Iowa State University Capstones, Theses and Retrospective Theses and Dissertations Dissertations

2005 Genetic profiling of Mycoplasma hyopneumoniae Melissa L. Madsen Iowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/rtd Part of the Microbiology Commons, Molecular Biology Commons, and the Veterinary Medicine Commons

Recommended Citation Madsen, Melissa L., "Genetic profiling of Mycoplasma hyopneumoniae " (2005). Retrospective Theses and Dissertations. 1793. https://lib.dr.iastate.edu/rtd/1793

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. NOTE TO USERS

This reproduction is the best copy available.

® UMI

Genetic profiling of Mycoplasma hyopneumoniae

by

Melissa L. Madsen

A dissertation submitted to the graduate faculty in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Molecular, Cellular and Developmental Biology

Program of Study Committee: F. Chris Minion, Major Professor Daniel S. Nettleton Gregory J. Phillips Eileen L. Thacker Eve Wurtele

Iowa State University

Ames, Iowa

2005 UMI Number: 3200480

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

UMI

UMI Microform 3200480 Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 ii

Graduate College Iowa State University

This is to certify that the doctoral dissertation of

Melissa L. Madsen has met the dissertation requirements of Iowa State University

Signature was redacted for privacy.

Major Professor

Signature was redacted for privacy.

For the Major Program iii

TABLE OF CONTENTS

LIST OF FIGURES vii

LIST OF TABLES viii

ABSTRACT ix

CHAPTER 1. GENERAL INTRODUCTION 1 Introduction 1 Dissertation Organization 1 Literature Review 2 Mycoplasmas 2 Mycoplasma hyopneumoniae 4 Disease characteristics 4 Detection 5 Vaccines 6 Virulence mechanisms 8 Microarrays 9 History 9 Types of arrays 10 Informatics 14 Applications 17 Genomic comparisons (within/without species) 17 Single nucleotide polymorphism detection 18 Gene expression 19 Pathogens 19 Food safety 20 Mycoplasmas 21 Summary with Goals and Hypothesis 22 References 23

CHAPTER 2. TRANSCRIPTIONAL PROFILING OF MYCOPLASMA HYOPNEUMONIAE DURING HEAT SHOCK USING MICROARRAYS Abstract 37 Introduction 38 Materials and Methods 39 Mycoplasma strains and culture conditions 39 Microarray construction 39 RNA isolation 43 Random primer set for M. hyopneumoniae 43 Target generation and hybridization 44 Image acquisition and data analysis 44 IV

Statistical analysis 45 Validation of microarray data 45 Results 46 Primer design and array construction 46 Heat shock studies 47 Validation of microarray data 53 Discussion 54 Acknowledgements 57 References 57

CHAPTER 3. TRANSCRIPTIONAL PROFILING OF MYCOPLASMA HYOPNEUMONIAE DURING IRON DEPLETION USING MICROARRAYS Abstract 61 Introduction 62 Materials and Methods 63 Mycoplasma strains and culture conditions 63 Microarray 64 Experimental design 64 RNA isolation 64 Target generation and hybridization 64 Data acquisition and normalization 65 Data analysis 65 Confirmation of microarray data 66 Results 67 Iron depletion studies 67 Microarray studies 67 Validation of microarray data 69 Discussion 72 Acknowledgements 75 References 75

CHAPTER 4. GENE EXPRESSION DIFFERENCES BETWEEN MYCOPLASMA HYOPNEUONIAE VIRULENT STRAIN 232 AND A VIRULENT STRAIN J Abstract 79 Introduction 79 Materials and Methods 81 Mycoplasma strains and culture conditions 81 Microarray 81 Experimental design 81 RNA isolation 82 Target generation and hybridization 82 Image acquisition and normalization 83 Data analysis 83 V

Results 84 Microarray studies 84 Discussion 86 Acknowledgements 87 References 88

CHAPTER 5. TRANSCRIPTIONAL VARIATION BETWEEN HIGH AND LOW ADHERENT VARIANTS OF MYCOPLASMA HYOPNEUMONIAE Abstract 91 Introduction 91 Materials and Methods 93 Mycoplasma strains and culture conditions 93 Microarray 93 Experimental design 94 RNA isolation 94 Target generation and hybridization 94 Image acquisition and normalization 95 Data analysis 95 Results 96 Microarray studies 96 Discussion 98 Acknowledgements 99 References 102

CHAPTER 6. GENETIC ANALYSIS OF FIELD ISOLATES BY MICROARRAY Abstract 107 Introduction 107 Materials and Methods 109 Mycoplasma strains and culture conditions 109 Microarray 109 Experimental design 109 DNA isolation 110 TempliPhi™ reactions 110 Nebulization 110 Target generation and hybridization 111 Data acquisition and normalization 111 Data analysis 111 Results 112 TempliPhi™ reactions 112 Microarray studies 113 Discussion 116 Acknowledgements 117 References 117 VI

CHAPTER 7. GENERAL CONCLUSIONS 121 General Discussion 121 Recommendations for Future Research 123 References 128

APPENDIX A. CIRCULAR REPRESENTATION OF THE MYCOPLASMA HYOPNEUMONIAE GENOME STRUCTURE 130

APPENDIX B. PRIMERS USED IN THE MYCOPLASMA HYOPNEUMONIAE MICROARRAY 132

APPENDIX C. PRINTING FILE FOR BIOROBOTICS MICROGRID PRINTING ROBOT 146

APPENDIX D. HEXAMER OLIGONUCLEOTIDE SEQUENCES USED FOR GENERATING CDNA FOR LABELING EXPERIMENTS 151

APPENDIX E. OPEN READING FAMES MISSING FROM THE MYCOPLASMA HYOPNEUMONIAE MICROARRAY 153

ACKNOWLEDGMENTS 156 vii

LIST OF FIGURES

Figure 2.1. Flow chart for PCR primer design 41 Figure 2.2. Example of agarose gel of purified PCR products 46 Figure 2.3. Determination of cross-linking energy 47 Figure 2.4. Volcano plot of transcriptional differences in M. hyopneumoniae during heat shock conditions 48 Figure 2.5. Semi-quantitative RT-PCR 53 Figure 3.1. Effect of 2,2'-dipyridyl on growth of M. pulmonis and M. gallisepticum 68 Figure 3.2. Volcano plot of transcriptional differences in M. hyopneumoniae during low iron growth conditions 69 Figure 3.3. Semi-quantitative RT-PCR 71 Figure 4.1. Volcano plot of transcriptional differences in M. hyopneumoniae strains 232 and J 84 Figure 5.1. Volcano plot of transcriptional differences in M. hyopneumoniae high-adherent variant 91-3 and low-adherent variant 60-3 98 Figure 6.1. Analysis of TempliPhi™ amplified and nebulized mycoplasma chromosomal DNA 113 Figure 6.2. Comparison of mean log signal intensity values of genomic vs TempliPhi™ amplified chromosomal DNA 114 Figure 7.1. Steps for microarray construction and utilization. 122 vin

LIST OF TABLES

Table 2.1. Initial Primer3 criteria for designing primers for the M. hyopneumoniae ORFs s. 42 Table 2.2. Heat shock up-regulated genes in M. hyopneumoniae 49 Table 2.3. Heat shock down-regulated genes in M. hyopneumoniae 51 Table 2.4. RT-PCR primers 53 Table 3.1. Differentially expressed genes of M. hyopneumoniae during growth under iron limiting conditions 70 Table 3.2. RT-PCR primer sequences. 71 Table 4.1. Transcriptional differences between M. hyopneumoniae strains 232 and J/KO.01 85 Table 5.1. Variation in signal intensities between variants 60-3 and 91-3 at

^<0.05 97 Table 6.1. Genetic variation of M. hyopneumoniae field isolates compared to strain 232 withp<0.05 115 Table. 7.1. Genes with significant (p<0.05) transcript values in both heat- shocked and iron-depleted studies 125 Table 7.2. Putative housekeeping genes 127 ix

ABSTRACT A microarray was constructed and applied to transcriptional profiling and genetic variation studies of Mycoplasma hyopneumoniae. The genome sequence enabled the construction of the microarray to allow a global approach to understanding fundamental processes in M. hyopneumoniae. These studies focused on whether M. hyopneumoniae regulates its genes under different environmental conditions and if genetic changes can be correlated with virulence. The microarray consisted of 632 open reading frames represented by polymerase chain reaction products and were used in a two-color experimental design. Data were analyzed using a mixed linear statistical model. Unique features implemented in these studies included the printing of two complete arrays per substrate, reducing the slide to slide variation; the scanning of each dye channel of each array at different laser power settings to increase the dynamic range of expression measurement; and the use of a unique set of hexamer primers to generate fluorescently labeled targets for microarray analysis. The first series of studies focused on transcriptional profiling during heat shock and iron deprivation, two environmental changes that M. hyopneumoniae encounters on the respiratory epithelial surface during disease. Key genes responsive to these stresses were identified, and interestingly, fifty-three were regulated in common to the two conditions at p<0.05. Since adherence to swine cilia is a prerequisite for colonization and disease, the next series of studies involved the comparison of an adherent, pathogenic strain 232 to a nonpathogenic, nonadherent type strain J. In conjunction with this study, two strain 232 adherence variants, one high and one low, were also compared. Thirty genes were identified that differed between the strains 232 and J. Nineteen genes were up-regulated in the high adherent variant including the heat shock protein DnaJ, and fifteen genes were up-regulated in the low adherent strain. In a comparison of eight field isolates to strain 232, two were indistinguishable from strain 232 and the others varied in an many as twenty-five loci and as few as one. In summary, these studies attempt to unravel some of the underlying mechanisms and pathways M. hyopneumoniae uses to infect the host and cause disease. 1

CHAPTER 1. GENERAL INTRODUCTION

Introduction Estimated annual economic loss to the US swine industry is greater than $200 million due to Mycoplasma hyopneumoniae, the etiological agent of swine pneumonia. Adherence of the organism to the cilia of the swine lung is necessary to cause disease, however, current research has been unable to determine the pathogenic mechanisms by which M. hyopneumoniae causes disease. Research has been hindered due to the fastidious growth of M. hyopneumoniae and the inability to study the organism using traditional molecular techniques. This challenge can be overcome to some extent using microarray technology, which gives researchers a tool to examine a genome and its transcriptional regulation on a global basis. The work outlined in the studies below will increase the understanding of the mechanisms by which M. hyopneumoniae regulates gene function in response to environmental conditions. Additionally, studies examining transcriptional variation among adherence variants will give insight into mechanisms of pathogenesis. Finally, genomic evaluation can provide insight into the conserved genome of the organism and genetic variation that may occur in the field. These studies will lay the foundation for future experiments that will continue to aid in our understanding of pathogenesis and the development of novel preventive and therapeutic strategies.

Dissertation Organization This dissertation is organized into several chapters, five of which will be submitted for publication and includes a final summary and direction of future studies. Chapter one is the introduction containing the background information regarding the disease, application of novel technology and justification of the research. Chapters 2 and 3 describe environmental studies examining the response of M. hyopneumoniae to heat shock and low-iron conditions. Chapter 4 examines transcriptional differences between the pathogenic strain 232 and the non-pathogenic strain J. Transcriptional differences between strain 232 and the high-adherent and low-adherent clone variants are discussed in Chapter 5. Evaluation of genetic variation found among field isolates and strain 232 are the focus of Chapter 6. General conclusions and ideas for future research comprises Chapter 7, and five appendices complete the dissertation. 2

Literature Review

Mycoplasmas Mycoplasmas infect a wide range of hosts, including humans, mammals, fish, reptiles, arthropods and plants (71). They are also often found as common contaminants of tissue culture cell lines. Mycoplasmas are members of the class Mollicutes, which includes eight genera including the pathogenic Mycoplasma, Ureaplasma and Spiroplasma (41, 99). There are more than 200 species that have been identified with additional species being reported every year (http://www.the-icsp.org/subcoms/mollicutes.htm). Many of these newly identified species are pathogens responsible for what are thought to be newly emerging diseases in animals and plants (10-12, 61). Some genera, i.e., Acholeplasma, are considered saprophytes, but even that has come into question (99). Mycoplasmas inhabit a wide variety of niches, but for the most part, they are host-specific. Unique features delineate the class Mollicutes. The defining characteristic of this class is the lack of cell walls, unique among the eubacteria (98). This deficiency is thought to be the result of degenerative evolution resulting in a significant reduction in genome size (70). In fact, with a diameter of 0.3-1 and genomes ranging from 580 to 2220 kilobase pairs, mycoplasmas are among the smallest self-replicating organisms known (97). Some of the ramifications of genome reduction have been the loss of the metabolic pathways including those involved in amino acid, purine and pyrimidine, and phospholipid synthesis. Also, the loss of the biosynthetic capacity to generate a cell wall has caused the mycoplasmas to be highly dependent on an osmotically stable environment, such as mucosal linings of their hosts or the internal regions of plants. Several species have been able, however, to retain a semblance of stable morphology indicating the presence of proteins that serve as cytoskeletal-like structures (99). Although not classified as such, mycoplasmas actually persist as obligate parasites obtaining their biosynthetic precursors from their hosts. For animal pathogens, these precursors are obtained from the mucosal surfaces upon which they reside, and from mucosal secretions and/or cell surfaces to which they are attached. For the sake of clarity, the generic term mycoplasmas will be used in reference to the members of the animal pathogenic genus Mycoplasma throughout the remainder of this dissertation. The need for the acquisition of macromolecular precursors limits the types of environments in which mycoplasmas can prosper and grow. While they may survive for short periods of time on external surfaces, they can only grow and survive long term on host tissues. There they reside on mucosal surfaces, occasionally becoming septicemic and moving to other tissues (74, 114). Required nutrients are available either on the cell surfaces 3 to which they are attached or in body secretions. They are also thought to reside intracellularly on occasion (5, 104), although that has not been rigorously tested; studies have only been reported using tissue culture systems (5, 104, 136). The exceptions are the newly reclassified hemotrophic mycoplasmas(73). With this in mind, mycoplasmas must have mechanisms for the acquisition of amino acids, purines and pyrimidines, phospholipids and cholesterol, the building blocks of macromolecules and membranes. Few studies have focused on surface activities with the potential to participate in these acquisition mechanisms (54, 76, 77), and it is thought these systems may play important roles in pathogenesis. Whether mycoplasmas reside on cell surfaces or are suspended in complex growth media, they must assimilate phospholipids and cholesterol from their environment and use those materials (sometimes with little modification) to produce their own membrane. This feature of mycoplasmas led to their use as models of membrane structure, and these models were instrumental in development of the mosiac model of membrane structure (96). The small size of the mycoplasma genome has also led researchers to use them as models for a minimal genome (40, 51, 52). These studies will help to define the minimal requirements of life (52) and possibly lead to the construction of the first completely synthetic cell (J. I. Glass, N. Alperovich, N. Assad-Garcia, H. Baden-Tillson, C. A. Hutchison III, H. Khouri, M. Lewis, F. C. Minion, W. C. Nierman, W. C. Nelson, C. Pfannkoch, K. Remington, V. Subba, S. Yooseph, H. O. Smith, and J. C. Venter. Abstr. 14th Congr. Int. Org. Mycoplasmol. abstr. 49, 2002). The mycoplasma genome is not only small in size, but it also has unusual features. Mycoplasmas contain a relatively low G+C content of 27-32 mol%. The one outlier is the human pathogen Mycoplasma pneumoniae with a 40 mol% G+C content (46). The low G+C content of the genome is thought to result from a strong AT-biased mutation pressure that has operated during the evolution of mycoplasmas (83). The genetic systems of mycoplasmas have unusual features as well. For example, they utilize UGA as a tryptophan codon rather than as a stop codon (83, 139). They also contain a minimal set of tRNAs (approximately 20 of the 62 potential), but yet can translate all available codons (83). How gene expression is controlled has yet to be defined in mycoplasmas. With the advent of high-throughput sequencing, many of the mycoplasmas have been among the first genomes sequenced. To date, ten mycoplasma genome sequences have been completed [M. pneumoniae (46), Mycoplasma genitalium (40), Mycoplasma pulmonis (18), Ureaplasma urealyticum (42), Mycoplasma gallisepticum (91), Mycoplasma penetrans (105), Mycoplasma mycoides subsp. mycoides SC (133), Phytoplasma asteris Onion Yellows strain (88), Mycoplasma mobile (53), and M. hyopneumoniae (79)] and more are under analysis (Mycoplasma alligatoris, 4

Mesoplasma florum, Mycoplasma arthritidis, Mycoplasma bovis, Mycoplasma capricolum subsp. capricolum, Mycoplasma fermentans, Mycoplasma haemofelis, Spiroplasma kunkelii, Spiroplasma citri, Mycoplasma agalactiae). These sequencing projects promise to accelerate our understanding of the molecular basis of pathogenesis. Much research has gone into the study of mycoplasmas associated with human disease. Two such organisms are M. pneumoniae and M. genitalium, which have been well studied and characterized. The sequencing of M. pneumoniae and M. genitalium has enhanced the molecular characterization of many of the surface adhesions, cytoskeleton proteins and their accessory proteins, allowing for greater understanding of their pathogenic mechanisms. Comparative studies also indicate their parasitic lifestyle makes them highly dependent on well-defined transport systems, delivering the necessary exogenous precursors and nutrients to the cells (99).

Mycoplasma hyopneumoniae

Disease characteristics Disease caused by M. hyopneumoniae, the causative agent of enzootic pneumonia, is characterized by low mortality but high morbidity of swine. Mycoplasma hyopneumoniae attaches to the cilia of the respiratory epithelium of the airways. The disease is characterized by a chronic, dry, non-productive cough 10-14 days post infection usually without any signs of lethargy, anorexia or fever (126). Lung lesions observed in pigs infected with M. hyopneumoniae appear as tan to dark purplish areas of lung consolidation occurring primarily in the cranioventral areas of the lung and are slow to develop, taking 2 to 3 weeks to appear (126). The economic losses associated with M. hyopneumoniae cost producers over $200 million annually. The losses are often associated with decreased feed efficiency, resulting in longer times to attain market weight. Mycoplasma hyopneumoniae has also been identified as a component of the porcine respiratory disease complex (PRDC). The damaging effects to lung tissue and a heightened immune response during co-infection with porcine reproductive and respiratory syndrome virus are exacerbated by infection with M. hyopneumoniae (125). Although commercially available vaccines decrease the number of organisms present, they do not completely prevent colonization of the lungs and upper respiratory tract. Prophylactic use of antibiotics does not prevent infection by M. hyopneumoniae. More effective vaccines and prevention practices are needed to limit the spread of M. hyopneumoniae. 5

Several obstacles prevent traditional molecular techniques from being used to study M hyopneumoniae. It is difficult to culture due to its fastidious growth and nutritional requirements that necessitate complex culture media. In addition, transformation studies are problematic because M. hyopneumoniae is difficult to grow on agar surfaces preventing isolation of single mutants or colonies. Mycoplasma hyopneumoniae like all mycoplasmas uses the stop codon UGA to code for tryptophan, making gene cloning and protein expression studies a challenge. With the completed genome sequencing of M. hyopneumoniae, annotated data suggests many gene products may contribute to M. hyopneumoniae pathogenesis (80).

Detection Much research has focused on detection methods for M. hyopneumoniae to use in diagnostic applications. Culturing M. hyopneumoniae from clinical samples is a challenge due to competition of multiple bacteria recovered from lung samples. Even when antibiotics are used to prevent growth of walled bacteria, Mycoplasma hyorhinis, a common pig colonizer, outgrows M. hyopneumoniae cultures preventing its isolation and detection. Serological detection of the organism is often unreliable due to the cross-reactions with other mycoplasmas (7). Also, the current enzyme linked immunosorbent assays (ELISA) have inherent problems. There is no correlation between antibody levels and protection. Also, a study by Erlandson et al. gave false negative rates for three commercially available ELISA kits in the range of 34.1% to 61.0% (K. Erlandson, B. Thacker, M. Wegner, R. Evans, and E. Thacker. Abstr. 17th Congr. Int. Pig Vet. Soc., abstr. 249, 2002). If a vaccine is given, those animals will test positive as a result of the antibodies they produce. There are no "marked" vaccines that can differentiate between vaccinated and non-vaccinated individuals. Detectable levels of antibodies in infected pigs can vary depending on the time when the animal sero-converts. Challenged pigs have shown variation in sero-conversion from two to four weeks post infection or longer (125). This variability, as well as a lack of agreement for consistent cut-off thresholds, is indicative of the need for better diagnostic tests. Despite these limitations, the ELISA continues to be the assay of choice for most diagnostic laboratories around the world. The use of polymerase-chain reaction (PCR) to detect organisms has greatly improved sensitivity and decreased diagnostic times (4, 65). Specific primers are designed flanking the target of interest, so some knowledge of the DNA sequence is required to design the oligonucleotide primers. The DNA is amplified exponentially using a thermal stabile polymerase, which allows for repeated rounds of denaturation, annealing and elongation 6

without adding additional polymerase after each round. Thermal cyclers allow consistent temperature and for streamlined automation. The amplified product can then be sized, sequenced, used for genotyping, and used as probes in hybridization experiments. The sensitivity of PCR makes it an appealing candidate for diagnostic tests, but it can be hampered by contamination and poor choice of sampling site. Since negative and positive controls are run in all evaluations, the false positive rate is low. Several research groups have identified individual genes with unique sequences for specific identification of M. hyopneumoniae by PCR (16, 65, 128), and the resulting tests range in sensitivity and specificity (65). During assay development, samples were collected from experimentally infected pigs, field isolates or from diagnostic labs. Sample sites are critical for the proper performance of the PCR test, however. Results can vary depending on how samples are collected, the environment, the disease state of the animal and the types of samples examined. A recent study by Kurth et al. showed that nasal swabs were not reliable for detecting M. hyopneumoniae infections, but bronchial alveolar lavage samples proved 100% accurate in detecting infected pigs using a nested PCR assay (65). At the point of necropsy, gross lesion characteristics and immuno-histochemistry tests can be used to identify diseased animals, but those with a low level of infection can be problematic. Thus, the inclusion of BAL samples in the diagnostic repertoire will improve the detection rate for M. hyopneumoniae-infected pigs. The detection systems that have been developed in the last several years are capable of identifying infected pigs, but most pigs throughout the world are considered to be M. hyopneumoniae positive. Thus, improved diagnostics has not led to decreased incidence of pneumonia nor improved herd health. The issue is no longer one of detection, but of protection.

