bacterial infections, so most researchers use in vitro-grown bacterial cultures. DeLeo says a number of labs are addressing this issue, including that of Pat Brown, PhD, in the department of biochemistry at Stanford University, Palo Alto, Calif., who recently published a study in which he and his colleagues performed studies on placenta where they had a standard set of control RNAs. “That seemed to work pretty well. It’s a tough question, and it is probably going to be very difficult to standardize,” DeLeo says.

The results of microarray analysis should be validated to determine if changes in gene expression are real or artifacts, and this is typically done using a real-time reverse transcriptase quantitative polymerase chain reaction (PCR) method.

However, because of the cost of the PCR enzyme, this assay can be expensive. So, in the interest of efficient use of limited research dollars, some researchers opt not to validate every microarray experiment if they already know their system well.

“But if you are characterizing a system for the first time, it is very important that you know that what you are seeing is real, especially if it is a new array or a system that you are not familiar with.”



ChIP on Chip
Historically, DNA microarrays, also known as DNA chips, were strictly used to detect changes in gene expression. Recently, however, researchers expanded the use of these arrays into other areas, including DNA-protein interactions. The interactions between transcriptional proteins and DNA are crucial to the regulation of cellular gene expression.

The chromatin immunoprecipitation method (ChIP) has been used primarily to isolate and identify DNA-protein interactions. The fusion of ChIP and DNA chip technologies creates a very powerful tool for identifying DNA-protein interactions on a global genomic scale.

Richard Jenner, PhD, a postdoctoral fellow in the laboratory of Richard Young, PhD, a professor of biology at the Massachusetts Institute of Technology, Cambridge, Mass., also incorporates DNA microarrays in his infectious disease research, but for a different reason. Jenner’s research focuses on how the human immune system is regulated and how pathogens such as bacteria and viruses can alter this regulation to maintain infection in immunocompetent individuals. Specifically, he uses DNA microarrays to understand the host cell’s transcriptional response to infection.

To understand these biological processes, Jenner developed an approach called “ChIP on Chip” which uses genomewide location analysis. This couples a custom-made, oligonucleotide-based DNA microarray analysis with a chromatin immunoprecipitation (ChIP) assay, allowing researchers to locate binding on the host genome of specific host-transcription factors and transcriptional regulators.

Jenner uses DNA microarrays to identify DNA fragments to which the immunoprecipitated protein of interest is bound. In his system, binding of a transcription factor or regulator to a specific gene is defined by an enrichment of signal for that gene in the ChIP sample relative to a whole cell genomic DNA control. However, the yield of DNA fragments from ChIP is not adequate for microarray analysis.

To circumvent this problem, the isolated DNA was amplified by ligation-mediated PCR (LMPCR) prior to microarray analysis in order to amplify the signal.

While working on these experiments, Jenner noticed a few limitations in the microarray technology. For one, the number of genes on the array, otherwise known as the array coverage, limits current human DNA microarray technologies.

“There’s still some problems with being able to get a probe to every part of the genome,” he said. “However, if you have a probe tiling across the whole genome, you have to have millions and millions of probes . . . and most microarray technologies don’t have the density for this level of coverage, causing researchers to need to use multiple arrays to obtain that level of coverage.”

One thing that would improve arrays and assist experiments would be to increase the number of features on a single array, reducing the number of arrays needed. “By adding more features to the arrays, you can start having redundant probes, that is, probes with different sequences representing the same area of the genome, to see how reproducible the results with the different oligos are.”


The in vivo factor
In order to determine the host- and pathogen-related factors that lead to human infectious disease, researchers must be able to study molecular pathogenesis in vivo in humans. Caroline Attardo Genco, PhD, professor of infectious diseases and microbiology at Boston University School of Medicine, studies gene expression of Neisseria gonorrhea during in vivo infection in humans. To study gene gonococcal expression, her laboratory uses a custom made DNA microarray based on partial sequence information of the N. gonorrhea genome.

“The problem with the gonococcus is that the full genome sequence has not been published yet. And so there was not an array available for this organism,” Genco says. However, there was an array available for Neisseria meningitides, an organism that is about 95% homologous with N. gonorrhea. Several strains of N. meningitides were sequenced in early 2000, however, and an array was made from them. “So, we did all of our work with N. meningitides array and we published that in vitro work with whole genome array. Then we took genes of interest from the meningococcal array and searched the gonococcal genome at the time and made our own specific gonococcal array,” Genco says.

The microarrays were used to determine whether particular genes are expressed during active infection in humans (mostly males) with gonorrhea. Genco says that while the majority of men who get infected with gonorrhea have symptoms, females infected with the bacteria don’t get inflammation or other symptoms of the disease. But women are the primary carriers of the disease because they are unaware they are infected and go on to pass it on to another partner.

“The million dollar question from people trying to understand gonococcal transmission is how is it that women get infected with these bacteria, but they don’t get disease and they don’t get inflammation?”

“What we are trying to understand is whether there is something in particular about the genes expressed by the gonnococcal bacteria in a male versus a female that allows it to colonize within a particular site but not induce inflammation. We think that by studying the bacteria directly from the specimen without culturing them to look at which genes are expressed is a great way to go about it,” Genco says.

There are two major problems that Genco faces in processing specimens for microarray analysis. First, once specimens are collected from the genital lesions, the RNA must be isolated immediately to avoid loss of RNA due to degradation.

Second, the number of samples that the clinic receives for a study varies over time, and because microarrays do not typically have a long shelf life, the microarray can expire long before all samples are collected.

“In real life in doing any kind of clinical work, you do have an issue of when you get the specimen, how long it sits, as well as the shelf life of the arrays themselves. So that’s the biggest issue we have had is the shelf life of arrays.”

Reproducibility is a fundamental requirement for any quantitative assay, including microarray analysis. Furthermore, because of the amount of data generated by a microarray analysis, as well as the degree of variability inherent in hybridization-based methodologies and in biological systems, some researchers find it difficult to obtain reproducible microarray data.



“I think that if you are talking about in vitro work, you just have variability from sample to sample—that’s inevitable. There’s nothing technical about it, and I think you just need to repeat an experiment at least four or five times. You just need to do enough replicates of it,” Genco says.

“When looking at in vivo work, there’s just variability when you look at samples from patient to patient. There is just so much variability that you need to do a lot of replicates to see statistically significant differences.”

Genco believes that looking at bacterial expression in vivo is the wave of the future, and she would like to eventually be able to correlate expression of specifi c bacterial genes with the type of infection that patients get.

“It’s an interesting project because no one has published anything looking at bacterial gene expression in humans. There’s been some work in animals where people give pathogens to animals and see how these genes are expressed in different mouse models of infection. But as far as a human infection goes, people just have not really tackled it.”