Vaccines Mycoplasma hyopneumoniae vaccines offer some protection against disease but they fail to eliminate the organism. The resulting low level of colonization has important ramifications for total herd health, particularly in relation to other infectious agents (87, 125, 126). Immunizing with whole cell lysates has been shown to provide protection against disease (30, 31, 69, 102). However, the treatment of these cells and the adjuvants used in the vaccine preparation are important in providing protection. For instance, immunization with cells that had undergone a freeze-thaw preparation did not provide sufficient immunity and in some cases increased lesion size (101). Subunit vaccines have not proven more effective against challenge than whole cell bacterins (38, 60), but given the complexity of the membrane 7 surface and the number of proteins exposed to the host immune system, perhaps the question that needs answering is which proteins are protective and how do you present them to the host in the most effective manner? Studies have been conducted on the detection of unique proteins and individual genes for developing more efficacious vaccines. Surface proteins are often chosen as potential vaccine targets because of their exposure to antibodies and because of their potential roles in virulence. Heat shock proteins have also been investigated for their immunogenicity. Several of these proteins have been identified by researchers as potential immune modulators in the hopes of developing a more protective vaccine. Species-specific monoclonal antibodies have been isolated for proteins P46 and P65 (19). P42 is another candidate protein used as a potential DNA vaccine. Studies have shown that antibodies to P42 can block growth of M hyopneumoniae (22). A variety of approaches have been used to enhance vaccine effectiveness including oral immunization with a recombinant Salmonella (39) or Erysipelothrix vaccine (113), DNA (molecular) vaccines (23), microencapsulation (67, 68), and fusions of M. hyopneumoniae proteins with immune stimulating sequences (20). Single protein vaccines, however, rarely protect against bacterial infections unless a toxin is involved. Additionally, antibody responses to M. hyopneumoniae in infected swine are not indicative of protection (127). One approach to vaccine development, expression library immunization (ELI), can provide a means of identifying sets of immunogenic proteins that can provide protection. Following construction of a random chromosomal or cDNA library, a preselection of clones with fragments identified by a poly-His tag reduces the total number of cDNA clones used for immunization, and thus the total number of animals needed to evaluate for vaccine efficacy (81). Several problems arise when screening mycoplasma libraries, however. Mycoplasma genes contain UGA codons that code for tryptophan; in other organisms it is a stop codon. This unusual codon usage prevents identification of many potentially protective antigens by routine genetic screens because of the failure to properly express in heterologous genetic systems. Also, the antibody screen requires only that a small peptide be produced as a fusion to the poly-histidine sequence. It may not represent an in-frame sequence or one in the proper orientation. Nevertheless, this technique in combination with other technologies will probably play a role in identifying protective antigens of M. hyopneumoniae.

Virulence mechanisms Little has been done to elucidate the mechanisms of disease in M. hyopneumoniae. The most significant progress has been in the area of adherence to host tissues. Mycoplasma 8 hyopneumoniae binds exclusively to the cilia of the epithelial cells lining the respiratory tract of pigs (122). Adherence to respiratory epithelium is required for infection and disease (26, 27, 144). Studies in the mid 1990s by Richard Ross and colleagues led to the identification and analysis of the cilium adhesin (141-143). Development of a cilium-binding assay led to the identification of monoclonal antibodies (Mabs) that blocked adherence of M. hyopneumoniae (143), and immunoblot analysis using those Mabs then led to the identification of the major adhesin, P97 (142). The Mabs were pivotal in cloning the gene for P97 (48) and the surrounding chromosomal region (50). Further studies involved the cloning and expression of functional P97, the first time that a mycoplasma adhesin had been cloned and expressed in a heterologous host in a functional form (48). The P97 protein contains two repeat regions near the carboxy terminus, R1 and R2. Additional analyses proved unequivocally that the R1 region of P97 was responsible for cilium binding and was the recognition site for adherence-blocking Mabs (49, 75). The function of the R2 region remains unknown. The P97 protein has several unusual features. Unlike the attachment proteins found in M. pneumonia and other mycoplasmas, the M. hyopneumoniae adhesin protein P97 does not directly attach to the mycoplasmal membrane (142). In all immunogold electron micrographs developed using the adherence-blocking Mabs, P97 appears to be attached to the membrane surface by thin fibers (29). This is in stark contrast to similar studies with PI, the adhesin of M. pneumoniae (43). The P97 protein is also posttranslationally cleaved, an unusual and provocative observation for mycoplasmas (29). None of the cleavage fragments have motifs suitable for interacting with membrane surfaces, suggesting that the P97 fragments must be interacting with other proteins to maintain their association with the mycoplasmal cell surface (29). More recent unpublished studies have shown that mycoplasma proteins and proteolytic fragments can bind host extracellular matrix proteins (T. Burnett, K. Dinkla, M. Rohde, G. S. Chhatwal, S. Cordwell, S. Geary, F. C. Minion, M. Walker, and S. Djordjevic. Abstr. 15th Congr. Int. Org. Mycoplasmol. abstr. 27, 2004; J. Wilton, K. K Stewart, F. C. Minion, P. Young, A. Collins, M. F. Walker, and S. P. Djordjevic. Abstr. 15th Congr. Int. Org. Mycoplasmol. abstr. 257, 2004) suggesting that these interactions serve an important role in pathogenesis. Finally, the recent completion of the M. hyopneumoniae genome sequence (80) revealed several paralogs of P97 lacking the R1 region suggesting that other regions of P97 have been duplicated. These paralogs are unusual in that they are gene fusions with non-homologous sequences and have been maintained within the genome suggesting that these gene sequences may be important to survival of the organism. 9

The interactions of M. hyopneumoniae with respiratory epithelium results in loss of cilia from the trachea, bronchi and bronchioles (72). Similar effects are seen in tracheal ring cultures (27). The mechanism by which M. hyopneumoniae can cause ciliostasis and cell death is unknown, but there is evidence that calcium release is involved (28, 92). Interesting, only strains that could bind cilia could induce calcium rises within the cell suggesting that the phenomenon is associated with a protein on the mycoplasma surface or associated with the cell (92).

Microarrays

History All groundwork for microarrays began in the 1960's with the fixation of DNA to nitrocellulose filters followed by detection with radiolabeled probes. In 1975, Southern expanded the technology to include separation of specific fragments prior to fixation to filters (119). In these experiments, the unknown nucleic acid sequences were transferred to membranes and the known labeled sequence or probe was allowed to interact. By doing so, Southern was able to detect specific sequences in complex populations of DNA digested with restriction enzymes. The technology is based upon the ability of single stranded DNA to hybridize to regions of homology forming stable hydrogen-bonded double-stranded molecules. Microarrays are based upon the same technology, but only in a much smaller footprint and in an opposite format. The known sequences are deposited on a solid substrate and the unknown nucleic acids (i.e., mRNAs, cDNAs) are labeled and interacted with the bound sequences. In contrast to Southern technology, however, microarrays offer a massively parallel gene expression and gene discovery approach to complex biological problems. One experiment can provide information on thousands of genes simultaneously. When cDNAs are used as targets, the level of signal intensity can be directly correlated with the steady state levels of mRNAs in the experimental system. Alternatively, DNA has been used to interrogate microarrays to identify non-homologous sequences. In these cases, fluorescence differences indicate sequence variation (13). The technology has been incorporated into the study and research of all types of organisms including viruses, bacteria, yeast, plants, animals and humans. The use of microarrays in molecular research is relatively new, being first described by P. Ekins and colleagues in the late 1980s (35). The principles on which microarrays are based have as their foundation in ligand assays relying on the binding of target molecules to a specific recognition reagent (37). The concept of miniaturized, multianalyte "microspot" 10 assays was first described by Ekins and coworkers (36), but it was not until 1995 when experiments using multiple genes in a miniature format was first reported in a study of 1,000 genes from Arabidopsis (108). The first genome-wide array was reported shortly after in 1997 on yeast (66). The pace of reports using microarrays has been accelerating over the past decade with over 3,300 articles that have used the Affymetrix platform for transcriptional analysis. Affymetrix has been producing commercially available microarrays called GeneChips® since 1994 (www.affymetrix. co m). Their technology revolves around the use of short oligonucleotides bound to glass slides. The oligonucleotides are produced in situ on glass substrates using photolithography as opposed to spotting presynthesized oligonucleotides using a pin based or ink jet style robot. The use of short oligonucleotides in microarrays presents some specific problems in terms of gene coverage and cross reactivity between genes, but computer algorithms have been developed to account for these adverse reactions. Construction of microarrays requires genome or cDNA sequence information. Early microarrays were based on cDNA sequences, but 251 genomes have been completely sequenced and another 1,087 are in the process (www.genomesonline.org). With the advent of whole genome sequencing, microarrays offer a low cost global approach to comparing within species variation, as well as between species variation either at the genomic or transcriptional level.

Types of arrays Microarrays can be divided into two basic types: oligonucleotide or cDNA. Oligonucleotide arrays can be constructed by photolithography in situ or they can consist of spotted preformed oligonucleotides. The former is a high-density microarray that is constructed using a combination of photolithography and combinatorial chemistry; tens of thousands of probes can be arranged on a single array (2, 85). Construction of such arrays begins with amine-modified glass substrates, which are further modified to contain the reactive group methynitropiperonyloxycarbonyl (MeNPOC)(106). This group is stable to a wide variety of chemical reagents, but it can be activated with ultraviolet light and thus is a photo-protecting group. Ultraviolet light removes the MeNPOC modification and allows the deprotected region to react with reactive phosphoramidite modified deoxyribose. By repeating the deprotection and coupling steps, any sequence of bases can be constructed on the glass surface. Photo masks allow the deprotection to occur at specific regions on the microarray surface. These masks can be manufactured in any configuration to allow the synthesis of DNA base pair at any location in the array. Affymetrix is printing arrays with 11 over 250,000 features (spots). The advantage of such arrays is that only four nucleotide building blocks are needed to construct the array, significantly reducing storage needs. The main disadvantage is that only short (<30 nucleotides) oligonucleotides can be reliably constructed using this technology (106). Because the oligonucleotides are relatively short, each probe is designed with a perfect match and a mismatch sequence to allow for a semi­ quantitative means of measuring each target. Multiple probes are designed to represent different regions in the gene or target sequence that has been identified from expressed sequence tag (EST) libraries or DNA sequence data. Algorithms help to identify probes best suited to identify all possible transcripts (www.affVmetrix.com). Maskless array synthesizer technology is also now available. This technology uses mirrors to direct an ultraviolet light beam to any location on the array (115). These mirrors are linked to computer memory substrates so that deflection of the mirror can be achieved by accessing individual memory bits. Microarray densities of-390,000 features/cm2 have been achieved using this technology (115). Spotted arrays are DNA sequences that represent individual genes, printed on treated glass substrates. The treatments of these substrates are to enhance DNA binding and include poly-L-lysine coating and glass modifying treatments such as gamma aminopropylsilane, aldehyde and epoxy 3-glycidoxypropyltrimethoxysilane epoxide. Additional coatings are now available to support the construction of protein arrays or other modified macromolecules (avidin, streptavidin, hydrophobic polymers). The amine-modified substrates are best used with long oligonucleotides and PCR products while the epoxy-modified substrates can accomodate shorter (~30-mers) oligonucleotides in addition to cDNA and longer oligonucleotides. The aldehyde-modified substrates can also be used with proteins. Variation in glass quality and flatness, autofluorescence due to glass purity, coatings or treatments, efficiency of probe binding resulting from treatment type, uniformity of coating or treatment, etc., are only some of the many parameters that must be stringently controlled during the manufacturing process of the substrates. The spotted DNA, referred to as probes, represent known DNA sequences. The sequence information needed to produce the specific probes can be obtained from genome sequencing projects, expressed sequence tag projects, or could be individual cDNA clones spotted as plasmids. The latter could be sequenced later once specific targets have been identified through array analysis. These probes can be PCR amplified sequences from whole genomes generally ranging in size from 200-500 bp, cDNA libraries in plasmids or individual oligonucleotides approximately 50-70 bp in length. The spotting process requires motion- control systems, robots, computer software, and deposition devices (106). Deposition can 12 occur using solid pins, split pins, capillary tubes, pin and ring devices, and non-contact piezoelectric technologies (ink jet technology). Each has its own advantages and disadvantages; split pins in a contact robot were used in the construction of the arrays used in these studies. The spotting process, like other aspects of microarray technology, has numerous elements that must be carefully controlled. Elements such as pin design, pin arrangement, pin cleaning efficiency, robot speed, pin contact speed, contact time, humidity during printing, etc. all impact the quality of the arrays (106). In addition, there is the issue of probe generation for microarray construction. This process must be carefully controlled as well, taking into consideration things like PCR primer design (for oligonucleotide arrays, oligonucleotide design is the single most important factor), PCR conditions, PCR product purification and purity, product concentration, and spotting buffer composition. Once spotted, arrays are allowed to dry in place, heated under vacuum to remove all water molecules and maximize chemical cross-linking, and then the arrays are cross-linked using ultraviolet light to maximize the retention of the probes to the glass surface during the subsequent hybridization and washing steps. A more detailed analysis of microarray construction can be obtained from Schena (106), which describes the dozens of variables involved in array construction from glass substrates and their modification, probe design, manufacturing technologies, detection systems, and informatics. The importance of the spotting buffer is often overlooked, but it is one of the most important factors in array construction because of the small volumes involved and the effect that evaporation plays in the spotting process. Evaporation occurs in two places, in the plate that contains the individual probes or oligonucleotides and in the pins used to transfer the probes from the plates to the glass surface. The former can result in increasing concentration of probe deposition during the spotting cycle and the latter can result in loss of probe at the end of the printing cycle. Too rapid evaporation of the substrate following spotting can also result in non-uniform spot morphology causing problems in data analysis. Spotting buffers that significantly reduce evaporation while allowing proper interaction of the probe with the modified glass surface are desired such as the spotting buffer produced by Coming (Coming, Inc.). Buffers like Tris-ethylenediamine tetraacetic acid are subject to rapid evaporation and can be problematic. Buffers that contain dimethyl sulfoxide (50%) are less subject to evaporation, but can cause DNA precipitation when plates containing the probes are frozen and thawed or if the concentration of dimethyl sulfoxide reaches 70% or more as a result of evaporation. These precipitates cannot be solubilized in an effective manner, and thus the probes are lost and must be produced again. 13

Once spotted, these microarrays can be hybridized with fluorescently labeled "targets", such as fragmented genomic DNA, cDNA, or RNA. They represent the unknown sequences or sequence quantities of your sample and are the elements of the study under investigation. Ideally, each labeled target is represented in the population in the same relative ratio or concentration as in the sample. A variety of labeling techniques are now available, which can be defined as direct and indirect labeling methods. For transcriptional studies, cDNA is generated from RNA preparations using reverse transcriptase and random primers, oligo-dT primers (to selectively identify eukaryotic mRNAs), or gene specific primers (used primarily in bacterial systems or where low numbers of targets are being interrogated by the array). The cDNA can then be labeled directly or indirectly. Direct labeling schemes incorporate the fluorescent tags directly into the probes during the reverse transcriptase or PCR reaction. This can occur either by using prelabeled primers or labeled nucleotides. The former results in probes of low signal strength and the latter are differentially incorporated into the product because of the large size differences of the fluorescent tags (i.e., Cy3 and Cy5) and their interference with the polymerization reaction. Consequently, dye "swap" experiments, where the labeled dyes are swapped with respect to the mRNA sources in two different experiments, are critical to minimize error associated with dye bias. Newer dyes such as the Alexa series from Molecular Probes are more similar in size and structure and present fewer problems during incorporation. These dyes also have higher fluorescence and photostability than the cyanine dye derivatives and are more suitable for microarray experiments where signal intensities are critical (see http://www.probes.invitrogen.com/handbook for more details on the structure and performance of the Alexa dyes). The indirect labeling schemes can be divided into two variations. Amine modified nucleotides can be incorporated into the DNA product and then reacted with fluorescent dyes. The amine modification has little effect on the polymerization reaction and such modified nucleotides are incorporated at high efficiency. The tag coupling reaction is simple and efficient resulting in strong signals. There is less need for dye swap experiments using this indirect method for labeling targets, but this should always be confirmed. In lieu of conjugating the fluorescent tag directly to the target, biotin can be incorporated during the target generation reaction and a streptavidin-phycoerythrin conjugate can be used to generate signals (107, 130). This approach is suitable only for one-color platforms such as Affymetrix. A second approach called Tyramide Signal Amplification (TSA™) utilizes biotin and dinitrophenol (DNP) or fluorescein epitopes as well as streptavidin and antibody conjugates linked to horseradish peroxidase (l)(Perkin Elmer, Boston, Mass.). The technology works on the principle of catalyzed reporter deposition where horseradish peroxidase oxidizes the 14 phenolic ring of fluorescently labeled tyramide conjugates to create reactive intermediates that subsequently bind to the biotin molecule. Two-color detection is accomplished using matched hapten/cyanine-tyramide pairs and sequential detection (130). This technology is reported to increase sensitivity 50-100 fold over other labeling methods. Bacterial systems present unique problems in fluorescent target generation because of the difficulty in identifying bacterial mRNA during the labeling reaction; they have no poly- adenylated tails that can be used to purify or selectively label the mRNA molecules. Thus, either total RNA must be labeled (mRNA represents only -5% of the sample) reducing the signal intensities significantly on the arrays, the ribosomal RNAs must be removed from the total RNA preparations, or gene-specific primers must be used to generate the labeled targets. Procedures and reagents are now available for removing the ribosomal RNAs from many prokaryotic total RNA preparations, but they are not useful for mycoplasmas (Ambion, Austin, TX). The microarray hybridization parameters such as buffer composition, pH, salt concentrations, and temperature are generally determined empirically. This is important particularly for arrays that represent probes from organisms with skewed G+C mol% content such as streptococci (high G+C) and mycoplasmas (low G+C). Betaine, dimethyl sulfoxide or glycerol is sometimes added to enhance hybridization specificity and efficiency. Following target labeling and hybridization, the fluorescent intensities of each spot should accurately represent the concentration of each probe in the sample. The fluorescent intensities from each spot can then be analyzed for statistical significance of genomic or transcriptional differences. The spotted arrays are usually hybridized with two or more labeled targets so that control and experimental samples can be analyzed on the same spot and on the same slide thus allowing background subtraction and normalization to be performed for both control and experimental data from the same array.

Informatics The development of microarray technology has brought needs for additional data management and analyses. Software development alone has been extensive and includes primer design, image acquisition and quantitation, database design and statistical analysis. Primer design software must produce primers that are unique and perform well depending on the type of array under construction. For long oligonucleotide-based arrays, the oligonucleotides must be specific and not allow cross-reactions with more than a single transcript. For the generation of PCR products for array construction, primers must bind to a single chromosomal region and must generate PCR products of a preselected size at high 15 efficiency. Thus, each primer must be tested against all other chromosomal sequences to ensure specificity. This problem increases exponentially as the number of open reading frames included in the array increases. Other problems may include limited sequence information, sources of samples and availability, and the restrictive costs of large experiments. Finally, the experimental design must take into account the variables within each of the subject areas above. Image analysis may seem straightforward, but several features of microarrays present problems. The processing of a scanned image requires three separate tasks: addressing or gridding, segmentation, and intensity extraction (106). Addressing is assigning the coordinates to each of the spots on the image. This is usually accomplished using a grid or spot array and displacing those to positions that overlap the spots on the image. Different programs vary in the way this is accomplished, but in all cases, small variations within the array symmetry require some manual input to the process. The goal here is to perform this aspect in an automated fashion with minimal human input to allow for high throughput. Segmentation classifies the pixels as foreground (within the spot) or background (outside the spot). This can be accomplished in various ways, including fixed circle, adaptive circle, adaptive shape, and by histogram. The fixed circle is usually the standard option in commercial software and is the easiest to implement. The adaptive circle method relies on individual sizing of the circles as a function of the spot size. This can be done automatically or it can be done manually (14). For large arrays, this can be time consuming. Adaptive shape segmentation might be the most accurate, but the method requires starting points from which the process grows. In most cases this can be accomplished from array images, but deviations in array structure do occur. This approach has the advantage of being able to cope with spot shapes that are not symmetrical. Histogram segmentation uses a target mask that is larger than the spots to analyze and from which the pixel values are calculated. Foreground and background values are estimated from these values as a predetermined percentile such as 5th and 20th percentile for background and the 80th and 95th percentile for foreground. Several implementations of this approach have been used including one that used the Mann-Whitney test to compute threshold values (21). Intensity extraction involves both summing the pixel intensities within the spot mask and the calculation of local background intensities that arise from nonspecific hybridization and fluorescence emitted from other chemicals on the glass. The process of background estimation can use several different approaches including those that measure each spot individually and those that use a global background for all spots. 16

Presentation and analysis of microarray data are rapidly evolving. Early studies presented false color images (109, 132), scatter plots (45), and cluster diagrams (34). These methods of analysis give an overview of the data, but do not provide critical evaluation of the data or its significance. To accurately test and evaluate data of such magnitude and complex design as significant, statistical methods needed to be adapted to answer the questions being asked using microarray technology. Microarray data is not normally distributed and therefore it is necessary to normalize the data. Normalization adjusts for systematic differences in the relative intensity of each signal channel, and corrects for differences in intensities within the same slide or between slides that are array dependent and not due to true biological variation (106). A number of approaches have been defined for normalization and are discussed by Schena (106). Usually a set of genes is identified for normalization, such as housekeeping genes, that are not expected to differ within the experiment. This rarely occurs so some investigators have used spiked controls. Several companies (i.e., Stratagene) sell kits that contain spike mixes and housekeeping genes for validation and normalization. An approach that is gaining popularity is to use all of the genes on the array with the assumptions that only a relatively few of the genes will vary significantly in expression, and that there is symmetry in the expression of up-regulated and down-regulated genes. Many features of array construction and design can help to ensure the quality of the array and hence the value of statistics resulting from an experimental study. Increasing the number of biological replicates is favored over increasing technical replicates from limited biological samples (25). Normalizing data compensates for the variation introduced by the technical aspects, such as unequal fluorescent dye incorporation and the array-to-array variation, but not variation due to biological differences (117). Normalization in two-color systems is almost always necessary due to the varied incorporation rates and stability of the fluorescent dyes. Normalizing data is completed before statistical analysis. A two-color array design is powerful because it has the ability to directly compare two samples (25). However, the degree of freedom is reduced because the two-color factor is included in the statistical model. Additional factors included in the model are the arrays, experimental treatments, genes, array by gene interaction, and the most interesting effect being that of treatment by gene (56). Kerr et al. describe the ease with which data is normalized and then analyzed to reflect the significant effects of treatment upon a specific gene (56). Using a logarithmic scale to transform the raw data and not the log ratio values also allows for ease of representing the data as a fold-change in gene expression with statistical significance associated with that change in expression. Relying strictly on fold- 17 change values results in numerous false positives and false negatives (123). Tanaka et al. found that differentially expressed genes in the 1.2 to 2-fold change can often be overlooked as non-significant, when they have actual biological and statistical significance in replicated data sets (123). A mixed-model includes factors that are fixed and random. In typical microarray analyses, factors considered fixed are the treatment effect and dye effect. Factors usually considered random are slides and slide-by-area. Generally, random factors account for variability so that differences can be determined for the factors of interest (i.e., fixed effects). Examples of variability are spatial variation caused by either defects on the slide surface or non-uniform coating of the slide surface, and dye incorporation rates. The probability of finding a gene statistically significant when in reality it is not significant (i.e., type I error) is substantially increased when multiple comparisons are made with data from a single experiment. Traditional analyses rely on methods like Bonferroni, Scheffe, and Tukey to control type I error rate (89). Storey et al. explains a method to determine type I errors in microarray analysis, the false discovery rate (FDR)(120). Determining the FDR helps researchers identify genes that are not transcriptionally differentiated even though a /-test indicates otherwise. False discovery rates are determined by calculating a q-value. The q-value is similar to a p-value, indicates the probability of committing a type I error and establishes a level of confidence in determining statistical significance. False discovery rate is used to control for experiment-wise error.

Applications Microarrays are used to elicit information about genetic or transcription differences represented on the array. For example, genetic variation in clinical or field isolates of pathogens, transcriptional variation and its effects in tumorgenesis, effects of stress on transcriptional variation in pathogens are all suitable topics for investigation using microarrays. Although too many to list, the number of publications utilizing microarray technology is growing rapidly and the possible experiments seem limitless.

Genomic comparisons (within/without species) Through examination of the genetic content of bacteria, variations often indicate the emergence of new strains and adaptations to new environmental niches. Genetic comparisons allow the medical field to track how bacterial species may acquire genes, increasing the occurrence of infections in human populations (124, 134). Additional studies outline the similarity of laboratory and clinical strains giving validity to in vitro assays using laboratory 18 strains (118). Snyder and colleagues also indicate that multi-strain, multi-species microarrays would truly make them a more informative tool compared to single genome arrays (118). With the heightened awareness of potential bioterrorism threats, microarrays are being developed to identify infectious agents. This progress is slow and only a limited number of arrays are currently being reported. Most arrays currently in use require some knowledge regarding the sequence of the organism. Belosludtsev et al. have designed an array to detect Bacillus subtilis, Yersinia pestis, Streptococcus pneumoniae, Bacillus anthracis and Homo sapiens using a universal probe set tested against the various species (6). Their results indicate arrays can be designed for species not yet sequenced and accurately identify unknown samples. The ability to identify various bacteria from mixed samples is relevant for environmental samples. Arrays fabricated by Kim et al. using random probes also distinguished individual species from mixed populations (57). As the need for surveillance of the environment continues, more random probe microarrays will continue to be developed to by-pass the need for sequencing entire genomes.

Single nucleotide polymorphism detection Genetic variation at the nucleotide level is readily amenable to microarray technology and can be used in several applications of patient health. Much of the single nucleotide polymorphism (SNP) studies focus on defining the natural variation present in a population versus mutations occurring at those loci. By detecting SNPs associated with disease, researchers are able to determine the predisposition to infection, cancer and other disorders. Genotypes have been previously associated with psychotropic medication treatment outcome (82). By determining the genetic sequence of a particular locus, doctors are able to better predict which medications to prescribe and their predicted outcomes. A course of treatment can then be tailored to a specific patient based on their genetic make-up. Once accurate probe detection is available, any subsequent genetic screening, especially for highly heritable traits in offspring, can be established (112). Epidemiological studies often provide data regarding disease states. Examining SNPs associated with a particular cancer and comparing that to normal tissue can identify unstable regions in the genome (47). Due to the loss of heterozygosity often associated with cancer and various diseases, current research is focusing on validating SNP microarrays using current gold standards (137). This is an ever-expanding field of study. As more SNPs are validated for detection of disease and therapeutic treatment of those diseases are identified, SNP microarrays will become the standard tool for doctors and researchers to examine the course of treatment. One area of research that is utilizing genomic arrays is that of breast 19 cancer research. Researchers are using arrays to compare genetic make-up of the tumors for the purposes of better classifying and treating them (84). Once multiple tumor types have been classified genetically, insight can be gained from the genetic component of oncogenesis, hopefully developing novel and effective preventatives and therapeutics. Forensic science also wishes to exploit SNP technology to identify unknown persons, suspects and/or victims in a particular criminal case. Research is ongoing to optimize current databases, and improve those regions of detection. Validation of SNP technology is ongoing in area of mitochondrial DNA sequencing and identification of suitable probes. Mitochondrial DNA is often an ideal choice for analysis given the degradation that occurs in many forensic samples, especially those of remains cases. Robust databases must be created to allow for greater comparisons of unknown samples and identification of individuals. NanoChip arrays have been developed to allow for minimal sample, increase throughput and sensitivity when analyzing forensic samples (8). This system will require validation before it can withstand the rigorous examination in the courtroom.

Gene expression Expression profiles for numerous bacteria have now been constructed using microarrays. Experiments designed to examine differences due to strain variation and environmental factors such as temperature, antibiotics, nutrient requirements and host effects, frequent the list of current publications (44, 116).

Pathogens In the realm of pathogenic bacterial microarrays, the studies seem to be limitless. Microarrays are used as a means of surveillance for the presence of bacteria, for identification of varying serovars, to examine host-pathogen interactions and to assess genetic contributions to pathogenicity (86, 90, 93, 94). Using microarrays, researchers have been able to study in vivo host-pathogen interactions and not just in vitro laboratory models. Historical samples can also be used, giving an indication of genetic variation during disease progression. Epidemiological studies can tract changes in an organism over both time and space. As the microarray data is gathered and analyzed, a more complete picture of the mechanisms of bacterial pathogenesis will be developed.

Food safety One area of research in which microarrays may prove to be beneficial is food-borne pathogenesis. Current research in the area has targeted Listeria monocytogenes, Escherichia coli 0157:H7, Campylobacter jejuni, and Salmonella enterica serovar Enteritidis, as well as 20 other pathogens associated with the preparation, distribution and storage of food. Microarrays can be used for pre- and post-harvest diagnostics, for epidemiological studies into sources of outbreaks and for studies into the pathogenesis of the diseases. Listeria monocytogenes microarrays have been able to identify the differences between serotypes. Genomic hybridizations were able to identify divergent lineages and screen for genetic markers in clinical samples (15). Additional research using a mixed genome microarray revealed data similar in resolution to pulsed field gel electrophoresis for epidemiologically linked strains (9). Research studies using C. jejuni microarrays have included environmental detection systems, genome mapping, and determining the effects of nutrients on the physiology of the bacteria. Global mapping studies allow for a selection of markers that can identify genetic variation in strains of C. jejuni and eventually may be used in genotyping methods (121) by analyzing physiological variation, mechanisms of survival and the physiological role particular elements play in pathogenesis. Microarray studies examining the functional groups associated with iron acquisition by C. jejuni reveals many of the same pathways are required for colonization of the gastrointestinal tract (121). Chickens are often the source for intestinal disorders due to C. jejuni, and as such, a focus for many detection studies and protocols. Danish researchers were able to use microarrays to expedite analyses of contamination levels in broiler chickens (55). In this study, the researchers were able to detect C. jejuni at the species level in all but 6% of the samples tested compared to 19% in PCR testing. The inherent specificity, low sample volume and lack of a need to culture samples make microarrays an efficient means of evaluating the presence of C. jejuni in food sources. Escherichia coli 0157:H7 microarray research has also focused its efforts on identifying the genes relevant to pathogenesis and to distinguish between pathogenic and non-pathogenic strains (138). As an emerging infectious agent having been identified in the 1980s, studies are now focused on identifying the etiology of the organism (134). By designing arrays with unique genes related to pathogenesis, the arrays can then be used to identify E. coli strains in clinical and environmental samples. Salmonella enterica serovar Enteritidis studies currently look at the spread of disease and identification of particular serovars in relation to outbreaks and prevalence (3, 135). By identifying regions of variation at the genetic level, new classifications can be made that determine pathogenicity and host range (94). The authors also coined the term "genovars" to differentiate variation at the genetic level between common serovars. Genomic analysis of several serovars of Typhimurium determined that intrastrain divergence is limited, but interstrain divergence exists and may play a role in virulence (100). In a study by Prouty et 21 al. the role of bile salts contributed to transcriptional regulation and probably pathogenesis as well (95). The array results confirmed the down-regulation of virulence and flagellular biosynthesis genes. The role of "swarming" has also been investigated using microarrays indicating that many of the genes associated with the type III secretion system, lipopolysaccharide synthesis and iron regulation are differentially transcribed compared to cells grown in culture (129). These findings give additional information and focus to studies of how serovar Typhimurium is able to colonize the host intestinal tract. Microarrays have traditionally examined the genetic make-up and overall transcription levels of a single organism. Another approach is to combine several gene-specific probes on a single array to screen for many types of contaminants. Sergeev et al. were able to detect Staphylococcus aureus enterotoxin genes, Listeria spp., Campylobacter spp. and Clostridium perfringens toxin genes using microarrays (111). This array combined the ability to detect low numbers of organisms in a given sample with the ability to screen for several different pathogens in one test. An area of research currently lacking in microarray research of food pathogens is the pre- and post-harvest analysis of foodstuffs. With the sequencing of many species of bacteria, sets of unique genes capable of identifying low levels of contaminants should be possible. This would be a significant addition to the monitoring process and protection of our food supply.

Mycoplasmas Seemingly overlooked in the realm of microarray studies are the mycoplasmas. With their endemic existence, broad host base, and small genomes, mycoplasmas should be actively studied using this technology. To date, only three reports have applied arrays in their studies. Using a high-density tissue macroarray, Chinese researchers examined the prevalence of M. hyorhinis in gastric tissue samples compared to non-cancerous tissues (140). Diagnostic analysis using an oligonucleotide array to various respiratory infections included M. hyorhinis (103). The final study examined the transcriptional response of M. pneumoniae to heat shock conditions (131). Herrmann and co-workers were able to identify transcriptional differences similar to the results from other bacterial heat shock studies. The minimal amount of microarray data for mycoplasmas is indicative of need for additional studies. 22

Summary with Goals and Hypothesis As described above, M. hyopneumoniae presents several problems when attempting to use classic molecular biology tools to study pathogenesis. This is true for mycoplasmas in general, although there are some exceptions where genetic tools are available (17, 24, 32, 33, 58, 59, 62-64, 78, 110). Consequently, progress in understanding the genetic basis of pathogenesis for M. hyopneumoniae has been slow, and little information is currently available regarding the molecular mechanisms of pathogenesis. The recent sequencing of the genome (Appendix A) has provided the information necessary to initiate new lines of investigation that have been heretofore closed to mycoplasmologists. Using this genome sequence information, we were able to construct microarrays to initiate studies of both genetic and transcriptional differences under changing environmental conditions. Since there has only been a single report of a global transcriptional study in mycoplasmas (131), this study offers new insight into mycoplasma gene regulation. In addition, the information gained from these studies may add insight as to the functions for hypothetical genes, and thus, targets for new, effective vaccines may be revealed. The overall hypothesis of this study is that variation occurs in M. hyopneumoniae that impacts its response to environmental changes, ability to adhere to swine cilia, and cause disease. Genetic variation may also be contributing to these factors. Therefore, the goal of this study was to assess the ability of a genome-wide microarray to measure changes at the genomic and transcriptional levels and reveal new information regarding mycoplasma gene regulation and genome variability. The first set of experiments reported in Chapters 2 and 3 focus on analyzing transcriptional differences during two stress-related conditions, heat shock and iron deprivation. These two stresses have been shown to impact an organism's ability to survive in vivo during disease. By providing insight into how M hyopneumoniae responds to stress in vitro, these studies further our understanding of fundamental genetic processes that seem common to all bacteria but only now are being revealed in mycoplasmas. These studies also impact our understanding of the ability of mycoplasmas to regulate transcription in response to environmental changes. Chapter 4 compares virulent strain 232 with laboratory-adapted virulent strain J. This study was undertaken to define differences in the transcription profiles between a virulent strain and an avirulent strain with the intent of identifying changes that may be related to pathogenesis. The studies reported in Chapter 5 examine transcriptional differences in cilium adherence variants obtained from M. hyopneumoniae strain 232. Transcriptional analysis of a low 23 adherent 60-3 variant and high adherent 91-3 variant should give insight into swine cilium adherence-related gene expression differences and possibly identify genes involved in the adherence process in addition to P97, the cilium adhesin. Since adherence is required to cause disease, these data should provide knowledge of the genes required by M hyopneumoniae in vivo. Chapter 6 reports the genetic differences in field isolates as revealed by this microarray. Since these isolates came from field cases they are considered to be pathogenic, but the extent of virulence has been assessed in only a few of the strains using a pig challenge model (Strait et al., unpublished). These data could also be used to define genomic regions suitable for development of better diagnostic reagents for M. hyopneumoniae. One of the drawbacks of this microarray is that only known genes in strain 232 are represented. Any gene acquisition by the field isolates would not be detected. Since mycoplasmas are thought to have evolved by degenerative evolution and are thought to be a model for the minimalist genome, gene acquisition should be limited. Additionally, there is no evidence that strain 232 had acquired any gene sequences in the recent past since the GC skew in the genome was unremarkable (80). This dissertation ends with the General Conclusions and Future Studies. Microarrays are often thought of as hypotheses generating tools in addition to their ability to measure variation in a massively parallel format. Unexpected results are common because we do not know how all of the pieces fit together in living systems to produce defined phenotypes. Microarrays provide not only the tools to examine change, but they also do so in accelerated fashion. These studies will enhance our understanding of basic mechanisms employed by M. hyopneumoniae and by mycoplasmas in general in unanticipated ways, the true hallmark of scientific investigation.

References 1. Adler, K., J. Braodbent, R. Garlick, R. Joseph, A. Khimani, A. Mikulskis, P. Rapiejko, and J. Killian. 2000. MICROMAX™: a highly sensitive system for differential gene expression on microarrays. In M. Schena (ed.), Microarray Biochip Technology. Biotechnqiues Books, Natick, N.Y. 2. Aitman, T. J. 2001. DNA microarrays in medical practice. Br. Med. J. 323:611-615. 3. Alvarez, J., S. Porwollik, I. Laconcha, V. Gisakis, A. B. Vivanco, I. Gonzalez, S. Echenagusia, N. Zabala, F. Blackmer, M. McClelland, A. Rementeria, and J. Garaizar. 2003. Detection of a Salmonella enterica serovar California strain spreading 24

in spanish feed mills and genetic characterization with DNA microarrays. Appl. Environ. Microbiol. 69:7531-7534. 4. Artiushin, S., and F. C. Minon. 1996. Arbitrarily primed PCR analysis of Mycoplasma hyponeumoniae field isolates demonstrates genetic heterogeneity. Int. J. Sys. Bacteriol. 46:324-328. 5. Baseman, J. B., M. Lange, N. L. Criscimagna, J. A. Giron, and C. A. Thomas. 1995. Interplay between mycoplasmas and host target cells. Microbial Pathogenesis 19:105- 116. 6. Belosludtsev, Y. Y., D. Bowerman, R. Weil, N. Marthandan, R. Balog, K. Luebke, J. Lawson, S. A. Johnston, C. R. Lyons, K. Obrien, H. R. Garner, and T. F. Powdrill. 2004. Organism identification using a genome sequence-independent universal microarray probe set. Biotechniques 37:654-658, 660. 7. Bolske, G., L. Strandberg, K. Bergsrtrom, and K. Johansson. 1987. Species-specific antigen of Mycoplasma hyopneumoniae and cross-reaction with other porcine mycoplasma. Curr. Microbiol. 15:233-239. 8. Borsting, C., J. J. Sanchez, and N. Morling. 2005. SNP typing on the NanoChip electronic microarray. Methods Mol. Biol. 297:155-168. 9. Borucki, M. K., S. H. Kim, D. R. Call, S. C. Smole, and F. Pagotto. 2004. Selective discrimination of Listeria monocytogenes epidemic strains by a mixed-genome DNA microarray compared to discrimination by pulsed-field gel electrophoresis, ribotyping, and multilocus sequence typing. J. Clin. Microbiol. 42:5270-5276. 10. Brown, D. R., J. M. Farley, L. A. Zacher, J. M. Carlton, T. L. Clippinger, J. G. Tully, and M. B. Brown. 2001. Mycoplasma alligatoris sp. nov., from American alligators. Int. J. Sys. Evol. Microbiol. 51:419-424. 11. Brown, D. R., D. F. Talkington, W. L. Thacker, M. B. Brown, D. L. Dillehay, and J. G. Tully. 2001. Mycoplasma microti sp. nov., isolated from the respiratory tract of prairie voles (Microtus ochrogaster). Int. J. Sys. Evol. Microbiol. 51:409-412. 12. Brown, M. B., D. R. Brown, P. A. Klein, G. S. McLaughlin, I. M. Schumacher, E. R. Jacobson, H. P. Adams, and J. G. Tully. 2001. Mycoplasma agassizii sp. nov., isolated from the upper respiratory tract of the desert tortoise (Gopherus agassizii) and the gopher tortoise (Gopherus polyphemus). Int. J. Sys. Evol. Microbiol. 51:413- 418. 13. Brown, P. O., and D. Botstein. 1999. Exploring the new world of the genome with DNA microarrays. Nat. Gen. 21:33-37. 25

14. Buhler, J., T. Ideker, and D. Haynor. 2000. Dapple: improved techniques for finding spots on DNA microarrays, p. http://www.cs.wustl.edu/~jbuhler/research/dapple/dapple-tr.pdf. University of Washington, Seattle. 15. Call, D. R., M. K. Borucki, and T. E. Besser. 2003. Mixed-genome microarrays reveal multiple serotype and lineage-specific differences among strains of Listeria monocytogenes. J. Clin. Microbiol. 41:632-639. 16. Calsamiglia, M., J. E. Collins, and C. Pijoan. 2000. Correlation between the presence of enzootic pneumonia lesions and detection of Mycoplasma hyopneumoniae in bronchial swabs by PCR. Vet. Microbiol. 76:299-303. 17. Cao, J., P. A. Kapke, and F. C. Minion. 1994. Transformation of Mycoplasma gallisepticum with Tn916, Tn4001, and integrative plasmid vectors. J. Bacteriol. 176:4459-4462. 18. Chambaud, I., R. Heilig, S. Ferris, V. Barbe, D. Samson, F. Calisson, I. Moszer, K. Dybvig, H. Wroblewski, A. Viari, E. P. Rocha, and A. Blanchard. 2001. The complete genome sequence of the murine respiratory pathogen Mycoplasma pulmonis. Nuc. Acids Res. 29:2145-2153. 19. Cheikh Saad Bouh, K., F. Shareck, and S. Dea. 2003. Monoclonal antibodies to Escherichia co/z'-expressed P46 and P65 membranous proteins for specific immunodetection of Mycoplasma hyopneumoniae in lungs of infected pigs. Clin. Diagn. Lab. Immunol. 10:459-468. 20. Chen, J. R., C. W. Liao, S. J. Mao, and C. N. Weng. 2001. A recombinant chimera composed of repeat region RR1 of Mycoplasma hyopneumoniae adhesin with Pseudomonas exotoxin: in vivo evaluation of specific IgG response in mice and pigs. Vet. Microbiol. 80:347-357. 21. Chen, Y., E. R. Dougherty, and M. L. Bittner. 1997. Ratio-based decisions and the quantitative analysis of cDNA microarray images. J. Biomed. Optics 2:364-374. 22. Chen, Y. L., S. N. Wang, W. J. Yang, Y. J. Chen, H. H. Lin, and 0. Shiuan. 2003. Expression and immunogenicity of Mycoplasma hyopneumoniae heat shock protein antigen P42 by DNA vaccination. Infect Immun 71:1155-1160. 23. Chen, Y. L., S. N. Wang, W. J. Yang, Y. J. Chen, H. H. Lin, and D. Shiuan. 2003. Expression and immunogenicity of Mycoplasma hyopneumoniae heat shock protein antigen P42 by DNA vaccination. Infect. Immun. 71:1155-1160. 26

24. Chopra-Dewasthaly, R., M. Zimmermann, R. Rosengarten, and C. Citti. 2005. First steps towards the genetic manipulation of Mycoplasma agalactiae and Mycoplasma bovis using the transposon Tn4001mod. Int. J. Med. Microbiol. 294:447-453. 25. Churchill, G. A. 2002. Fundamentals of experimental design for cDNA microarrays. Nat. Genet. 32 Suppl:490-495. 26. Debey, M. C., and R. F. Ross. 1989. Ciliary and metabolic responses of porcine tracheal rings to Mycoplasma hyopneumoniae. Proc. Am. Soc. Microbiol. 89:138. 27. DeBey, M. C., and R. F. Ross. 1994. Ciliostasis and loss of cilia induced by Mycoplasma hyopneumoniae in porcine tracheal organ cultures. Infect. Immun. 62:5312-5318. 28. Debey, M. C., J. A. Roth, and R. F. Ross. 1993. Enhancement of the increase in intracellular calcium concentration in stimulated neutrophils by Mycoplasma hyopneumoniae. Vet. Res. Commun. 17:249-257. 29. Djordjevic, S. P., S. J. Cordwell, M. A. Djordjevic, J. Wilton, and F. C. Minion. 2004. Proteolytic processing of the Mycoplasma hyopneumoniae cilium adhesin. Infect. Immun. 72:2791-2802. 30. Djordjevic, S. P., G. J. Eamens, L. F. Romalis, P. J. Nicholls, V. Taylor, and J. Chin. 1997. Serum and mucosal antibody responses and protection in pigs vaccinated against Mycoplasma hyopneumoniae with vaccines containing denatured membrane antigen pool and adjuvant. Aust. Vet. J. 75:504-511. 31. Dohoo, I. R., and M. E. Montgomery. 1996. A field trial to evaluate a Mycoplasma hyopneumoniae vaccine: effects on lung lesions and growth rates in swine. Canadian Vet. J. 37:299-302. 32. Dybvig, K., and G. H. Cassell. 1987. Transposition of gram-positive transposon Tn916 in Acholeplasma laidlawii and Mycoplasma pulmonis. Science 235:1392- 1394. 33. Dybvig, K., C. T. French, and L. L. Voelker. 2000. Construction and use of derivatives of transposon Tn4001 that function in Mycoplasma pulmonis and Mycoplasma arthritidis. J. Bacteriol. 182:4343-4347. 34. Eisen, M. B., P. T. Spellman, P. O. Brown, and D. Botstein. 1998. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95:14863-14868. 35. Ekins, R., F. Chu, and E. Biggart. 1990. Fluorescence spectroscopy and its application to a new generation of high sensitivity, multi-microspot, multianalyte, immunoassay. Clinica Chimica Acta 194:91-114. 27

36. Ekins, R. P. 1998. Ligand assays: from electrophoresis to minaturized microarrays. Clin. Chem. 44:2015-2030. 37. Ekins, R. P., and F. Chu. 1994. Devloping multianalyte assays. Trends Biotechnol. 12:89-94. 38. Fagan, P. K., S. P. Djordjevic, G. J. Eamens, J. Chin, and M. J. Walker. 1996. Molecular characterization of a ribonucleotide reductase (nrdF) gene fragment of Mycoplasma hyopneumoniae and assessment of the recombinant product as an experimental vaccine for enzootic pneumonia. Infect. Immun. 64:1060-1064. 39. Fagan, P. K., M. J. Walker, J. Chin, G. J. Eamens, and S. P. Djordjevic. 2001. Oral immunization of swine with attenuated Salmonella typhimurium aroA SL3261 expressing a recombinant antigen of Mycoplasma hyopneumoniae (NrdF) primes the immune system for a NrdF specific secretory IgA response in the lungs. Microb. Path. 30:101-110. 40. Fraser, C. M., J. D. Gocayne, O. White, M. D. Adams, R. A. Clayton, R. D. Fleischmann, C. J. Bult, A. R. Kerlavage, G. Sutton, J. M. Kelley, J. L. Fritchman, J. F. Weidman, K. V. Small, M. Sandusky, J. Fuhrmann, D. Nguyen, T. R. Utterback, D. M. Saudek, C. A. Phillips, J. M. Merrick, J.-F. Tomb, B. A. Dougherty, K. A. Bott, P.-C. Hu, T. S. Lucier, S. N. Peterson, H. O. Smith, C. A. Hutchison, and J. C. Venter. 1995. The minimal gene complement of Mycoplasma genitalium. Science 270:397-403. 41. Garrity, G. M. (ed.). 2001. Bergey's Manual of Systematic Bacteriology, 2 ed, vol. 1. Springer-Verlag, New York. 42. Glass, J. I., E. J. Lefkowitz, J. S. Glass, C. R. Heiner, E. Y. Chen, and G. H. Cassell. 2000. The complete sequence of the mucosal pathogen Ureaplasma urealyticum. Nature 407:757-762. 43. Hahn, T.-W., M. J. Willby, and D. C. Krause. 1998. HMW1 is required for cytadhesin PI trafficking to the attachment organelle in Mycoplasma pneumoniae. J. Bacteriol. 180:1270-1276. 44. Herold, S., J. Siebert, A. Huber, and H. Schmidt. 2005. Global expression of prophage genes in Escherichia coli 0157:H7 strain EDL933 in response to norfloxacin. Antimicrob. Agents Chemother. 49:931-944. 45. Hilsenbeck, S. G., W. E. Friedrichs, R. Schiff, P. O'Connell, R. K. Hansen, C. K. Osborne, and S. A. Fuqua. 1999. Statistical analysis of array expression data as applied to the problem of tamoxifen resistance. J. Natl. Cancer Inst. 91:453-459. 28

46. Himmelreich, R., H. Hilbert, H. Plagens, E. Pirkl, B.C. Li, and R. Herrmann. 1996. Complete sequence analysis of the genome of the bacterium Mycoplasma pneumoniae. Nucleic Acids Res. 24:4420-4449. 47. Hoque, M. O., C. C. Lee, P. Cairns, M. Schoenberg, and D. Sidransky. 2003. Genome-wide genetic characterization of bladder cancer: a comparison of high- density single-nucleotide polymorphism arrays and PCR-based microsatellite analysis. Cancer Res. 63:2216-2222. 48. Hsu, T., S. Artiushin, and F. C. Minion. 1997. Cloning and functional analysis of the P97 swine cilium adhesin gene of Mycoplasma hyopneumoniae. J. Bacteriol. 179:1317-1323. 49. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infect. Immun. 66:4762-4766. 50. Hsu, T., and F. C. Minion. 1998. Molecular analysis of the P97 cilium adhesin operon of Mycoplasma hyopneumoniae. Gene 214:13-23. 51. Hutchison III, C. A., and M. G. Montague. 2002. Mycoplasmas and the minimal genome concept, p. 221-253. In S. Razin and H. Herrmann (ed.), Molecular biology and pathogenicity of mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 52. Hutchison III, C. A., S. N. Peterson, S. R. Gill, R. T. Cline, O. White, C. M. Fraser, H. O. Smith, and J. C. Venter. 1999. Global transposon mutagenesis and a minimal mycoplasma genome. Science 286:2165-2169. 53. Jaffe, J. D., N. Stange-Thomann, C. Smith, D. Decaprio, S. Fisher, J. Butler, S. Calvo, T. Elkins, M. G. Fitzgerald, N. Hafez, C. D. Kodira, J. Major, S. Wandg, J. Wilkinson, R. Nicol, C. Nusbaum, B. Birren, H. C. Berg, and G. M. Church. 2004. The complete genome and proteome of Mycoplasma mobile. Genome Res. 14:1447- 1461. 54. Jarvill-Taylor, K., C. VanDyk, and F. C. Minion. 1999. Cloning of mnuA, a membrane nuclease encoding gene of Mycoplasma pulmonis, and analysis of its expression m Escherichia coli. J. Bacteriol. 181:1853-1860. 55. Keramas, G., D. D. Bang, M. Lund, M. Madsen, H. Bunkenborg, P. Telleman, and C. B. Christensen. 2004. Use of culture, PCR analysis, and DNA microarrays for detection of Campylobacter jejuni and Campylobacter coli from chicken feces. J. Clin. Microbiol. 42:3985-3891. 56. Kerr, M. K., M. Martin, and G. A. Churchill. 2000. Analysis of variance for gene expression microarray data. J. Comput. Biol. 7:819-837. 29

57. Kim, B. C., J. H. Park, and M. B. Gu. 2005. Multiple and simultaneous detection of specific bacteria in enriched bacterial communities using a DNA microarray chip with randomly generated genomic DNA probes. Anal. Chem. 77:2311-2317. 58. King, K. W., and K. Dybvig. 1991. Plasmid transformation of Mycoplasma mycoides subspecies mycoides is promoted by high concentrations of polyethylene glycol. Plasmid 26:108-115. 59. King, K. W., and K. Dybvig. 1994. Transformation of Mycoplasma capricolum and examination of restriction modification in M. capricolum and Mycoplasma mycoides subsp. mycoides. Plasmid 31:308-311. 60. King, K. W., D. H. Faulds, E. L. Rosey, and R. J. Yancey. 1997. Characterization of the gene encoding Mhpl from Mycoplasma hyopneumoniae and examination of Mhpl's vaccine potential. Vaccine 15:25-35. 61. Kirchhoff, H., K. Mohan, R. Schmidt, M. Runge, D. R. Brown, M. B. Brown, C. M. Foggin, P. Muvavarirwa, H. Lehmann, and J. Flossdorf. 1997. Mycoplasma crocodyli sp. nov, a new species from crocodiles. Int. J. Syst. Bact. 47:742-746. 62. Knudtson, K. L., and F. C. Minion. 1993. Construction of 7n4001 lac derivatives to be used as promoter probe vectors in mycoplasmas. Gene 137:217-222. 63. Knudtson, K. L., and F. C. Minion. 1993. Use of lac gene fusions in the analysis of Acholeplasma upstream gene regulatory sequences. J. Bacteriol. 176:2763-2766. 64. Krause, D. C., T. Proft, C. T. Hedreyda, H. Hilbert, H. Plagens, and R. Herrmann. 1997. Transposon mutagenesis reinforces the correlation between Mycoplasma pneumoniae cy to skeletal protein HMW2 and cytadherence. J. Bacteriol. 179:2668- 2677. 65. Kurth, K. T., H. Tsungda, E. Snook, E. L. Thacker, B. J. Thacker, and F. C. Minion. 2002. Use of a Mycoplasma hyopneumoniae nested polymerase chain reaction test to determine the optimal sampling sites in swine. J. Vet. Diagn. Invest. 14:463-469. 66. Lashkari, D. E., J. L. DeRisi, J. H. McCusker, A. F. Namath, C. Gentiles, S. Y. Hwang, P. O. Brown, and R. W. Davis. 1997. Yeast microarrays for genome wide parallel genetic and gene expression analysis. Proc. Natl. Acad. Sci. USA 94:13057- 13062. 67. Lin, J. H., M. J. Pan, C. W. Liao, and C. N. Weng. 2002. In vivo and in vitro comparisons of spray-drying and solvent- evaporation preparation of microencapsulated Mycoplasma hyopneumoniae for use as an orally administered vaccine for pigs. J. Vet. Med. Res. 63:1118-1123. 30

68. Lin, J. H., C. N. Weng, C. W. Liao, K. S. Yeh, and M. J. Pan. 2003. Protective effects of oral microencapsulated Mycoplasma hyopneumoniae vaccine prepared by co-spray drying method. J. Vet. Med. Sci. 65:69-74. 69. Maes, D., H. Deluyker, M. Verdonck, F. Castryck, C. Miry, A. Lein, B. Vrijens, and A. de Kruif. 1998. The effect of vaccination against Mycoplasma hypopneumoniae in pig herds with a continuous production system. Zentralbl. Veterinarmed. B. 45:495- 505. 70. Maniloff, J. 2002. Phylogeny and evolution, p. 31-44. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 71. Maniloff, J., R. N. McElhaney, L. R. Finch, and J. B. Baseman. 1992. Mycoplasma. Molecular biology and pathogenesis. American Society for Microbiology, Washington, D. C. 72. Mebus, C. A., and N. R. Underdahl. 1977. Scanning electron microscopy of trachea and bronchi from gnotobiotic pigs inoculated with Mycoplasma hyopneumoniae. Am. J. Vet. Res. 43:1249-1254. 73. Messick, J. B. 2004. Hemotrophic mycoplasmas (hemoplasmas): a review and new insights into pathogenic potential. Vet. Clin. Pathol. 33:2-13. 74. Minion, F. C. 2002. Molecular pathogenesis of animal mycoplasma respiratory pathogens. Front. Biosci. 7:dl410-1422. URL: http://www.bioscience.Org/2002/vl417/d/minion/fulltext.htm. 75. Minion, F. C., C. Adams, and T. Hsu. 2000. R1 region of P97 mediates adherence of Mycoplasma hyopneumoniae to swine cilia. Infect Immun 68:3056-3060. 76. Minion, F. C., and K. Jarvill-Taylor. 1994. Membrane-associated hemolysin activities in mycoplasmas. FEMS Microbiol. Lett. 116:101-106. 77. Minion, F. C., K. J. Jarvill-Taylor, D. E. Billings, and E. Tigges. 1993. Membrane- associated nuclease activities in mycoplasmas. J. Bacteriol. 175:7842-7847. 78. Minion, F. C., and P. A. Kapke. 1998. Transformation of mycoplasmas, p. 227-234. In R. Miles and R. Nicholas (ed.), Methods in Molecular Biology: Mycoplasma Protocols, vol. 104. Humana Press, Totowa. 79. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. (In press) 186. 31

80. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 81. Moore, R. J., C. Lenghaus, S. A. Sheedy, and T. J. Doran. 2001. Improved vectors for expression library immunization — application to Mycoplasma hyopneumoniae infection in pigs. Vacc. 20:115-120. 82. Murphy, G. M., Jr., C. Kremer, H. E. Rodrigues, and A. F. Schatzberg. 2003. Pharmacogenetics of antidepressant medication intolerance. Am. J. Psychiatry 160:1830-1835. 83. Muto, A., and C. Ushida. 2002. Transcription and translation, p. 323-345. In S. Razin and R. Herrmann (ed.), Molecular biology and pathogenicity of mycoplasmas. Plenum Press, New York. 84. Nessling, M., K. Richter, C. Schwaenen, P. Roerig, G. Wrobel, S. Wessendorf, B. Fritz, M. Bentz, H. P. Sinn, B. Radlwimmer, and P. Lichter. 2005. Candidate genes in breast cancer revealed by microarray-based comparative genomic hybridization of archived tissue. Cancer Res. 65:439-447. 85. Nguyen, D. V., A. Bulak Arpat, N. Wang, and R. J. Carroll. 2002. DNA microarray experiments: Biological and technological aspects. Biometrics 58:701-717. 86. Nilsson, C., A. Sillen, L. Eriksson, M. L. Strand, H. Enroth, S. Normark, P. Falk, and L. Engstrand. 2003. Correlation between cag pathogenicity island composition and Helicobacter pylori-associated gastroduodenal disease. Infect Immun 71:6573-6581. 87. Opriessnig, T., E. L. Thacker, S. Yu, M. Fenaux, X. J. Meng, and P. G. Halbur. 2004. Experimental reproduction of postweaning multisystemic wasting syndrome in pigs by dual infection with Mycoplasma hyopneumoniae and porcine circovirus type 2. Vet. Pathol. 41:624-640. 88. Oshima, K., S. Kakizawa, N. Nishigawa, H. Y. Jung, W. Wei, S. Suzuki, R. Arashida, D. Nakata, S. Miyata, M. Ugaki, and S. Namba. 2004. Reductive evolution suggested from the complete genome sequence of a plant-pathogenic phytoplasma. Nat. Genet. 36:27-29. 89. Ott, R. L. 1993. An introduction to statistical methods and data analysis, 4 ed. Wadsworth, Inc., Belmont. 90. Panicker, G., D. R. Call, M. J. Krug, and A. K. Bej. 2004. Detection of pathogenic Vibrio spp. in shellfish by using multiplex PCR and DNA microarrays. Appl. Environ. Microbiol. 70:7436-7444. 32

91. Papazisi, L., T. S. Gorton, G. Kutish, P. F. Markham, G. F. Browning, D. K. Nguyen, S. Swartzell, A. Madan, G. Mahairas, and S. J. Geary. 2003. The complete genome sequence of the avian pathogen Mycoplasma gallisepticum strain R(low). Microbiol. 149:2307-2316. 92. Park, S.-C., S. Yibchok-Anun, H. Cheng, T. F. Young, E. L. Thacker, F. C. Minion, R. F. Ross, and W. H. Hsu. 2002. Mycoplasma hyopneumoniae increases intracellular calcium release in porcine ciliated tracheal cells. Infect. Immun. 70:2502-2506. 93. Plant, L., V. Asp, L. Lovkvist, J. Sundqvist, and A. B. Jonsson. 2004. Epithelial cell responses induced upon adherence of pathogenic Neisseria. Cell Microbiol. 6:663- 670. 94. Porwollik, S., E. F. Boyd, C. Choy, P. Cheng, L. Florea, E. Proctor, and M. McClelland. 2004. Characterization of Salmonella enterica subspecies I genovars by use of microarrays. J. Bacteriol. 186:5883-5898. 95. Prouty, A. M., I. E. Brodsky, J. Manos, R. Belas, S. Falkow, and J. S. Gunn. 2004. Transcriptional regulation of Salmonella enterica serovar Typhimurium genes by bile. Immunol. Med. Microbiol. 41:177-185. 96. Razin, S. 1975. The mycoplasma membrane. Prog. Surf. Membr. Sci. 9:257-312. 97. Razin, S. 1978. The mycoplasmas. Microbiol. Rev. 42:414-470. 98. Razin, S., and E. A. Freundt. 1984. The mycoplasmas, p. 740-793. In N. R. Kreig and J. G. Holt (ed.), Sergey's manual of systematic bacteriology, vol. 1. Williams and Wilkins, Baltimore. 99. Razin, S., D. Yogev, and Y. Naot. 1998. Molecular biology and pathogenicity of mycoplasmas. Microbiol. Mol. Biol. Rev. 62:1094-1156. 100. Reen, F. J., E. F. Boyd, S. Porwollik, B. P. Murphy, D. Gilroy, S. Fanning, and M. McClelland. 2005. Genomic comparisons of Salmonella enterica serovar Dublin, Agona, and Typhimurium strains recently isolated from milk filters and bovine samples from Ireland, using a Salmonella microarray. Appl. Environ. Microbiol. 71:1616-1625. 101. Ross, R. F., B. J. Zimmermann-Erickson, and T. F. Young. 1984. Characteristics of protective activity of Mycoplasma hyopneumoniae vaccine. Am. J. Vet. Res. 45:1899-1905. 102. Ross, R. F., B. J. Zimmermann-Erickson, and T. F. Young. 1984. Characteristics of protective activity of Mycoplasma hyopneumoniae vaccine. Am. J. Vet. Med. 45:1899-1905. 33

103. Roth, S. B., J. Jalava, O. Ruuskanen, A. Ruohola, and S. Nikkari. 2004. Use of an oligonucleotide array for laboratory diagnosis of bacteria responsible for acute upper respiratory infections. J. Clin. Microbiol. 42:4268-4274. 104. Rottem, S. 2002. Invasion of mycoplasmas into and fusion with host cells, p. 391 - 401. In S. Razin and R. Herrmann (ed.), Molecular biology and pathogenicity of mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 105. Sasaki, Y., J. Ishikawa, A. Yamashita, K. Oshima, T. Kenri, K. Furuya, C. Yoshino, A. Horino, T. Shiba, T. Sasaki, and M. Hattori. 2002. The complete genomic sequence of Mycoplasma penetrans, an intracellular bacterial pathogen in humans. Nucleic Acids Res. 30:5293-5300. 106. Schena, M. 2003. Microarray analysis. John Wiley & Sons, Inc., New Jersey. 107. Schena, M., and R. W. Davis. 2000. Technology standards for microarray research. In M. Schena (ed.), Microarray Biochip Technology. Biotechniques Books, Natick, N.Y. 108. Schena, M., D. Shalon, R. W. Davis, and P. O. Brown. 1995. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270:467-470. 109. Schena, M., D. Shalon, R. Heller, A. Chai, P. O. Brown, and R. W. Davis. 1996. Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc. Natl. Acad. Sci. U.S.A. 93:10614-10619. 110. Schmidt, J. A., G. F. Browning, and P. F. Markham. 2004. Mycoplasma hyopneumoniae p65 surface lipoprotein is a lipolytic enzyme with a preference for shorter-chain fatty acids. J. Bacteriol. 186:5790-5798. 111. Sergeev, N., M. Distler, S. Courtney, S. F. Al-Khaldi, D. Volokhov, V. Chizhikov, and A. Rasooly. 2004. Multipathogen oligonucleotide microarray for environmental and biodefense applications. Biosens. Bioelectron. 20:684-698. 112. Shaffer, L. G., and B. A. Bejjani. 2004. A cytogeneticist's perspective on genomic microarrays. Hum. Reprod. Update 10:221-226. 113. Shimoji, Y., E. Oishi, Y. Muneta, H. Nosaka, and Y. Mori. 2003. Vaccine efficacy of the attenuated Erysipelothrix rhusiopathiae YS-19 expressing a recombinant protein of Mycoplasma hyopneumoniae P97 adhesin against mycoplasmal pneumonia of swine. Vaccine 21:532-537. 114. Simecka, J. W., J. K. Davis, M. K. Davidson, S. E. Ross, C. T. K.-H. Stadtlander, and G. H. Cassell. 1992. Mycoplasma diseases of animals, p. 391-415. In J. Maniloff, R. N. McElhaney, L. R. Finch, and J. B. Baseman (ed.), Mycoplasmas: molecular biology and pathogenesis. American Society for Microbiology, Washington, D. C. 34

115. Singh-Gasson, S., R. D. Green, Y. Yue, C. Nelson, F. Blattner, M. R. Sussman, and F. Cerrina. 1999. Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror array. Nat. Biotechnol. 17:974-978. 116. Smajs, D., M. McKevitt, J. K. Howell, S. J. Norris, W. W. Cai, T. Palzkill, and G. M. Weinstock. 2005. Transcriptome of Treponema pallidum', gene expression profile during experimental rabbit infection. J. Bacteriol. 187:1866-1874. 117. Smyth, G. K., and T. Speed. 2003. Normalization of cDNA microarray data. Methods 31:265-273. 118. Snyder, L. A., J. K. Davies, and N. J. Saunders. 2004. Microarray genomotyping of key experimental strains of Neisseria gonorrhoeae reveals gene complement diversity and five new neisserial genes associated with Minimal Mobile Elements. BMC Genomics 5:23. 119. Southern, E. M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol. 98:503-517. 120. Storey, J. D., and R. Tibshirani. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA 100:9440-9445. 121. Taboada, E. N, R. R. Acedillo, C. D. Carrillo, W. A. Findlay, D. T. Medeiros, O. L. Mykytczuk, M. J. Roberts, C. A. Valencia, J. M. Farber, and J. H. Nash. 2004. Large- scale comparative genomics meta-analysis of Campylobacter jejuni isolates reveals low level of genome plasticity. J. Clin. Microbiol. 42:4566-4576. 122. Tajima, M., and T. Yagihashi. 1982. Interaction of Mycoplasma hyopneumoniae with the porcine respiratory epithelium as observed by electron microscopy. Infect. Immun. 37:1162-1169. 123. Tanaka, T. S., S. A. Jaradat, M. K. Lim, G. J. Kargul, X. Wang, M. J. Grahovac, S. Pantano, Y. Sano, Y. Piao, R. Nagaraja, H. Doi, W. H. Wood, 3rd, K. G. Becker, and M. S. Ko. 2000. Genome-wide expression profiling of mid-gestation placenta and embryo using a 15,000 mouse developmental cDNA microarray. Proc. Natl. Acad. Sci. U.S.A. 97:9127-9132. 124. Tettelin, H., V. Masignani, M. J. Cieslewicz, J. A. Eisen, S. Peterson, M. R. Wessels, I. T. Paulsen, K. E. Nelson, I. Margarit, T. D. Read, L. C. Madoff, A. M. Wolf, M. J. Beanan, L. M. Brinkac, S. C. Daugherty, R. T. DeBoy, A. S. Durkin, J. F. Kolonay, R. Madupu, M. R. Lewis, D. Radune, N. B. Fedorova, D. Scanlan, H. Khouri, S. Mulligan, H. A. Carty, R. T. Cline, S. E. Van Aken, J. Gill, M. Scarselli, M. Mora, E. T. Iacobini, C. Brettoni, G. Galli, M. Mariani, F. Vegni, D. Maione, D. Rinaudo, R. Rappuoli, J. L. Telford, D. L. Kasper, G. Grandi, and C. M. Fraser. 2002. Complete 35

genome sequence and comparative genomic analysis of an emerging human pathogen, serotype V Streptococcus agalactiae. Proc. Natl. Acad. Sci. U.S.A. 99:12391-12396. 125. Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 126. Thacker, E. L., B. J. Thacker, and B. H. Janke. 2001. Interaction between Mycoplasma hyopneumoniae and swine influenza virus. J. Clin. Microbiol. 39:2525- 2530. 127. Thacker, E. L., B. J. Thacker, M. Kuhn, P. A. Hawkins, and W. R. Waters. 2000. Evaluation of local and systemic immune responses induced by intramuscular injection of a Mycoplasma hyopneumoniae bacterin to pigs. Am. J. Vet. Res. 61:1384-1389. 128. Vicca, J., D. Maes, L. Thermote, J. Peeters, F. Haesebrouck, and A. de Kruif. 2002. Patterns of Mycoplasma hyopneumoniae infections in Belgian farrow-to-finish pig herds with diverging disease-course. J. Vet. Med. B. Infect. Dis. Vet. Public Health 49:349-353. 129. Wang, Q., J. G. Frye, M. McClelland, and R. M. Harshey. 2004. Gene expression patterns during swarming in Salmonella typhimurium: genes specific to surface growth and putative new motility and pathogenicity genes. Mol. Microbiol. 52:169- 187. 130. Warrington, J. A., S. Dee, and M. Trulson. 2000. Large-scale genomic analysis using affymetrix GeneChip probe assays, p. 149-166. In M. Schena (ed.), Microarray Biochip Technology. Biotechnqiues Books, Natick, N.Y. 131. Weiner III, J., C.-U. Zimmerman, H. W. H. Gohlmann, and R. Herrmann. 2003. Transcription profiles of the bacterium Mycoplasma pneumoniae grown at different temperatures. Nucleic Acids Res. 31:6306-6320. 132. Welford, S. M., J. Gregg, E. Chen, D. Garrison, P. H. Sorensen, C. T. Denny, and S. F. Nelson. 1998. Detection of differentially expressed genes in primary tumor tissues using representational differences analysis coupled to microarray hybridization. Nucleic Acids Res. 26:3059-3065. 133. Westberg, J., A. Persson, A. Holmberg, A. Goesmann, J. Lundegerb, K.-E. Johansson, B. Pettersson, and M. Uhlaen. 2004. The genome sequence of Mycoplasma mycoides subsp. mycoides SC type strain PG1T, the causative agent of contagious bovine pleuropneumonia (CBPP). Genome Res. 14:221-227. 36

134. Wick, L. M., W. Qi, D. W. Lâcher, and T. S. Whittam. 2005. Evolution of genomic content in the stepwise emergence of Escherichia coli 0157:H7. J. Bacteriol. 187:1783-1791. 135. Willse, A., T. M. Straub, S. C. Wunschel, J. A. Small, D. R. Call, D. S. Daly, and D. P. Chandler. 2004. Quantitative oligonucleotide microarray fingerprinting of Salmonella enterica isolates. Nucleic Acids Res. 32:1848-1856. 136. Winner, F., R. Rosengarten, and C. Citti. 2000. In vitro cell invasion of Mycoplasma gallisepticum. Infect Immun 68:4238-4244. 137. Wong, K. K., Y. T. Tsang, J. Shen, R. S. Cheng, Y. M. Chang, T. K. Man, and C. C. Lau. 2004. Allelic imbalance analysis by high-density single-nucleotide polymorphic allele (SNP) array with whole genome amplified DNA. Nucleic Acids Res. 32:e69. 138. Wu, C. F., J. J. Valdes, W. E. Bentley, and J. W. Sekowski. 2003. DNA microarray for discrimination between pathogenic 0157:H7 EDL933 and non-pathogenic Escherichia coli strains. Biosens. Bioelectron. 19:1-8. 139. Yamao, F., A. Muto, Y. Kawauchi, M. Iwami, S. Iwagami, Y. Azumi, and S. Osawa. 1985. UGA is read as tryptophan in Mycoplasma capricolum. Proc. Natl. Acad. Sci. USA 82:2306-2309. 140. Zhang, J., Y. Wang, C. Shou, G. Xu, X. Chen, J. Wu, Y. Xie, J. Li, S. So, and J. Jiafu. 2002. Detection of Mycoplasma hyorhinis in gastric cancer using bio-chip technology. Zhonghua Yi Xue Za Zhi 82:961-965. 141. Zhang, Q., T. F. Young, and R. F. Ross. 1994. Glycolipid receptors for attachment of Mycoplasma hyopneumoniae to porcine respiratory ciliated cells. Infect. Immun. 62:4367-4373. 142. Zhang, Q., T. F. Young, and R. F. Ross. 1995. Identification and characterization of a Mycoplasma hyopneumoniae adhesin. Infect Immun 63:1013-1019. 143. Zhang, Q., T. F. Young, and R. F. Ross. 1994. Microliter plate adherence assay and receptor analogs for Mycoplasma hyopneumoniae. Infect. Immun. 62:1616-1622. 144. Zielinski, G. C., and R. F. Ross. 1990. Effect of growth in cell cultures and strain on virulence of Mycoplasma hyopneumoniae for swine. Am. J. Vet. Res. 51:344-348. 37

CHAPTER 2. TRANSCRIPTIONAL PROFILING OF MYCOPLASMA HYOPNEUMONIAE DURING HEAT SHOCK USING MICROARRAYS

A paper to be submitted to Infection and Immunity

Melissa L. Madsen,1 Dan Nettleton,2 Eileen L. Thacker,1 Robert Edwards,3 and F. Chris Minion1

Departments of Veterinary Microbiology and Preventive Medicine1 and Statistics2, Iowa State University, Ames, IA; San Diego State University, San Diego, CA3

Abstract Bacterial pathogens undergo stress during host colonization and disease processes. These stresses result in changes in gene expression to compensate for potentially lethal environments developed in the host during disease. Mycoplasma hyopneumoniae colonizes swine epithelium and causes a pneumonia that predisposes the host to enhanced disease from other pathogens. How M. hyopneumoniae responds to changing environments in the lung during disease progression is not known. In fact, little is known concerning the capabilities of mycoplasmas to respond to changing growth environments. With limited genes, mycoplasmas are thought to lack complex regulatory circuits and few mechanisms for gene regulation. A microarray consisting of 632 of the 698 open reading frames of M. hyopneumoniae was constructed and used to study gene expression differences during a temperature shift from 37°C to 42°C, a temperature swing that might be encountered during disease. To enhance sensitivity, a unique hexamer primer set was employed for generating cDNA from only mRNA species. Our analysis identified ninety-one genes that were either up- or down-regulated in response to heat shock conditions (p <0.01) with an estimated false discovery rate of four percent. Many of the heat shock proteins previously characterized in other bacteria were identified as significant in this study as well. A proportion of the identified genes (54 of 91) had no assigned function. 38

Introduction Mycoplasma hyopneumoniae is the causative agent of porcine enzootic pneumonia and is a major contributor to the porcine respiratory diseases complex (24, 28). Like other mycoplasmas, M. hyopneumoniae has a small genome with limited biosynthetic potential (18). Mycoplasmas in general are rarely exposed to severe environmental changes; they are transmitted animal to animal and do not survive long outside their host. They do, however, encounter changing environmental conditions within the host as the immune response develops and responds with the recruitment of large numbers of neutrophils and macrophages to the infection site and the host becomes febrile (29). Some invasive mycoplasma species, i.e. Mycoplasma bovis, encounter different environments within the tissues it colonizes after becoming systemic. Thus, it is of interest to study potential growth conditions that might impact the mycoplasmas^ ability to adapt to and survive these changing conditions. Heat shock proteins (Hsps) are a group of proteins induced during stressful conditions that result in denaturation of polypeptides. Temperature extremes, high pressure and exposure to toxic compounds can induce expression of these proteins, conditions that result in protein unfolding and denaturation (21). The Hsps are generally thought of as chaperones, proteins involved in the correct folding of polypeptides as they emerge from the ribosome. They operate by binding to exposed hydrophobic residues in improperly folded protein structures. They can also be involved in the assembly or disassembly of multimeric protein structures and translocation of proteins across membranes. An additional class of Hsps includes proteases that degrade misfolded or abnormal proteins (9, 10). The Hsps are highly conserved and appear to be universal, having functions required for normal cell growth. The regulation of heat shock proteins falls into four general classes, proteins regulated by sigma32 (rpoH) and sigma24 (rpoE) (10), those regulated by transcriptional repressors (20), and those with unknown regulatory mechanisms (31). Mycoplasmas, however, do not have the typical sigma factors involved in responses to high temperature shifts; only a single sigma factor has been identified in any mycoplasma genome (2, 12, 18, 22, 33). Consequently the current thought on gene regulation in mycoplasmas is that the majority of genes are constitutively expressed and some of these may undergo phase shifting due to genetic changes occurring during DNA replication (36), and that those that undergo regulation are controlled by a limited number of transcriptional regulators (19). Heat shock has long been considered a stress condition that many pathogens encounter in their pathogen-host relationship. While much has been learned about the Hsp genes and their regulation in many bacterial species (10), few studies have examined mycoplasmas for heat shock responses. In 1990, two reports of heat shock studies involved identification of heat 39 shock proteins (Hsps) by gel electrophoresis in Mycoplasma capricolum subsp. capricolum and Acholeplasma laidlawii (1,3, 4). More recently, transcriptional profiling was performed on Mycoplasma pneumoniae (32) to identify a set of genes that undergo regulation in response to changing growth temperatures. The results of that study indicated that 47 genes were up-regulated and 30 genes were down-regulated during a temperature shift from 32°C to 42°C at ap-value <0.001. This report is also notable in that it was the first genome-wide transcriptional profiling study performed in a mycoplasma species. Despite its importance to swine production, there have been few studies of potential molecular mechanisms of pathogenesis with M. hyopneumoniae and how it might respond to environmental changes. This has probably been due to the difficulty in growing the organism and its recalcitrance to genetic manipulation. The few reports have focused on mechanisms of adherence to swine cilia (13, 14, 17, 38) with no reports of potential stress-related proteins. Analysis of the genome sequence has identified putative Hsps DnaK (mhp072), DnaJ (mhp073), ClpB (mhp278), GrpE (mhpOl 1), Lon (mhp541), and a putative heat shock regulator HrcA (mhpOlO) (18). To better understand the mechanisms involved in stress responses, we have determined M. hyopneumoniae's response to heat shock using a microarray approach. Our studies show significant changes in transcriptional activities in numerous genes suggesting that M. hyopneumoniae like other bacteria tightly regulate their genes in response to the environment.

Materials and Methods Mycoplasma strains and culture conditions Pathogenic M. hyopneumoniae strain 232, a derivative of strain 11, was used in this study (16). Cultures used were passed fewer than 15 times in vitro in Friis media as previously described (8). Two hundred fifty ml flasks containing 125 ml cultures were grown at 37°C to early-log phase as determined by medium color change and optical density. Flasks were shifted to 30°C for 2 h and then paired flasks shifted to 37 or 42°C for 30 min. There were six replicates. Cells were pelleted by centrifugation at 24,000 x g, and 500 (il of RNAlater (Ambion, Inc. Austin, Tex.) was added to pellet. Pellets were stored at -70°C until the total RNA was isolated.

Microarray construction The M. hyopneumoniae microarray consists of polymerase chain reaction (PCR) products (probes) spotted to glass substrates. To identify primers for PCR reactions, a FASTA file was 40 first generated from the M. hyopneumoniae strain 232 sequence (18) containing the DNA sequences for all putative open reading frames (ORFs). This file was then analyzed with Primer3 software (26) using a set of scripts to allow for iteration and unique primer design across the genome. The software began the design process (Figure 2.1) by reading and storing the sequences for each of the ORFs. The program then iterated through each ORF, determining those for which valid primers have been identified. At this point, a mispriming library file was created, which consisted of the sequences of all ORFs except for the one for which primers were to be designed. This ensured that any primers found would not match any other ORFs besides the ORF requiring primers. Next, the Primer3 parameters were written to a file (Table 2.1), and this file and the mispriming library were passed to Primer3 and the results appended to another file. The program then moved on to the next ORF. Once each ORF had been examined, any ORFs still needing primers were identified. The parameters passed to Primer3 (i.e., melting temperature, primer product size, mispriming score, etc.) were relaxed, and the process was then repeated a total of three times, iterating through each of the ORFs and re-running Primer3 on any ORFs that did not already have primers. At the end of the third evaluation, any ORFs for which unique primers could not be identified were not included in the microarray design. Two sets of primers were identified for each ORF. Primers were purchased from Integrated DNA Technologies (Coralville, la.) and were obtained as a frozen solution of 100 picomoles per microliter in water in 96 well plates. The primer sequences used to generate PCR products for array spotting are given in the Appendix B. Polymerase chain reactions were performed in 96 well plates in a MJ Research Dyad Thermal Cycler (Bio-Rad Laboratories, Inc., Waltham, Mass.) on all sets of PCR primers in 100 (j.1 reactions using the following conditions: IX PCR Buffer; 2.5 mM MgCh; 2.0 mM each dATP, dCTP, dTTP and dGTP (Roche Diagnostics Corp., Indianapolis, Ind.); 2 pmol of each primer, forward and reverse; 2.5 units Taq polymerase (New England Biolabs, Beverly, Mass.); and 100 ng M. hyopneumoniae strain 232 chromosomal DNA, which had been isolated by phenol :chloroform extraction. The thermal cycling conditions were denaturation at 95°C for 5 min; and then 30 cycles of denaturation at 94°C for 1 min, annealing at 50°C for 1 min, and elongation at 72°C for 30 sec; and a final elongation at 72°C for 10 min. Products were analyzed by 2% agarose gel electrophoresis to confirm that a single product of correct size was present. For failed reactions, a second round of PCR was performed with the first primer pair with lowered temperatures or the second set of PCR primers was used with the same reaction conditions. Forty-eight PCR products were sequenced to verify that the correct product was being made. The PCR products were purified by combining the PCR 41

Read in and store sequences for all ORFs

• Itéra e n ugh all

No

No

Yes

Run Phmer3 on (his ORF sequence No

Figure 2.1. Flow chart for PCR primer design. 42

Table 2.1. Initial Primer3 criteria for designing primers for the M. hyopneumoniae ORFs.

PRIMER DESIGN OPTIONS CRITERIA PRIMER MISPRIMING LIBRARY screen,seq PRIMER MAX MISPRIMING 12.0 PRIMER PAIR MAX MISPRIMING 24.0 PRIMER PRODUCT SIZE RANGE 250-350 100-500 PRIMER TASK pick_pcr_primers PRIMER GC CLAMP 1 PRIMER MIN SIZE 18 PRIMER OPT SIZE 20 PRIMER MAX SIZE 25 PRIMER MIN TM 55.0 PRIMER OPT TM 61.0 PRIMER MAX TM 63.0 PRIMER MAX DIFF TM 5.0 PRIMER MIN GC 30.0 PRIMER OPT GC PERCENT 50.0 PRIMER MAX GC 70.0 PRIMER NUM RETURN 2 PRIMER FIRST BASE INDEX 1 products with 3 volumes of 6 M potassium iodide and applying to FilterEX™96 well glass filter plates (Corning #3510; Corning, Inc., Big Flats, N.Y.). The plate was subjected to vacuum for 1 min and each well was washed with 100 gl of 80% ethanol, twice. After the second wash, the plates were dried under vacuum for 2 min and tapped to remove any residual ethanol droplets in the nozzle. Each plate was placed on its side and air dried at room temperature for 90 min. The DNA was eluted by adding 80 |xl of 10 mM Tris pH 8.5 per well and centrifuged for 2 min at 782 x g. The eluted DNA was captured in a 96 well plate (Coming #3365). Purified PCR products were quantified using a Spectramax 250 spectrophotometer (Molecular Devices, Sunnyvale, Calif.) and CoStar UV transparent 96- well flat-bottom plates (Coming #3635). Samples were dried using a Savant vacuum centrifuge with a microtiter plate adapter and resuspended to an approximate concentration of 200 ng per jj.1 in spotting buffer (Coming). A 1 pi volume of probe was diluted in 5 jul distilled water with 1 pi of loading dye (10 mM Tris-HCl (pH 7.6), 0.03% bromophenol blue, 43

0.03% xylene cyanol FF, 60% glycerol, 60 mM ethylenediamine tetraacetic acid) and electrophoresed in 2% agarose gels to confirm the concentration and purity of each product. Products were arrayed into 384 V-well plates (#781280; Greiner Bio-One, Inc., Longwood, Fl.) to prepare for printing. The microarrays were printed using a BioRobotics MicroGrid (Genomic Solutions, Ann Arbor, Mich.)(see Appendix C for the printing file) onto Corning UltraGAPS™ substrates. Each probe was printed in triplicate, in non-adjacent spots to reduce non-uniform substrate errors in replicated probes. The array consisted of 16 subarrays, with a 12-column x 12-row configuration. Two arrays were printed per slide. Humidity was increased during printing and pin strike speed slowed according to Coming's recommendation. The slides were allowed to dry on the printing deck overnight and stored for 48 h in a dessicator prior to transport and crosslinking. Slides were UV cross-linked at 450 mJ in a Stratalinker (Stratagene, La Jolla, Calif.) to immobilize DNA. Preliminary staining experiments were performed to determine the optimal cross-linking energy needed for each batch of arrays. Arrays were cross-linked with 350, 450 and 550 mJ and stained with SpotQC (Integrated DNA Technologies) per manufacturer's instructions. The energy giving the most uniform and highest signal strength was chosen for the batch treatment. Prior to hybridization, arrays were prehybridized with sodium borohydride to reduce background according to methods outlined by Raghavachari et al. (23).

RNA isolation RNA was isolated from frozen cell pellets using the Versagene™ RNA Purification System (Gentra Systems, Minneapolis, Minn.) according to the manufacturer's protocol. The optional step of DNase treatment was routinely performed on column according to the manufacturer's recommendation. With a cut-off of 150 bp, 5S rRNA and tRNAs were removed from the samples, limiting interference in downstream manipulations. Samples were quantified and checked for purity using the Nanodrop® ND-1000 Spectrophotometer (Nanodrop, Wilmington, Del.). If necessary, samples were concentrated using Microcon YM-30 micro-concentrators (Millipore, Billerica, Mass.) for optimal cDNA generation.

Random primer set for M. hyopneumoniae A set of 129 six-mer oligonucleotide primers was designed to generate cDNA sequences from M. hyopneumoniae mRNA (Appendix D). The hexamer oligonucleotide sequences were screened to remove any sequences that would potentially amplify rRNA or tRNA, which make up 95% or more of the total RNA preparation. All 4,096 potential hexamer 44 oligonucleotide sequences were generated in silico. Those sequences that were homologous with the 41 tRNA or rRNA genes were removed from the analysis. The remaining 327 primers were compared to the 698 ORFs in the annotated sequence. Primers were identified that would theoretically hybridize to the coding strand of the mRNA, and a minimal set of primers that would amplify all ORFs was identified. One hundred picomoles of each primer (Integrated DNA Technologies) were then combined to form a primer mixture for cDNA generation.

Target generation and hybridization Targets were generated from total RNA extracted from cell pellets as described above. Fluorescently labeled cDNA targets were generated and purified using the SuperScript™ Indirect cDNA Labeling System (Invitrogen Corp., Carlsbad, Calif.). Targets were labeled with Alexa Fluor™ 555 Reactive Dye or Alexa Fluor™ 647 Reactive Dye (Molecular Probes, Inc., Eugene, Ore.). Following purification of the labeled cDNA, samples were dried in a vacuum centrifuge and then resuspended in 10 pi Pronto! cDNA/long oligo hybridization solution (Coming). Targets were denatured at 95°C for 5 min and centrifuged at 13,000 x g for 2 min at room temperature. Targets from a control and heat shocked culture were then combined, pipetted to an array, and covered with a 22 x 22 mm HybriSlip™ (Schleicher & Schuell, Keene, N. H.). Slides were placed in a Coming hybridization chamber and incubated in a 42°C water bath for 12-16 h. Slides were washed according to Coming's UltraGAPS™ protocol and dried by centrifugation.

Image acquisition and data analysis Each dye channel of each array was scanned using a ScanArray Express laser scanner (Applied BioSystems, Inc., Foster City, Calif.) three times under varying laser power and PMT gain settings to increase the dynamic range of expression measurement (6). Images were quantified using the softWorRx Tracker analysis software package (Applied Precision, Inc., Issaquah, Wash.). Spot-specific mean signals were corrected for local background by subtracting spot-specific median background intensities. The natural logarithm of the background-corrected signals from a single scan were adjusted by an additive constant so that all scans of the same array-by-dye combination would have a common median. The median of these adjusted-log-background-corrected signals across multiple scans was then computed for each spot to obtain one value for each combination of spot, array, and dye channel. These data for the two dye channels on any given array were normalized using LOWESS normalization to adjust for intensity-dependent dye bias (7, 35). Following 45

LOWE SS adjustment, the data from each channel were adjusted by an additive constant so that the median for any combination of array and dye would be the same for all array-by-dye combinations. The normalized values for triplicate spots were averaged within each array to produce one normalized measure of expression for each of the 632 probe sequences and each of the 12 RNA samples.

Statistical analysis A separate mixed linear model analysis was conducted for each probe sequence using the normalized data (34). Each mixed model included fixed effects for treatment (heat shock versus control), slide region (upper versus lower), and dye (Alexa 555 versus 647) as well as random effects for slide and slide-by-region interaction. A t-test for differential expression across treatments was conducted for each probe as part of our mixed linear model analyses. The 632 p-values from these /-tests were converted to ^-values using the method of Storey and Tibshirani (27). These ^-values can be used to obtain approximate control of the False Discovery Rate (FDR) at a specified value. For example, declaring probes with ^-values less than or equal to 0.05 to be differentially expressed produces a list of significant results for which the FDR is estimated to be approximately 5%. Along with ^-values, estimates of fold- change were computed for each probe by taking the inverse natural log of the mean treatment difference estimated as part of our mixed linear model analyses.

Validation of microarray data To confirm significant transcriptional differences between genes, semi-quantitative PCR analysis was performed. PCR was performed in 96 well plates in a MJ Research Dyad Thermal Cycler (Bio-Rad Laboratories, Inc., Waltham, Mass.) in 20 pi reactions, with the 6 study replicates for each ORF from control and heat shocked samples, using the following conditions: IX PCR Buffer; 2 mM MgCli; 1.0 mM each dATP, dCTP, dTTP and dGTP (Fisher Bioreagents, Fairlawn, NJ); 5 pmol of each primer, forward and reverse; 2.5 units Taq polymerase (New England Biolabs, Beverly, Mass.); and 2 pi cDNA from heat shocked samples diluted 1:50 or 2 pi cDNA from control samples diluted 1:50. The thermal cycling conditions were denaturation at 95°C for 5 min; and then 25 cycles of denaturation at 94°C for 1 min, annealing at 50°C for 1 min, and elongation at 72°C for 30 sec; and a final elongation at 72°C for 10 min. The reactions were mixed with 4 pi 6X loading dye and 10 pi were analyzed by 1.5% agarose gel electrophoresis then stained with 0.5 pg/ml ethidium bromide to confirm that a single product of correct size was present. The gel was visualized, 46 digitized and band density was measured with FlourChem 8000 version 2.0 software (Alpha Innotech Corp., San Leandro, Calif.). Analysis of variance was performed to determine differences between band density values of heat shocked versus control samples. Band densities were background corrected by subtracting the background density within the reaction lane from the sample band density. Estimated differences were considered significant if the /rvalue of a /-test was <0.05.

Results Primer design and array construction Primers were designed for PCR reactions based on the M. hyopneumoniae genome sequence. Of the 698 putative ORFs, primers could be designed for 686 ORFs based on the criteria shown in Table 2.1. Polymerase chain reaction conditions were optimized for product yield using a temperature gradient program (data not shown). Also, the deoxynucleotide concentrations were optimized by increasing to 2 mM, which significantly improved product concentration. This was probably necessary because of the 28.6% A+T content of the genome. When the first set of primers failed to give positive PCR products, a second set of primers were tested. In some instances a third set of primers were designed with reduced stringencies. The 66 missing ORFs (Appendix E) included those with no unique primers, multiple products, incorrect product size, or failed reactions. Ninety-two percent of the ORFs with designed primer pairs (632/686) gave positive reactions and were spotted on the array. Following purification, PCR reaction products were analyzed by agarose gel for consistency in concentration and product size (Figure 2.2).

Figure 2.2. Example of agarose gel of purified PCR products. 47

Microarrays were printed to Corning UltraGAPS™ substrates and allowed to dry on the printing platform overnight. For each batch of substrates, the substrates were cross-linked at 350, 450 and 550 mJ, prehybridized, and several slides per batch were stained with SpotQC to determine the spot morphology, general array quality, and optimal cross-linking energy. Figure 2.3 gives an example of spot morphology and optimization of the cross-linking energy.

Figure 2.3. Determination of cross-linking energy. Following cross-linking, slides were stained with SpotQC and scanned with the Cy3 channel. The image shown represents the same subarray from three slides cross-linked at 350, 450 and 550 mJ (left to right). The signal is given in false color to indicate intensity. Not all spots reacted with the dye equally because of the sequence variation.

Heat shock studies Six independent RNA samples from heat-shocked cells were paired with six independent RNA samples from control cells for hybridization on six two-color microarrays. The six arrays were printed on a total of three slides. Each slide was divided into two regions (upper and lower), and each region contained the full array of spots. Each of the 632 sequences in the array was represented by at least three replicate spots. For three of the six arrays, the control sample was labeled with Alexa 555 dye and compared to the heat-shocked sample labeled with Alexa 647 dye. The dye assignment to control and treated samples was reversed for the other three arrays to account for variation due to labeling efficiencies. The three slides were hybridized under identical conditions as described below. 48

Each sample was isolated and independently labeled prior to hybridization. Total RNA concentrations post purification were approximately 7-18 jag. In each cDNA reaction, 7-12 (ig of total RNA was added yielding cDNA concentrations of 2-7 (ig. Control versus heat shock samples were paired and had the same concentration of total RNA added to the cDNA reactions. Three samples from each treatment were labeled with Alexa Flour 555 and three with Alexa Fluor 647. The label incorporation of fluorescent dyes ranged from 20-81 bp/dye molecule for each dye, which is well within the recommended range of label incorporation (http://probes.invitrogen.com/ resources/calc/basedyeratio.html). To account for any variation due to dye incorporation, the experimental design included a dye swap. Some amount of error was attributed to the dye, but was not significant. Data from each of the six replicates was used in the statistical analysis. Statistical analysis indicated that 91 genes had significant transcriptional differences with a />value<0.01 and an estimated false discovery rate of four percent. Results are presented in Tables 2.2 and 2.3. The results are also displayed as a volcano plot, with up and down regulated genes indicated (Fig. 2.4).

W)cl, :

0.5 1 1.5 2.5 3 3.5 4 Log 2 Fold Change

Figure 2.4. Volcano plot of transcriptional differences in M. hyopneumoniae during heat shock conditions. Data represent individual gene responses plotted as Log2 fold change versus -Log p-value. Points above -Log p-value = 2 are significantly up- or down-regulated at pO.Ol. 49

Table 2.2. Heat shock up-regulated genes in M. hyopneumoniae. Gene Gene* Description P- q- Fold ID value value Change mhp032 CH Conserved Hypothetical 0.0015 0.0192 1.27 mhp034 parC Topoisomerase IV Subunit A 0.0001 0.0135 1.83 mhp057 CH Conserved Hypothetical 0.0034 0.0251 1.23 mhp060 # Signal Recognition Particle Protein 0.0086 0.0378 1.52 mhp064 CH Conserved Hypothetical 0.0007 0.0135 1.24 mhp069 CH Conserved Hypothetical 0.0061 0.0325 1.10 mhp071 UH Unique Hypothetical 0.0019 0.0224 1.39 mhp072 dnaK Chaperone Protein DnaK 0.0000 0.0064 13.75 mhp073 dnaJ Heat-Shock Protein 0.0002 0.0135 1.91 mhp078 lepA 30 kDa GTP-Binding Protein LepA 0.0040 0.0266 1.47 mhpl28 serS Seryl-tRNA Synthetase 0.0042 0.0267 1.85 mhpl44 CH Conserved Hypothetical 0.0080 0.0362 1.70 mhpl45 CH Possible D-Ribose-Binding Protein 0.0049 0.0286 1.75 mhpl47 rbsA Ribose Transport ATP-Binding Protein 0.0053 0.0302 1.68 mhpl48 CH Conserved Hypothetical 0.0014 0.0192 1.33 mhpl49 iolD Myo-Inositol Catabolism 0.0045 0.0273 2.13 mhpl50 CH Conserved Hypothetical 0.0011 0.0179 2.33 mhplSl CH Conserved Hypothetical 0.0013 0.0192 2.31 mhpl52 iolC Myo-Inositol Catabolism 0.0041 0.0266 2.04 mhpl53 mmsA Methylmalonate-Semialdehyde 0.0005 0.0135 2.53 Dehydrogenase mhpl71 CH Conserved Hypothetical 0.0054 0.0304 1.36 mhpl72 thdF Thiophene and Furan Oxidation Protein 0.0008 0.0135 1.38 mhp209 map Methionine Amino Peptidase 0.0077 0.0362 1.31 mhp211 rpsM 3 OS Ribosomal Protein S13 0.0015 0.0192 1.54 mhp212 rpsK 30S Ribosomal Protein SI 1 0.0005 0.0135 1.41 mhp214 50S Ribosomal Protein LI 7 0.0034 0.0251 1.32 mhp221 DeoB Phosphopentomutase 0.0034 0.0251 1.34 mhp224 CH Conserved Hypothetical 0.0069 0.0339 1.22 mhp254 CH Conserved Hypothetical 0.0080 0.0362 1.16 mhp271 CH Conserved Hypothetical, Putative 0.0005 0.0135 1.61 Adhesin mhp275 CH Hypothetical Protein PI02 0.0023 0.0224 1.40 mhp277 tpiA Triosephosphate Isomerase 0.0079 0.0362 1.40 mhp278 clpB ATP-Dependent Serine Proteinase, Heat 0.0003 0.0135 6.29 Shock mhp289 CH Conserved Hypothetical 0.0006 0.0135 1.40 * UH= unique hypothetical; CH=conserved hypothetical 50

Table 2.2. Continued. P- q- Fold Gene ID Gene* Description value value Change mhp308 UH Unique Hypothetical 0.0023 0.0224 1.38 mhp334 UH Unique Hypothetical 0.0024 0.0224 1.24 mhp356 UH Unique Hypothetical 0.0099 0.0405 1.21 mhp369 g/pF Glycerol Uptake Facilitator Protein 0.0072 0.0346 1.44 mhp370 g#: Glycerol kinase 0.0087 0.0378 1.43 mhp374 CH Conserved Hypothetical 0.0010 0.0173 1.41 mhp399 CH Conserved Hypothetical 0.0003 0.0135 2.64 mhp445 CH Conserved Hypothetical 0.0020 0.0224 1.55 mhp446 CH Conserved Hypothetical 0.0007 0.0135 1.81 mhp459 rplK Ribosomal Protein LI 1 0.0003 0.0135 2.01 mhp476 atpD ATP Synthase Beta Chain 0.0055 0.0307 1.28 mhp497 asnS Asparaginyl-tRNA Synthetase 0.0024 0.0224 1.57 mhpSOO oppC Oligopeptide Transport System 0.0072 0.0346 2.00 Permease mhpSOl oppB Oligopeptide Transport System 0.0049 0.0286 1.72 Permease mhp511 46kD Surface Antigen Precursor 0.0022 0.0224 1.70 mhp596 CH Conserved Hypothetical 0.0066 0.0336 1.30 * UH= unique hypothetical; CH=conserved hypothetical 51

Table 2.3. Heat shock down-regulated genes in M. hyopneumoniae. Fold ID Gene Description pvalue qvalue Change mhpOOS CH Conserved Hypothetical 0.0033 0.0251 1.53 mhpOO? UH Unique Hypothetical 0.0007 0.0135 1.41 mhp013 UH Unique Hypothetical 0.0065 0.0336 1.55 Ribosomal Large Subunit Pseudouridine 1.23 mhpO 15 rluC Synthase 0.0016 0.0200 Putative ABC Transporter ATP-Binding 1.03 mhp023 CH Protein 0.0041 0.0266 Glu-tRNA(Gln) Amidotransferase, 1.69 mhp030 ga# Subunit B 0.0032 0.0251 mhp038 CH Conserved hypothetical 0.0056 0.0307 1.19 mhpl35 rpsT 30S Ribosomal Protein S20 0.0058 0.0316 1.42 mhpl36 CH Conserved Hypothetical 0.0066 0.0336 1.32 mhpl76 CH Conserved Hypothetical 0.0031 0.0251 1.48 mhp235 CH Conserved Hypothetical 0.0093 0.0390 1.16 mhp240 CH Conserved Hypothetical 0.0043 0.0268 1.27 mhp241 glutamyl tRNA synthetase 0.0011 0.0179 1.24 mhp310 UH Unique Hypothetical 0.0027 0.0249 1.39 mhp330 mod Site-specific DNA-methyltransferase 0.0030 0.0251 1.13 mhp337 UH Unique Hypothetical 0.0020 0.0224 1.28 mhp353 CH conserved hypothetical 0.0002 0.0135 2.03 mhp373 CH Conserved Hypothetical 0.0039 0.0266 1.07 mhp393 CH Conserved Hypothetical 0.0022 0.0224 1.09 mhp397 proS Prolyl aminoacyl-tRNA synthetase 0.0037 0.0257 1.56 mhp410 metG Methionyl-tRNA Synthetase 0.0029 0.0249 1.31 mhp419 CH Conserved Hypothetical 0.0005 0.0135 1.28 mhp454 acpD Acyl Carrier Protein Phosphodiesterase 0.0092 0.0390 1.33 mhp465 CH Conserved Hypothetical 0.0015 0.0192 1.34 mhp466 CH Conserved Hypothetical 0.0007 0.0135 2.05 mhp467 hcrA ABC Transporter 0.0092 0.0390 1.34 mhp468 CH Conserved Hypothetical 0.0029 0.0249 1.26 mhp474 CH Conserved Hypothetical 0.0032 0.0251 1.32 mhp485 mgtE Mg2+ Ion Transporter 0.0081 0.0362 1.52 mhp486 CH Conserved Hypothetical 0.0067 0.0336 2.00 mhp509 UH Unique Hypothetical 0.0036 0.0257 1.38 mhp526 CH Conserved Hypothetical 0.0003 0.0135 1.40 mhp584 UH Conserved Hypothetical 0.0041 0.0266 1.65 mhp598 polA DNA polymerase I 0.0007 0.0135 1.47 * UH= unique hypothetical; CH=conserved hypothetical Table 2.3. Continued. ======^_===_==^=___= Fold ID Gene Description pvalue qvalue Change mhp619 CH Conserved Hypothetical 0.0087 0.0378 1.37 mhp620 CH Conserved Hypothetical 0.0001 0.0135 2.13 mhp621 CH Conserved Hypothetical 0.0006 0.0135 1.40 mhp658 UH Unique Hypothetical 0.0046 0.0278 1.17 mhp672 rplM 50S Ribosomal Subunit Protein L13 0.0099 0.0405 1.39 mhp674 CH Conserved Hypothetical 0.0022 0.0224 1.43 mhp697 CH Conserved Hypothetical 0.0068 0.0337 1.34 53

Validation of the microarray data Analysis indicated the four genes verified were significant using semi-quantitative PCR. The primers used in generating the RT-PCR products are given in Table 2.4. The results are shown in Figure 2.5.

Table 2.4. RT-PCR primers. Gene ID Primer Name Primer Sequence (5'-) rRNA RNA-F AGACGATGAT GTTT AGCGGGG RNA-R TGCTGCTCTTTGTAGTAGCCATG mhp072 072-F ATT AT CCGGT GG AACCTTCG 072-R AGTTCAACGTT AATCGGCCC mhp073 073-F T GGTTTTGGT GGCTC AC A AG 073-R TTTTTCCACGGCACTTTTTG mhp511 511-F A AGGT CTTTCACTTGCT GCG 511-R GATT GG ATT CTTT GACCGGC mhp620 620-F TTACTTACAACCAGCGCGAC 620-R AAACAGTT G ACGGCTTTTCG

p value -ÇeneL 1 3 5 7 9 11 i3 ' mhp16S 0.474

mhp072 Q 0.011

mhp073 <0.001

mhp511 <0.001

mhp620 <0.001 Figure 2.5. Semi-quantitative RT-PCR. Semi-quantitative RT-PCR was performed on the m RNA transcripts in control and heat shocked samples. Lanes 1, 3, 5, 7 and 9 are control samples from M. hyopneumoniae incubated in normal growth media at 37°C. Lanes 2, 4, 6, 8 and 10 are samples from M. hyopneumoniae incubated in media at 42°C. Lane 13 is the genomic DNA control. The gene designation is given on the left and the /j-value of the densitometry readings is given on the right. 54

Discussion Little is known about the mechanisms M. hyopneumoniae uses to colonize the respiratory epithelium and circumvent the host immune response. During infection, M. hyopneumoniae must encounter different environments as the numbers of organisms increase and the host responds accordingly with an influx of macrophages and neutrophils. If mycoplasmas are like other bacterial pathogens, they have mechanisms to respond to the changing in vivo environment, ensuring the organism's survival during active host immune responses. The host employs both innate and adaptive immune response strategies to protect it against pathogens. These environmental changes are generally thought of as stress conditions because they are often detrimental to the pathogen. For instance, cytokine release by host immune effector cells can lead to temperature fluxes in the host that may impact the organism's ability to maintain itself in the respiratory tract. Most bacteria have several genes that are induced to respond to this "heat shock". These genes usually encode proteins involved in protein folding and the prevention of protein aggregation, and in protein degradation of improperly folded proteins. The increased transcription of these genes during temperature shifts is indicative of an important role in survival. Denatured or nascent proteins do not function properly and would eventually lead to bacterial death. Fifty two-percent of the up-regulated genes (Table 2.2) have been assigned a function based on their sequence homology. These include functions involved in protein folding, metabolism, and translation. dnaK (mhp072), a member of the Escherichia coli hsp70 family, was identified as the most significantly up-regulated gene. Its role as a chaperone to protect proteins from improperly folding is well known. It also serves a regulatory function by sequestering a32 from the cytoplasm and presenting it to proteases. During heat shock, DnaK binds to denatured proteins thus releasing a32 to interact with RNA polymerase and up- regulate heat shock genes (37). Other heat shock-related genes,parC (mhp034), dnaJ (mhp073), and clpB (mhp278), were also significantly up-regulated in M. hyopneumoniae. These genes function to assist in protein folding, in degrading improperly folded proteins and in cell partitioning activities associated with DN A structure. Some of the genes appear to be members of opérons, e.g. mhpl44 through mhp 153 whose products are thought to be involved in carbohydrate metabolism, and mhp211 through mhp214 that code for ribosomal proteins and RpoA (mhp213), the a subunit of RNA polymerase. The structure of this operon is different than that found in E. colv, the M. hyopneumoniae operon lacks S4, the transcriptional regulator that binds to the leader sequence upstream of S13 (15). In other bacteria the ribosomal protein genes in this operon, the a operon, are regulated by translational coupling, but not rpoA. Thus, the mode of regulation for this operon in M. 55 hyopneumoniae is unknown, but it appears different than in other bacteria. mhp213 had a p- value of 0.013266 and just missed our arbitrary cutoff (p<0.01) for inclusion in Table 2.2. Forty-eight percent of the genes had unassigned functions and may represent important stress-related functions required for host colonization and persistence. One of the more interesting genes up-regulated during heat stress was ffh, the/ifty-/our kilodalton subunit homologue of the eukaryotic signal recognition particle. This is the first report of its regulation in response to heat stress and may represent a need to increase protein translocation in M. hyopneumoniae either to stabilize the membrane or maintain the integrity of membrane proteins, ffh is normally thought not to undergo regulation, but recent evidence suggests that in Gram positive bacteria it may be regulated in response to acid stress (11). In addition,^ was not regulated in M. pneumoniae during heat stress (32), suggesting that mycoplasmas may regulate different sets of genes during heat shock as a consequence of host adaptation. At least 41 genes were down-regulated in response to elevated temperatures. A larger proportion (73%) of these genes had unassigned functions compared to those up-regulated (48%). Many of these genes are involved in translation and DNA replication showing that temperature stress slows translation and replication, reducing energy needs, and slowing cellular physiology. These genes include rluC (mhp015), rpsT (mhpl35), and rplM (mhp672), as well, as DNA polymerase I (mhp598). This suggests that ribosomal proteins may have alternative functions since some are up-regulated in response to heat shock and some are down-regulated . Other genes fall into the classes of transporters, mgtE (mhp485) and bcrA (mhp467); carrier proteins, acpD (mhp454); and tRNA synthases, metG (mhp410), gatB (mhp030) andproS (mhp397). This is similar to other bacterial pathogens and suggests that reduction in energy requiring processes in the bacterial cell during heat stress is a fundamental property of pathogenesis and bacterial physiology. Little is known about the response of mycoplasmas to heat shock, particularly at the transcriptional level. Several studies have identified heat shock proteins or specific gene products in mycoplasmas (4, 25, 32). Weiner et al. analyzed the response of M. pneumoniae to heat shock conditions on a global scale using microarrays and identified 47 up-regulated genes (32). No other studies have taken a global approach to identifying heat shock related genes in mycoplasmas, however. A report that Hsp60 was present in M. hyopneumoniae (25) could not be substantiated by the genome sequence of strain 232, the strain used in these studies (18). In fact, both GroEL and GroES are missing in the M. hyopneumoniae genome. The experimental design to use a temperature shift of 37 to 42°C following equilibration at 32°C was based on several reasons. A temperature of 32°C is the usual starting 56 temperature for heat shock studies and was the starting temperature used by Weiner et al. for their studies with M. pneumoniae (32). Unlike M. pneumoniae, however, M. hyopneumoniae does not grow at measurable levels at 32°C. Mycoplasma pneumoniae was able to replicate albeit slowly at 32°C and thus, they were able to perform their studies starting at 32°C after an extended (144 h) culture period. In the reported studies, the 32°C temperature was used to synchronize the cultures prior to shifting to the experimental temperatures. A two hour time point was chosen because it represents an approximate generation time at 37°C. The temperature differential of 37 to 42°C, which was chosen to reproduce the effects of elevated temperatures in the infected host, was maintained for 30 min to allow the mRNA transcripts of the slow growing organisms to come to steady state. While a temperature shift of 5 degrees is unusual for heat shock studies, M. hyopneumoniae is host adapted and cannot survive for any length of time in the environment. Thus, it would normally be exposed only to temperature fluctuations within the host. These studies place maximum stress on M. hyopneumoniae in the form of an upper temperature limit of 42°C to maximize transcriptional variation and increase the probability of identification of regulated genes. A temperature higher than 42°C, e.g. the 43°C used by Weiner et al. (32), would not be reached within the host. These data indicate that 91 M. hyopneumoniae genes are either up-regulated or down-regulated in response a 37 to 42°C temperature shift at a significance level of /?<0.01 (Tables 2.2 and 2.3). It suggests that M. hyopneumoniae responds to temperature changes by altering transcriptional activities of specific genes through unknown mechanisms. It also suggests that mycoplasmas respond differently depending on the species and possibly host. A large percentage of heat shock responsive genes (-60%) have no functional assignment. Future studies will be required to identify either the roles of these gene products in physiology and pathogenesis. How these genes are controlled in the absence of typical heat shock regulators (i.e. rpoH) is not known, but perhaps mechanisms operative in related Gram positive bacteria might be involved. However, now that these genes have been identified, studies of upstream sequences may provide insight into regulatory mechanisms despite the lack of genetic tools in M. hyopneumoniae. These studies focused on transcriptional changes, but other mechanisms involving ribosome conformation (30) or posttranslational processing (5) might also be involved.

Acknowledgements We would like to thank Hui-Hsien Chou for assistance with primer design. He wrote the scripts used with Primer3 software. We also thank Nancy Upchurch and Barb Erickson for 57 assistance with mycoplasma cultures. Monica Perez contributed to the construction of the microarray. Mike Carruthers assisted with spot finding. Josh Pitzer assisted in the RT-PCR reactions validating the microarray results. Funding for this project was provided in part by the Iowa Healthy Livestock Advisory Council.

References 1. Borchsenius, S. N., E. V. Budantseva, and M. S. Vonsky. 1990. The heat-shock proteins of Acholeplasma laidlawii. Zentralbl. Bakeriol. Suppl. 20:657-658. 2. Chambaud, I., R. Heilig, S. Ferris, V. Barbe, D. Samson, F. Calisson, I. Moszer, K. Dybvig, H. Wroblewski, A. Viari, E. P. Rocha, and A. Blanchard. 2001. The complete genome sequence of the murine respiratory pathogen Mycoplasma pulmonis. Nucleic Acids Res 29:2145-2153. 3. Dascher, C. C., and J. Maniloff. 1992. Heat shock response, p. 349-354. In J. Maniloff, R. N. McElhaney, L. R. Finch, and J. B. Baseman (ed.), Mycoplasmas: molecular biology and pathogenesis. American Society for Microbiology, Washington, D C. 4. Dascher, C. C., S. K. Poddar, and J. Maniloff. 1990. Heat shock response in mycoplasmas, genome-limited organisms. J. Bacteriol. 172:1823-1827. 5. Djordjevic, S. P., S. J. Cordwell, M. A. Djordjevic, J. Wilton, and F. C. Minion. 2004. Proteolytic processing of the Mycoplasma hyopneumoniae ci Hum adhesin. Infect. Immun. 72:2791-2802. 6. Dudley, A. M., J. Aach, M. A. Steffen, and G. M. Church. 2002. Measuring absolute expression with microarrays with a calibrated reference sample and an extended signal intensity range. Proc. Natl. Acad. Sci. USA 99:7554-7559. 7. Dudoit, S., Y. H. Yang, M. J. Callow, and T. P. Speed. 2000. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. http://www.stat.Berkeley.edu/users/terry/zarray/Html/papersindex.html. 8. Friis, N. F. 1975. Some recommendations concerning primary isolation of Mycoplasma hyopneumoniae and Mycoplasma flocculare, a survey. Nord. Veterinaermed. 27:337-339. 9. Gottesman, S., and M. R. Maurizi. 1992. Regulation by proteolysis: energy- dependent proteases and their targets. Microbiol. Rev. 56:592-621. 10. Gross, C. A. 1996. Function and regulation of the heat shock proteins. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli 58

and Salmonella: cellular and molecular biology. American Society for Microbiology, Washington, D C. 11. Gutierrez, J. A., P. J. Crowley, D. G. Cvitkovitch, L. J. Brady, I. R. Hamilton, J. D. Hillman, and A. S. Bleiweis. 1999. Streptococcus mutans ffh, a gene encoding a homologue of the 54 kDa subunit of the signal recognition particle, is involved in resistance to acid stress. Microbiology - UK 145:357-366. 12. Himmelreich, R., H. Hilbert, H. Plagens, E. Pirkl, B.-C. Li, and R. Herrmann. 1996. Complete sequence analysis of the genome of the bacterium Mycoplasma pneumoniae. Nucleic Acids Res. 24:4420-4449. 13. Hsu, T., S. Artiushin, and F. C. Minion. 1997. Cloning and functional analysis of the P97 swine cilium adhesin gene of Mycoplasma hyopneumoniae. J. Bacteriol. 179:1317-1323. 14. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infec. Immun. 66:4762-4766. 15. Keener, J., and M. Nomura. 1996. Regulation of ribosome synthesis, p. 1417-1431. In F. C. Neidhardt (ed.), Esherichia coli and Salmonella cellular and molecular biology. ASM Press, Washington, D C. 16. Mare, C. J., and W. P. Switzer. 1965. New Species: Mycoplasma hyopneumoniae, a causative agent of virus pig pneumonia. Vet. Med. Small Anim. Clin. 60:841-846. 17. Minion, F. C., C. Adams, and T. Hsu. 2000. R1 region of P97 mediates adherence of Mycoplasma hyopneumoniae to swine cilia. Infect Immun 68:3056-3060. 18. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 19. Muto, A., and C. Ushida. 2002. Transcription and translation, p. 323-345. In S. Razin and R. Herrmann (ed.), Molecular biology and pathogenicity of mycoplasmas. Plenum Press, New York. 20. Narberhaus, F. 1999. Negative regulation of bacterial heat shcok genes. Mol Microbiol 31:1-8. 21. Neidhardt, F. C., and R. A. VanBogelen. 1987. Heat shock response. In F. C. Neidhardt, J. L. Ingraham, K. B. Lo, B. Magasanik, M. Schaechter, and E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D. C. 22. Papazisi, L., T. S. Gorton, G. Kutish, P. F. Markham, G. F. Browning, D. K. Nguyen, S. Swartzell, A. Madan, G. Mahairas, and S. J. Geary. 2003. The complete genome 59

sequence of the avian pathogen Mycoplasma gallisepticum strain R(low). Microbiology 149:2307-2316. 23. Raghavachari, N., and W. E. Fahl. 2002. Targeted gene delivery to skin cells in vivo: a comparative study of liposomes and polymers as delivery vehicles. J. Pharm. Sci. 91:615-622. 24. Ross, R. F. 1993. Mycoplasma-animal pathogens, p. 69-109. In I. Kahane and A. Adoni (ed.), Rapid diagnosis of mycoplasmas. Plenum Press, New York. 25. Scherm, B., G. F. Gerlach, and M. Runge. 2002. Analysis of heat shock protein 60 encoding genes of mycoplasmas and investigations concerning their role in immunity and infection. Vet. Microbiol. 89:141-150. 26. Skaletsky, R. S. 2000. Primer3 on the WWW for general users and for biologist programmers, p. 365-386. In S. Krawetz and S. Misener (ed.), Bioinformatics methods and protocols: methods in molecular biology. Humana Press, Totowa, N.J. 27. Storey, J. D., and R. Tibshirani. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA 100:9440-9445. 28. Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 29. Thanawongnuwech, R., B. Thacker, P. Halbur, and E. L. Thacker. 2004. Increased production of proinflammatory cytokines following infection with porcine reproductive and respiratory syndrome virus and Mycoplasma hyopneumoniae. Clin Diagn Lab Immunol 11:901-908. 30. VanBogelen, R. A., and F. C. Neidhardt. 1990. Ribosomes as sensors of heat and cold shock in Escherichia coli. Proc. Natl. Acad. Sci. USA 87:5589-5593. 31. Versteeg, S., A. Escher, A. Wende, T. Wiegert, and W. Schumann. 2003. Regulation of the Bacillus subtilis heat shock gene htpG is under positive control. J. Bacteriol. 185:466-474. 32. Weiner III, J., C.-U. Zimmerman, H. W. H. Gohlmann, and R. Herrmann. 2003. Transcription profiles of the bacterium Mycoplasma pneumoniae grown at different temperatures. Nucleic Acids Res. 31:6306-6320. 33. Westberg, J., A. Persson, A. Holmberg, A. Goesmann, J. Lundegerb, K.-E. Johansson, B. Pettersson, and M. Uhlaen. 2004. The genome sequence of Mycoplasma mycoides subsp. mycoides SC type strain PG1T, the causative agent of contagious bovine pleuropneumonia (CBPP). Genome Res. 14:221-227. 60

34. Wolfinger, R. D., G. Gibson, E. D. Wolfinger, L. Bennett, H. Hamadeh, P. Bushel, C. Afshari, and R. S. Paules. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comp. Biol. 8:625-637. 35. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed. 2002. Normalization for cDNA microarray data: a robust composite method for adressing single and multiple slide systematic variation. Nucleic Acids Res. 30:el5. 36. Yogev, D., G. F. Browning, and K. S. Wise. 2002. Genetic mechanisms of surface variation, p. 417-443. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 37. Yura, T., M. Kanemori, and M. Morita. 2000. The heat shock response: regulation and function. In G. Storz and R. Hengge-Aronis (ed.), Bacterial stress response. ASM Press, Washington, D C. 38. Zhang, Q., T. F. Young, and R. F. Ross. 1995. Identification and characterization of a Mycoplasma hyopneumoniae adhesin. Infect Immun 63:1013-1019. 61

CHAPTER 3. TRANSCRIPTIONAL PROFILING OF MYCOPLASMA HYOPNEUMONIAE DURING IRON DEPLETION USING MICROARRAYS

A paper to be submitted to Infection and Immunity

Melissa L. Madsen,1 Dan Nettleton,2 Eileen L. Thacker,1 Robert Edwards,3 and F. Chris Minion1

Departments of Veterinary Microbiology and Preventive Medicine1 and Statistics2, Iowa State University, Ames, IA; San Diego State University, San Diego, CA3

Abstract Mycoplasma hyopneumoniae, the causative agent of swine enzootic pneumonia and a major component of the porcine respiratory disease complex, continues to confound swine producers despite control programs worldwide. The disease is chronic and self-limiting, but the host is subject to immunopathologic changes that potentiate respiratory disease associated with other pathogens. The response of M hyopneumoniae to environmental stress is of interest because of the relevance to virulence mechanisms in other bacterial pathogens. One of these stressors, iron deprivation, is a prominent feature of the host innate immune response, and most certainly impacts growth of mycoplasmas in vivo. To study this, microarray technology was applied to the transcriptome analysis of M hyopneumoniae during iron deprivation. An array consisting of 632 of the 698 open reading frames in the genome was used to compare the mRNA isolated from organisms grown under normal laboratory conditions with those subjected to iron deprivation with the chelator 2,2'- dipyridyl. We employed a unique hexamer primer set for generating cDNA from only mRNA species. Our analysis identified 27 genes that were either up- or down-regulated in response to low iron growth conditions (p <0.01) with an estimated false discovery rate below ten percent. Among these were transport proteins, enzymes involved in energy metabolism, and components of the translation process. A proportion of the identified genes (10 of 27) had no 62 assigned function. These studies indicate that M. hyopneumoniae can respond to changes in environmental conditions, but the mechanism employed remains unknown.

Introduction Mycoplasma hyopneumoniae is the causative agent of swine enzootic pneumonia (23) and contributes significantly to porcine respiratory disease complex (29). Despite aggressive vaccination programs, M. hyopneumoniae continues to induce lesions in as many as 80% of the pig population (24). It colonizes the respiratory epithelia where it attaches exclusively to the cilia (17) causing ciliostasis and cell death (7, 14). Like all mycoplasmas, M. hyopneumoniae must acquire essential nutrients from its host because it lacks biosynthetic pathways due to the limited coding capacity of its genome (18). One of these nutrients, iron, is an essential metal for almost all living systems. The one known exception is the agent of Lyme disease, Borrelia burgdorferi (20). Iron serves as a cofactor or as a prosthetic group for essential enzymes that are involved in basic cellular functions (19). Free iron, however, does not exist in useful concentrations under neutral pH conditions since it is readily oxidized forming the hydroxide Fe(OFI)^, which is soluble only at 1.4 x 10"9 M at pH 7 (2). Free iron catalyzes the formation of free oxygen radicals and as such is highly toxic to cells. Therefore, iron is normally bound to specific proteins in the host such as transferrin and lactoferrin to prevent its precipitation in aqueous systems and to protect the cell from damaging UV light and lipid peroxidation reactions. In this way, iron is both sequestered for use by the host and prevents its use by invading pathogens. Most bacteria require iron concentrations in the range of 10"6 to 10"7 M for their metabolic processes (33). Therefore, bacteria have evolved mechanisms to compete with the host for available iron. These mechanisms usually involve multiple components that are highly coordinated to achieve their goal. In most bacteria iron binding molecules, siderophores, are secreted to compete with the host transferrin or lactoferrin for iron, and then the iron-siderophore complex is taken up by specific surface receptors and porin-like transporters (26). Since, considerable energy investment is needed by the bacteria to express competitive iron acquisition systems, their expression is often tightly regulated by iron availability (5). Unlike other bacteria, mycoplasmas have minimal gene coding capacity (12) and lack cell walls, so it is thought that their interactions with the host are less complex, including the acquisition of essential nutrients. Pathogenic mycoplasmas pose interesting questions concerning iron metabolism because they lack a host-free lifestyle, they have single limiting membranes and few studies have been performed on iron requirements. While it has not been 63 unequivocally shown that iron is essential for growth, Bauminger et al. showed iron was stored in Mycoplasma capricolum (4) and Try on and Baseman have shown that Mycoplasma pneumoniae can take up human lactoferrin (30). Mycoplasmas also contain enzymes that have been shown in other systems to contain iron such as dehydrogenases. Whether mycoplasmas contain complex iron acquisition mechanisms is not known. Iron deprivation in other bacterial systems triggers not only the genes required for its acquisition but also in many pathogens is a signal for controlling virulence factor expression (13, 22, 26, 28). Thus, virulence factors can sometimes be identified by their response to low iron growth conditions. These studies show that iron deprivation in vitro leads to transcriptional changes in M. hyopneumoniae. The iron chelator 2,2'-dipyridyl was used to chelate iron in the serum rich media, and the transcriptional responses to iron sequestration were measured using microarrays. Genes that were both up-regulated and down-regulated were identified in this study.

Materials and Methods Mycoplasma strains and culture conditions Pathogenic M. hyopneumoniae strain 232, a derivative of strain 11, was used in this study (16). Cultures used were passed fewer than 15 times in vitro in Friis media as previously described (11). Twelve 250 ml flasks containing 125 ml cultures were grown at 37°C to early-log phase as determined by medium color change and optical density. To six flasks, 2,2'-dipyridyl (Sigma Chemical Co., St. Louis, Mo.) was added to a final concentration of 1 mg/ml. Six flasks were left untreated, and all flasks were incubated for 2 h at 37°C. Cells were pelleted by centrifugation at 24,000 x g, and 500 jal of RNAlater (Ambion, Inc. Austin, Tex.) was added to pellet. Pellets were stored at -70°C until the total RNA was isolated. Mycoplasma gallisepticum (ATCC19610) and Mycoplasma pulmonis UAB6510 (6) were grown in modified Hayflick's media as described except that the media was supplemented with 25 pg/ml cefoperazone (Pfizer, New York) instead of thallium acetate and benzylpenicillin (10). To assess the effect that various iron chelators would have on mycoplasmas growth, Hayflick's media was supplemented with 2,2'-dipyridyl (0.2 and 1 mg/ml), sodium citrate (1, 2, 5 and 7.5 mM) and desferoxamine (desferyl) mesylate (0.2 and 1 mg/ml). Growth was determined by plating 10-fold serial dilutions of these broth cultures at different time points onto Hayflick's agar media lacking chelators. Colony forming units were determined after incubation at 37°C for 4-5 days. 64

Microarray The M. hyopneumoniae microarray consists of polymerase chain reaction (PCR) products (probes) spotted to Corning (Corning, Inc., Big Flats, N.Y.) UltraGAPS glass substrates. Ninety-one percent (632/698) open reading frames (ORFs) are represented on the array as PCR products of 125-350 bp in length. Each product is a unique sequence even within paralogous families as described by Minion et al. (18). The 66 missing ORFs are due to an inability to design suitable primers (12 ORFs) or failed PCR reactions due to incorrect product size, multiple bands or no band (54 ORFs). Only ORFs greater than 125 bp are represented on the array and no tRNAs or rRNas were included. The primer design, array construction and validation have been described (Chapter 2).

Experimental design Six independent RNA samples from iron-depleted cells were paired with six independent RNA samples from control cells for hybridization on six two-color microarrays. The six arrays were spotted on a total of three slides. Each slide was divided into two regions (upper and lower), and each region contained the full array of spots. Each of the 632 sequences in the array was represented by three replicate spots. For three of the six arrays, the control sample was labeled with Alexa 555 dye and compared to the iron-depleted sample labeled with Alexa 647 dye (Molecular Probes, Inc., Eugene, Ore.). The dye assignment to control and treated samples was reversed for the other three arrays. The three slides were hybridized under identical conditions as described below.

RNA isolation RNA was isolated from frozen cell pellets using the Versagene™ RNA Purification System (Centra Systems, Minneapolis, Minn.) according to the manufacturer's protocol. The optional step of DNase treatment was performed on column according to the manufacturer's recommendation. With a cut-off of 150 bp, 5S rRNA and tRNAs were removed from the samples, limiting interference in downstream manipulations. Samples were quantified and checked for purity using the Nanodrop® ND-1000 Spectrophotometer (Nanodrop, Wilmington, Del.). If necessary, samples were concentrated using Microcon YM-30 micro- concentrators (Millipore, Billerica, Mass.) for optimal cDNA generation.

Target generation and hybridization Targets were generated from total RNA extracted from cell pellets as described above. Fluorescently labeled cDNA targets were generated and purified using the SuperScript™ 65

Indirect cDNA Labeling System (Invitrogen Corp., Carlsbad, Calif.) with a set of 129 ORF- specific six-mer oligonucleotide primers designed as previously described (Chapter 2). Targets were labeled with Alexa Fluor™ 555 Reactive Dye or Alexa Fluor™ 647 Reactive Dye (Molecular Probes, Inc.). Following purification of the labeled cDNA, samples were dried in a vacuum centrifuge and then resuspended in 10 |il Pronto! cDNA/long oligo hybridization solution (Coming). Targets were denatured at 95°C for 5 min and centrifuged at 13,000 rpm for 2 min at room temperature. Targets from a control and iron-deprived culture were then combined, pipetted to an array, and covered with a 22x22 mm HybriSlip™ (Schleicher & Schuell, Keene, N. H.). Slides were placed in a Coming hybridization chamber and incubated in a 42°C water bath for 12-16 h. Slides were washed according to Coming's UltraGAPS protocol and dried by centrifugation.

Data acquisition and normalization Each array was scanned with each dye channel using a ScanArray Express laser scanner (Applied BioSystems, Inc., Foster City, Calif.) at least three times under varying laser power and PMT gain settings to increase the dynamic range of expression measurement (8). Images were quantified using the softWorRx Tracker analysis software package (Applied Precision, Inc., Issaquah, Wash.). Spot-specific mean signals were corrected for local background by subtracting spot-specific median background intensities. The natural logarithms of the background-corrected signals from a single scan were adjusted by an additive constant so that all scans of the same array-by-dye combination would have a common median. The median of these adjusted-log-background-corrected signals across multiple scans was then computed for each spot to obtain one value for each combination of spot, array, and dye channel. These data for the two dye channels on any given array were normalized using LOWE SS normalization to adjust for intensity-dependent dye bias (9, 35). Following LOWESS adjustment, the data from each channel were adjusted by an additive constant so that the median for any combination of array and dye would be the same for all array-by-dye combinations. The normalized values for triplicate spots were averaged within each array to produce one normalized measure of expression for each of the 632 probe sequences and each of the 12 RNA samples.

Data analysis A separate mixed linear model analysis was conducted for each probe sequence using the normalized data (34). Each mixed model included fixed effects for treatment (iron depletion versus control), slide region (upper versus lower), and dye (Alexa 555 versus 647) as well as 66 random effects for slide and slide-by-region interaction. A /-test for differential expression across treatments was conducted for each probe as part of our mixed linear model analyses. The 632 p-values from these /-tests were converted to q-values using the method of Storey and Tibshirani (27). These ^-values can be used to obtain approximate control of the False Discovery Rate at a specified value. For example, declaring probes with ^-values less than or equal to 0.05 to be differentially expressed produces a list significant results for which the False Discovery Rate is estimated to be approximately 5%. Along with ^-values, estimates of fold-change were computed for each probe by taking the inverse natural log of the mean treatment difference estimated as part of our mixed linear model analyses.

Confirmation of microarray data To confirm significant transcriptional differences between genes, semi-quantitative RT- PCR analysis was performed on a subset of genes that showed significant up- or down- regulation in the microarray analysis. The PCR was performed in 96 well plates in a MJ Research Dyad Thermal Cycler (Bio-Rad Laboratories, Inc., Waltham, Mass.) in 20 jxl reactions. The six cDNA preparations for each control and iron depleted sample was used with the following conditions: IX PCR Buffer; 2 mM MgCla; 1.0 mM each dATP, dCTP, dTTP and dGTP (Fisher Bioreagents, Fairlawn, NJ); 5 pmol of each forward and reverse primer; 2.5 units Taq polymerase (New England Biolabs, Beverly, Mass.); and 2 p.1 cDNA from either control or iron depleted samples diluted 1:50. The thermal cycling conditions were as follows: denaturation at 95°C for 5 min; and then 25 cycles of denaturation at 94°C for 1 min, anneal at 50°C for 1 min, and elongation at 72°C for 30 sec. A final elongation step at 72°C for 10 min was performed. The reactions were mixed with 4 |il 6X loading dye [10 mM Tris-HCl (pH 7.6), 0.03% bromophenol blue, 0.03% xylene cyanol FF, 60% glycerol, 60 mM ethylenediamine tetraacetic acid], and 10 jil were analyzed by 1.5% agarose gel electrophoresis. Gels were stained with 0.5 (J-g/ml ethidium bromide to confirm that a single product of correct size was present. The gel was visualized and digitized, and each band density was measured with FlourChem 8000 version 2.0 software (Alpha Innotech Corp., San Leandro, Calif). Analysis of variance was performed to determine differences between band density values of iron-depleted vs control samples. Band densities were background corrected by subtracting the background density within the reaction lane from the sample band density. Estimated differences were considered significant if the p-value of a /-test was <0.05. 67

Results Iron depletion studies Preliminary growth studies were performed with M. gallisepticum and M. pulmonis to determine optimal concentrations of 2,2'-dipyridyl, sodium citrate or desferyl mesylate to use in studies with M. hyopneumoniae. The rationale for these studies was that M hyopneumoniae growth is difficult to quantitate since it grows poorly on agar surfaces giving rise to pin point size colonies only under undefined ideal growth conditions. Both M. gallisepticum and M. pulmonis grow well in comparison to M. hyopneumoniae, readily form colonies on agar surfaces, and like M. hyopneumoniae, have a host-adapted life style. It was thought that they would react in a similar fashion to M. hyopneumoniae under iron limiting conditions since all three species grow in an environment replete with transferrin and lactoferrin. Since little is known about iron sequestering mechanisms in mycoplasmas, three chelators with different iron binding mechanisms and affinities were used in the preliminary studies. Our goal was to identify a chelator that could slow or stop the growth of M gallisepticum and M. pulmonis while growing in serum-rich medium. Figure 3.1 shows the results of supplementing Hayflick's growth media with 2,2'-dipyridyl at 0.2 and 1 mg/ml. This demonstrates that 2,2'-dipyridyl consistently slowed growth but only at the highest concentration (1 mg/ml). Neither sodium citrate nor desferyl mesylate had an effect on growth at any concentration tested (data not shown). Assuming the same effect on M hyopneumoniae growth, subsequent iron-limiting studies with M. hyopneumoniae were performed with 2,2'-dipyridyl supplemented at a concentration of 1 mg/ml.

Microarray studies We assessed the transcriptional profile of M. hyopneumoniae using microarrays containing 632 of the 698 ORFs in the genome (18)(also see Chapter 2). These arrays were constructed by spotting PCR products of 150-350 bp in size to glass substrates. Each of the PCR products were unique in sequence, even those of the paralog families (18). In a few instances, unique primers could not be designed or primers failed to produce the proper PCR products during reactions. These 66 ORFs were eliminated from our analysis. 68

M. pulmonis M. gallisepticum

3 1x10" oi c ••p 1x10'

1x10

0 10 20 0 20 30 Hours Figure 3.1. Effect of 2,2'-dipyridyl on growth of M. pulmonis and M gallisepticum. Overnight cultures of both mycoplasmas were diluted 1:20 in fresh culture media containing 100 pg/ml (open circles), 200 pg/ml (closed circles) and 1 mg/ml (open squares) and incubated at 37°C for the indicated time. Samples were removed, serially diluted and plated on mycoplasma agar lacking chelator for colony counts at the times indicated.

To assess the effect of 2,2'-dipyridyl on M. hyopneumoniae global transcriptional profiles, broth-grown cultures were supplemented with the chelator and incubated for 2 h at 37°C. Following centrifugation of the cells, total RNA was extracted from six replicate control and six supplemented cultures, and cDNA reactions were performed independently on each of the twelve RNA preparations. Total RNA concentrations post purification were approximately 15-20 pg total RNA, and cDNA concentrations ranged from 2-6 |ig total cDNA. Control and iron-depleted samples were paired and equal concentrations of total RNA were added to cDNA reactions. Three samples from each treatment were labeled with Alexa Flour 555 and three with Alexa Fluor 647. The label incorporation of fluorescent dyes ranged from 20-65 base pairs/dye molecule, which is well within the recommended range of label incorporation (http://probes.invitrogen.com/ resources/calc/basedyeratio.html). To account for variation due to dye incorporation, the experimental design included a dye swap with the three remaining sample pairs. One control and the paired 2,2'-dipyridyl supplemented labeled cDNA samples were mixed and hybridized to a single M. hyopneumoniae array overnight. Following scanning, the spot finding was performed to obtain signal intensity values at the two wavelengths representing the two dye adsorption maxima. Data from each of the six replicates was used in the statistical analysis. Some amount of error was attributed to the dye, but the effect was not significant in the experimental model as 69 analyzed by ANOVA. Statistical analysis indicated that 27 genes had significant transcriptional differences with a p-value <0.01. The results are presented in Table 3.1 and in Figure 3.2. Nine genes demonstrated a significant increase in transcriptional activity under iron limiting conditions as compared to control cells grown under normal in vitro conditions. Eighteen genes, however, were down regulated under these conditions. Figure 3.2 displays the results of the 632 genes as a volcano plot, with up and down regulated genes marked.

4.5

4

3.5

U3 3

Log2 Fold Change

Figure 3.2. Volcano plot of transcriptional differences in M. hyopneumoniae during low iron growth conditions. Data represent individual gene responses plotted as Log; fold change vs. -Log p-value. Points above -Log 7;-value = 2 are significantly up or down regulated at /?<0.01.

Validation of microarray data Semi-quantitative RT-PCR was performed on a subset of genes to validate the microarray results. RT-PCR reactions were performed using the 12 RNA samples from the control and low iron studies. The reactions contained equal RNA concentrations and were performed at the same time to reduce experimental error. Reaction products were separated on an agarose gel, stained, and the band intensities quantitated. The twelve experimental replicates for each gene (six control and six low iron samples) were analyzed for significance. The RT-PCR product signal intensities for genes mhpl40, mhpl51 and mhp558 from low iron conditions was significantly different than control at p <0.01. Genes mhp275 and mhp639 were not significantly different at p = 0.01. The primers used for the RT-PCR reactions are given in Table 3.2. The results are shown in Figure 3.3. 70

Table 3.1. Differentially expressed genes of M. hyopneumoniae during growth under iron limiting conditions. Gene ID Gene Description jD-value g-value FC Upregulated mhpl40 UH Unique Hypothetical 0.0083 0.0957 1.29 mhpl51 CH Conserved Hypothetical 0.0069 0.0957 2.35 mhpl52 WO Myo-Inositol Catabolism 0.0083 0.0957 2.15 mhp275 CH Hypothetical Protein, PI02 Paralog 0.0076 0.0957 1.68 mhp317 gW Glycerophosphoryl Diester 0.0024 0.0957 1.42 Phosphodiesterase mhp319 mglA ABC Transporter ATP Binding Subunit 0.0013 0.0633 0.98 mhp505 ackA Acetate Kinase 0.0049 0.0957 1.95 mhp510 UH Unique Hypothetical 0.0041 0.0957 1.34 mhp558 potB Spermidine/Putrescine Transport Permease 0.0007 0.0588 1.33 Downregulated mhp030 gatB Glu-tRNA Amidotransferase, Subunit B 0.0063 0.0957 1.63 mhpOSl UH Unique Hypothetical 0.0055 0.0957 1.23 mhp083 fusA GTP-Binding Chain Elongation Factor EF- 0.0012 0.0633 1.63 G mhp085 rpsL 30S Ribosomal Protein S12 0.0002 0.0347 1.34 mhp087 CH Conserved Hypothetical 0.0031 0.0957 1.35 mhp 117 hpt Hypoxanthine Phosphoribosyl Transferase 0.0097 0.0957 1.53 mhp 167 oppD Oligopeptide Transport System Permease 0.0042 0.0957 2.41 mhp168 oppC Oligopeptide Transport System Permease 0.0062 0.0957 2.12 mhp169 oppB Oligopeptide Transport System Permease 0.0050 0.0957 1.50 mhp 176 CH Conserved Hypothetical 0.0009 0.0631 1.50 mhp190 rpL2 50S Ribosomal Protein L2 0.0100 0.0957 1.76 mhp337 UH Unique Hypothetical 0.0062 0.0957 1.17 mhp411 CH Conserved Hypothetical 0.0079 0.0957 1.13 mhp450 metK S-Adenosylmethionine Synthetase 0.0100 0.0957 1.36 mhp637 rplL 50S Ribosomal Protein L7/L12 0.0028 0.0957 1.46 mhp639 CH Conserved Hypothetical 0.0001 0.0333 1.60 mhp651 ushA 5'-Nucleotidase 0.0003 0.0347 2.46 mhp666 CH Conserved Hypothetical 0.0047 0.0957 1.31 * UH= unique hypothetical; CH=conserved hypothetical; FC = Fold Change 71

Table 3.2. RT-PCR primer sequences. Gene ID Primer Name Primer Sequence (5'-) rRNA RNA-F AGACGATGATGTTTAGCGGGG RNA-R T GCTGCTCTTT GT AGT AGCCAT G mhp182 182-F T A AAA ACCGTGATT G AGGGC 182-R GCT GTTCAAATGCTT GT CCC mhp275 275-F T GAT ATTGTT GATCTAGTCGACGG 275-R GGGCTTAC ACCTT CTTT GGC mhp374 374-F TTTGGTTATT CTAATT GGCG 374-R AACCACTT GGTTTT AGTTCGAC mhp558 558-F GAAGCCGGGACTG ATTTAGG 558-R AGGGTCACCAAAACAAGCG mhp639 639-F AGGTCATCGAAAACGGGC 639-R ATAGTTTCAAGGGCTTGGCG mhp677 677-F CC AG AAT AT GTTT GGT GGGG 677-R TT GATC AACTCGGG ACATCG

mm; 1 3 11 mnuiG U L'91

miplc-1

mnc275

mnp <0.GC1 mhp:JJ l'PM'f'M'M'I'l r 51

Figure 3.3. Semi-quantitative RT-PCR. Semi-quantitative RT-PCR was performed on the mRNA transcripts in control and iron depleted samples. Lanes 1, 3 ,5, 7 and 9 are control samples from M. hyopneumoniae incubated in normal growth media. Lanes 2, 4, 6, 8 and 10 are samples from M. hyopneumoniae incubated in media containing 2,2'-dipyridyl. Lane 13 is the genomic DNA control. The gene designation is given on the left and the p- value of the densitometry readings is given on the left. 72

Discussion Mycoplasma hyopneumoniae is found worldwide and is transmitted via aerosol from pig to pig. In order to survive, M. hyopneumoniae must establish itself in the upper respiratory tract of its host in the face of significant innate immune defense mechanisms. One of these, iron sequestration, operates by preventing access of the pathogen to the essential metal by binding available iron to transferrin, lactoferrin and lactotransferrin (21). The host uses these molecules to prevent free iron from undergoing oxidation reactions that are detrimental and by providing a pool for host purposes. Iron is vital to the survival of almost all known organisms because of its requirement as a cofactor or a prosthetic group for essential enzymes (26). By sequestering iron, the host is able to limit the growth of pathogens, providing a level of security. Iron exists primarily in the oxidized ferric form with a solubility constant of 1.4 x 10"9 M (21). Transferrin, lactoferrin and lactotransferrin have iron binding constants in the range of~1020(l, 3), and transferrin is never fully saturated under normal conditions, giving the host additional iron sequestration capabilities during bacterial infections. Bacteria, however, have evolved mechanisms for iron acquisition that are tightly controlled by the level and availability of iron in the environment (21). These mechanisms involve transcriptional regulators such as Fur and cognate DNA binding sites (Fur binding sites) at genes involved in iron uptake (32). The function of these genes involve binding of host transferrin or lactoferrin and its reduction and uptake, or siderophore production, secretion, and transport systems (21). The Fur regulated genes are also thought to contribute to virulence in many pathogens, because the existence of multiple iron acquisition systems allow for alternate means of iron acquisition and virulence under changing environmental conditions within the host during disease. The regulation of iron acquisition genes linked with virulence factors allows for cell invasion, and possibly cell death, increasing the iron available for uptake by the bacteria (25). Sequence analysis of the M. hyopneumoniae genome has not identified any putative/wr genes or their regulators (18), so it was of interest to determine if M. hyopneumoniae would respond to an environment in which iron was sequestered by a chelator other than transferrin or lactoferrin. The goal of this study was to determine the steady-state levels of individual mRNA species on a global level during growth of M. hyopneumoniae under normal laboratory conditions and compare those levels with organisms grown under iron limiting conditions. This would allow the definition of iron-regulated genes in M. hyopneumoniae as a first step towards identifying their mechanism of regulation and possible role in virulence. Like other mycoplasmas, the medium for M. hyopneumoniae is supplemented with serum, a rich source 73 of iron-binding transferrin. The medium used for M. hyopneumoniae growth is supplemented with 20% swine serum, a higher level than normally used in most other mycoplasma media, which is approximately 5-10%. Under in vivo conditions, M. hyopneumoniae is exposed to mucosal secretions, a rich source of iron-binding transferrin and lactoferrin and similar to the in vitro growth conditions in terms of iron availability. If mycoplasmas are to respond to host iron sequestration mechanisms during disease, they must contain mechanisms for binding and transporting iron across their membrane. Little research has been done to examine the binding of iron-acquisition proteins lactoferrin and transferrin by mycoplasmas. Whether M. hyopneumoniae or any other mycoplasma requires iron for growth is not known, but it is likely that enzymes with iron containing prosthetic groups are present in mycoplasmas. One study has examined the interaction of M. pneumoniae and Mycoplasma genitalium with these proteins and found that M. pneumoniae binds lactoferrin but not transferrin (30). Mycoplasma genitalium did not bind either protein. Iron storage occurs in M. capricolum (4), but iron acquisition or iron requirements have not been studied further in the mycoplasmas. The rationale for using chelators in this study was to expose the mycoplasma to conditions that prevent iron uptake through normal pathways. This should be reflected in slower or lack of growth under the iron limiting conditions assuming there is a requirement for iron. Our initial studies used M. gallisepticum and M. pulmonis with three different iron chelators to estimate the need for iron among the mycoplasmas and to identify a chelator and concentration that might provide a slower growth response. Both of these species grow well and readily form colonies on agar surfaces for quantitative purposes. They are also both respiratory pathogens, and like M. hyopneumoniae are exposed to the transferrin-lactoferrin mediated iron sequestration mechanisms of the host. Since both species demonstrated a growth effect with 1 mg/ml 2,2'-dipyridyl (Fig. 3.1), this chelator and concentration were used in transcriptional profiling studies with M. hyopneumoniae. We also chose a single time point, 2 hours, because it represents an approximate generation time for M. hyopneumoniae under the growth conditions employed. The effect of iron deprivation on mRNA levels in M. hyopneumoniae is shown in Table 3.1 and Figure 3.2. Analysis of the data indicates that nine genes were up-regulated during iron deprivation. These genes include iolC (mhp 152), glpQ (mhp317), mlgA (mhp319), ackA (mhp505), potB (mhp558) in addition to four hypothetical genes (mhp140, mhpl51, mhp275, and mhp510). One of these, mhp 275, is a member of the PI 02 paralog family (18). The function of PI02 or any of its paralogs is not known, but it is interesting to speculate that one of these might have a role in iron uptake. Some of the known genes (mglA and potB) are 74 involved in transport, while others (iolC, glpQ and ackA) are involved in metabolism. One of the genes, mhp 140, has a prokaryotic membrane lipid attachment site and is thought to be a lipoprotein (18). Eighteen genes were down-regulated during iron limiting conditions. Seven of these (mhp081, mhp087, mhp176, mhp337, mhp411, mhp639 and mhp666) are hypothetical genes with no identified function. The remaining genes are identified as gatB (mhp030), fusA (mhp083), rpsL (mhp085), hpt (mhp 117), oppB-D (mhp 167-169), rpL2 (mhp 190), metK (mhp450), rplL (mhp637), and ushA (mhp651). Resistance to fusidic acid in fusA mutants of Salmonella enterica serovar Typhimurium, alters growth both in vivo and in vitro (15). So, although the serovar Typhimurium fusA mutants confer some antibiotic resistance, their ability to grow is also stunted. The down-regulated genes identified in this study have roles in translation and cell growth. The regulation or role of ushA in not clearly defined during transcription, however, it is known to increase 2-fold as cells enter the stationary phase (31). In M. hyopneumoniae this protein has a prokaryotic lipid attachment site and is thought to be a lipoprotein. Since ushA is down-regulated in this study, this could indicate a response to iron depletion or an instance where the majority of cells had not entered the stationary phase of growth, hpt is involved in the salvage of guanine for guanine nucleotide synthesis (36). Since cellular metabolism and growth are slowed during iron deprivation, the synthesis of nucleotides is expected to decrease as well. Our studies show that iron deprivation can reduce growth of M. gallisepticum and M. pulmonis suggesting that these mycoplasmas require iron for growth. It is thought that M. hyopneumoniae also requires iron, but the studies needed to demonstrate this have not been reported. The present studies do show, however, that M. hyopneumoniae alters mRNA levels of specific genes in response to low iron growth conditions, suggesting that regulation occurs. The degree of transcriptional change that occurs during low iron growth conditions may indicate that iron acquisition systems in mycoplasmas are tightly controlled. What mechanism(s) might be employed is not known. The hypothetical genes, though similar in number to the 44% of hypothetical genes in the genome (11 of 27), may play important roles in iron stress. Since there is no clear indication of the functions of those genes, additional studies are required to determine how they may be involved in iron acquisition or control of cellular metabolism. The evaluation of samples taken from the lungs of swine at the height of infection may be informative in conjunction with the data found here. However, laboratory grown organisms may be competing for iron in a vastly different ways than those found in their host niche. How M. hyopneumoniae secures sufficient levels of iron for its metabolism 75 and whether it is able to respond to low iron conditions through transcriptional regulation is not yet understood, but studies on the regulation mechanism(s) of several genes identified during this study may begin to unravel some of the mystery.

Acknowledgements We would like to thank Nancy Upchurch and Barb Erickson for assistance with mycoplasma cultures. Monica Perez contributed to the construction of the microarray. Mike Carruthers assisted with spot finding. Funding for this project was provided in part by the Iowa Healthy Livestock Advisory Council.

References 1. Aisen, P., and I. Listowsky. 1980. Iron transport and storage proteins. Ann. Rev. Biochem. 49:357-393. 2. Andrews, S. C., A. K. Robinson, and F. Rodriquez-Quinones. 2003. Bacterial iron homeostasis. FEMS Microbiol. Rev. 27:215-237. 3. Baker, E. N., B. F. Anderson, H. M. Baker, C. L. Day, M. Haridas, G. E. Norris, S. V. Rumball, C. A. Smith, and D. H. Thomas. 1994. Three-dimensional structure of lactoferrin in various functional states. Adv. Exp. Med. Biol. 357:1-12. 4. Bauminger, E. R., S. G. Cohen, F. Labenski-Dekanter, A. Levy, S. Aster, M. Kessel, and S. Rottem. 1980. Iron storage in Mycoplasma capricolum. J. Bacteriol. 141:378- 381. 5. Crosa, J. H. 1989. Genetics and molecular biology of siderophore-mediated iron transport in bacteria. Microbiol. Rev. 53:517-530. 6. Davis, J. K., K. M. Delozier, D. K. Asa, F. C. Minion, and G. H. Cassell. 1980. Interactions between murine alveolar macrophages and Mycoplasma pulmonis in vitro. Infect. Immun. 29:590-599. 7. DeBey, M. C., and R. F. Ross. 1994. Ciliostasis and loss of cilia induced by Mycoplasma hyopneumoniae in porcine tracheal organ cultures. Infect. Immun. 62:5312-5318. 8. Dudley, A. M., J. Aach, M. A. Steffen, and G. M. Church. 2002. Measuring absolute expression with microarrays with a calibrated reference sample and an extended signal intensity range. Proc. Natl. Acad. Sci. USA 99:7554-7559. 76

9. Dudoit, S., Y. H. Yang, M. J. Callow, and T. P. Speed. 2000. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. http://www.stat.Berkeley.edu/users/terry/zarray/Html/papersindex.html. 10. Freundt, E. A. 1983. Culture media for classic mycoplasmas, p. 127-135. In S. Razin and J. G. Tully (ed.), Methods in mycoplasmology. vol. 1. Mycoplasma characterization. Academic Press, New York. 11. Friis, N. F. 1975. Some recommendations concerning primary isolation of Mycoplasma hyopneumoniae and Mycoplasma flocculare, a survey. Nord. Veterinaermed. 27:337-339. 12. Hutchison III, C. A., and M. G. Montague. 2002. Mycoplasmas and the minimal genome concept, p. 221-253. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 13. Lamont, I. L., P. A. Beare, U. Ochsner, A. I. Vasil, and M. L. Vasil. 2002. Siderophore-mediated signaling regulates virulence factor production in Pseudomonas aeruginosa. Proc. Natl. Acad. Sci. USA 99:7072-7077. 14. Livingston, C. W., E. L. Stair, N. R. Underdahl, and C. A. Mebus. 1972. Pathogenesis of mycoplasmal pneumonia in swine. Am. J. Vet. Res. 33:2249-2258. 15. Macvanin, M., J. Bjorkman, S. Eriksson, M. Rhen, D. I. Andersson, and D. Hughes. 2003. Fusidic acid-resistant mutants of Salmonella enterica serovar Typhimurium with low fitness in vivo are defective in RpoS induction. Antimicrob. Agents Chemother. 47:3743-3749. 16. Mare, C. J., and W. P. Switzer. 1965. New Specieis: Mycoplasma hyopneumoniae, a causitive agent of virus pig pneumonia. Vet. Med. Small Anim. Clin. 60:841-846. 17. Mebus, C. A., and N. R. Underdahl. 1977. Scanning electron microscopy of trachea and bronchi from gnotobiotic pigs inoculated with Mycoplasma hyopneumoniae. Am. J. Vet. Res. 43:1249-1254. 18. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 19. O'Halloran, T. V. 1993. Transition metals in control of gene expression. Science 261:715-725. 20. Posey, J. E., and F. C. Gherardini. 2000. Lack of a role for iron in the Lyme disease pathogen. Science 288:1651-1653. 77

21. Ratledge, C., and L. G. Dover. 2000. Iron metabolism in pathogenic bacteria. Annu. Rev. Microbiol. 54:881-941. 22. Rodriguez, G. M., and I. Smith. 2003. Mechanisms of iron regulation in mycobacteria: role in physiology and virulence. Mol Microbiol 47:1485-1494. 23. Ross, R. F. 1993. Mycoplasma-animal pathogens, p. 69-109. In I. Kahane and A. Adoni (ed.), Rapid diagnosis of mycoplasmas. Plenum Press, New York. 24. Ross, R. F. 1992. Mycoplasmal disease, p. 537-551. In A. D. Leman, B. E. Straw, W. L. Mengeling, S. D'Allaire, and D. J. Taylor (ed.), Diseases of Swine. Iowa State University Press, Ames. 25. Salyers, A. A., and D. D. Whitt. 2002. Bacterial Pathogeneis: A Molecular Approach, Second ed. ASM Press, Washington, D. C. 26. Schaible, U. E., and S. H. E. Kaufmann. 2004. Iron and microbial infection. Nature Rev. 2:946-953. 27. Storey, J. 0., and R. Tibshirani. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA 100:9440-9445. 28. Szczebara, F., L. Dhaenens, S. Armand, and M. O. Husson. 1999. Regulation of the transcription of genes encoding different virulence factors in Helicobacter pylori by free iron. FEMS Microbiol. Lett. 175:165-170. 29. Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 30. Try on, V. V., and J. B. Baseman. 1987. The acquisition of human lactoferrin by Mycoplasma pneumoniae. Microbial Pathog. 3:437-443. 31. Wanner, B. L. 1996. Phosphorous assimilation and contol of the phosphate regulon, p. 1362. In F. C. Neidhardt (ed.), Escherichia coli and Salmonella Cellular and molecuolar biology, vol. 1. ASM Press, Washington, DC. 32. Weinberg, E. D. 1995. Acquisition of iron and other nutrients in vivo. In J. A. Roth, C. A. Bolin, K. A. Brogden, F. C. Minion, and M. J. Wannemuehler (ed.), Virulence mechanisms of bacterial pathogens. ASM Press, Washington, D C. 33. Weinberg, E. D. 1978. Iron and Infection. Microbiol. Rev. 42:45-66. 34. Wolfinger, R. D., G. Gibson, E. D. Wolfinger, L. Bennett, H. Hamadeh, P. Bushel, C. Afshari, and R. S. Paules. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comp. Biol. 8:625-637. 78

35. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed. 2002. Normalization for cDNA microarray data: a robust composite method for adressing single and multiple slide systematic variation. Nucleic Acids Res. 30:el5. 36. Zalkin, H., and P. Nygaard. 1996. Biosynthesis of purine nucleotides, vol. 1. ASM Press, Washington, DC. 79

CHAPTER 4. GENE EXPRESSION DIFFERENCES BETWEEN MYCOPLASMA HYOPNEUONIAE VIRULENT STRAIN 232 AND AVIRULENT STRAIN J

This is a paper to be submitted to Infection and Immunity

Melissa L. Madsen,1 Dan Nettleton,2 Eileen L. Thacker,1 and F. Chris Minion1

Departments of Veterinary Microbiology and Preventive Medicine1 and Statistics2, Iowa State University, Ames, IA

Abstract Mycoplasma hyopneumoniae causes swine pneumonia and contributes significantly to porcine respiratory disease complex. A lack of genetic tools prevents identification of virulence-related genes by classical genetic approaches. Newly emerging technology in the assessment of transcriptional activities across genomes, microarrays, provides an opportunity to compare strains of M. hyopneumoniae that differ significantly in their disease potential to identify virulence-related genes. Microarrays containing 632 open reading frames of M. hyopneumoniae were used to identify differences in transcriptional levels between virulent, low passage strain 232 and avirulent, high passage strain J. Thirty genes were identified that differed in mRNA levels between the two strains. In strain 232, 6 of 8 up-regulated genes were hypothetical with no known function. In strain J, nine ribosomal proteins and two elongation factor genes were up-regulated. These studies provide a striking contrast and may provide clues as to how mycoplasmas acclimate to growth in vitro and to which genes are needed for colonization and persistence in vivo.

Introduction Mycoplasma hyopneumoniae is the causative agent of enzootic pneumonia in swine and a major component of the porcine respiratory disease complex. The annual economic loss to the swine industry is significant and has been estimated to be as high as $200 million per year in the US (15, 21). Surveys have established that M. hyopneumoniae-mduced lesions are present in over 80% of the swine slaughtered in the US (18). While vaccines are available 80 and do limit disease, they do not effectively prevent colonization and are not always protective against outbreaks of mycoplasma disease. In addition to the pneumonia it causes, the organism predisposes the host to other pathogens. For instance, studies by Thacker et al. have shown that M. hyopneumoniae colonization significantly enhances the disease due to porcine reproductive and respiratory syndrome virus (22). Enhancement of porcine circovirus 2 disease has also been shown to occur as a result of M. hyopneumoniae infection (16). How M. hyopneumoniae effects this change in disease potential of other pathogens is not understood, but M. hyopneumoniae has been shown to adversely affect the pig immune system and perhaps the host's ability to respond to other infectious agents as well. Adherence to the ciliated epithelia is necessary to cause disease associated with M. hyopneumoniae. Previous studies have shown that strains of M. hyopneumoniae that adhere to swine cilia induce pneumonia in pigs while J strain, which is nonadherent, is incapable of colonization (29, 31). Once attached to cilia, M. hyopneumoniae causes cilia clumping and sloughing, with an accompanying reduction in mucociliary apparatus function (2). The mechanism involved has not been determined, but it may involve the release of calcium within ciliated cells (17). Thus, it is clear that attachment alone is not sufficient to cause disease. Other factors play a role in the virulence of M. hyopneumoniae. How the organism is able to affect host immune system function is not understood, but it may relate to gene products of the mycoplasma that interact with host immune effector cells resulting in abherent cellular responses. The lack of genetic systems to study pathogenesis in M. hyopneumoniae is a detriment to solving these issues. Progress, however, has been made in understanding the molecular basis for adherence to swine cilia. Following identification of the cilium adhesin and its description as a 97 kilodalton protein, P97 (29), a series of subsequent studies established not only the role for P97 in cilium binding (9) but also characterized the cilium binding site within the protein (10, 13). P97 undergoes posttranslational processing on the surface of the cell (3), and the proteolytic fragments remain cell associated through unknown mechanisms. This occurs for both the virulent strain 232 and the nonvirulent strain J. Processing of J strain P97, however, appears different than that for strain 232 (3). Additional proteins undergo proteolytic cleavage on the cell surface as well, but the extent to which this occurs has not been characterized. Differences in proteolytic processing can profoundly affect electrophoretic profiles and immunoblot analyses. The full complement of proteins that M. hyopneumoniae expresses can be estimated by analysis of the mRNA transcripts produced. By comparing virulent 232 and avirulent J strains, clues to the failure of J strain to produce disease can be obtained. To this end, a 81 microarray was constructed and used to assess transcriptional differences between strain 232 and J grown under normal laboratory conditions. The data shows that 30 genes differ in transcriptional activity between the two strains. The majority are up-regulated in J strain relative to strain 232. Most of these genes are related to translational processes and could indicate that J strain is functioning at a higher metabolic rate than the virulent strain under the conditions examined.

Materials and Methods Mycoplasma strains and culture conditions Pathogenic M. hyopneumoniae strain 232, a derivative of strain 11, was used in this study (12, 24) along with strain J, which is the type strain of M. hyopneumoniae (ATCC strain 25934) (7). Cultures for strain 232 were passaged fewer than 15 times in vitro. J strain is the type specific strain and has been extensively passed in vitro. Both strains were grown in Friis media as previously described (6). Twelve individual cultures were established in 500 ml flasks containing 250 ml of Friis media and were grown at 37°C- to mid-log phase as determined by medium color change and optical density. Six of the cultures were strain 232 and six were strain J. Mycoplasmas were pelleted by centrifugation at 24,000 x g, and one ml of RNAlater (Ambion, Inc. Austin, Tex.) was added to pellet. Pellets were stored at -70°C until the total RNA was isolated.

Microarray The M. hyopneumoniae microarray consists of polymerase chain reaction (PCR) products (probes) spotted to Coming UltraGAPS™ glass substrates (Coming, Inc., Big Flats, N.Y.). Ninety-one percent (632/698) of the total open reading frames (ORFs) of M. hyopneumoniae are represented on the array as PCR products of 125-350 bp in length. Each product is a unique sequence even within paralogous families as previously described (14). The 66 missing ORFs are due to an inability to design suitable primers (12 ORFs) or failed PCR reactions due to incorrect product size, multiple bands or no band (54 ORFs). Only ORFs greater than 125 bp are represented on the array and no tRNAs or rRNAs were included. The primer design, array construction, and validation have been described (Chapter 2).

Experimental design Six independent RNA samples from strain J cells were paired with six independent RNA samples from strain 232 (control) cells for hybridization on six two-color microarrays. The 82 six arrays were spotted on a total of three slides. Each slide was divided into two regions (upper and lower), and each region contained the full array of spots. Each of the 632 sequences in the array was represented by three replicate spots. For three of the six arrays, the control sample was labeled with Alexa 555 dye and compared to the iron-depleted sample labeled with Alexa 647 dye (Molecular Probes, Inc., Eugene, Ore.). The dye assignment to control and treated samples was reversed for the other three arrays. The three slides were hybridized under identical conditions as described below.

RNA isolation RNA was isolated from frozen cell pellets using the Versagene™ RNA Purification System (Centra Systems, Minneapolis, Minn.) according to the manufacturer's protocol. The optional step of DNase treatment was routinely performed on column according to the manufacturer's recommendation. With a cut-off of 150 bp, 5S rRNA and tRNAs were removed from the samples, limiting interference in downstream manipulations. Samples were quantified and checked for purity using the Nanodrop® ND-1000 Spectrophotometer (Nanodrop, Wilmington, Del.). If necessary, samples were concentrated using Microcon YM-30 micro-concentrators (Millipore, Billerica, Mass.) for optimal cDNA generation.

Target generation and hybridization Targets were generated from total RNA extracted from cell pellets as described above. Fluorescently labeled cDNA targets were generated and purified using the SuperScript™ Indirect cDNA Labeling System (Invitrogen Corp., Carlsbad, Calif.) with a set of 129 ORF- specific six-mer oligonucleotide primers designed as previously described (Chapter 2). Targets were labeled with Alexa Fluor™ 555 Reactive Dye or Alexa Fluor™ 647 Reactive Dye (Molecular Probes, Inc.). Prior to cDNA generation, strain 232 and J RNA samples were paired and equal concentrations of total RNA were added to the cDNA reactions. All of the resulting cDNA was then fluorescently labeled. Following purification of the labeled cDNA, samples were dried in a vacuum centrifuge and then resuspended in 10 pi Pronto! cDNA/long oligo hybridization solution (Coming). Targets were denatured at 95°C for 5 min and centrifuged at 13,000 rpm for 2 min at room temperature. Targets from a control 232 and J strain culture were then combined, pipetted to an array, and covered with a 22 x 22 mm HybriSlip™ (Schleicher & Schuell, Keene, N. H.). Slides were placed in a Corning hybridization chamber and incubated in a 42°C water bath for 12-16 h. Slides were washed according to Coming's UltraGAPS™ protocol and dried by centrifugation. 83

Image acquisition and normalization Each array was scanned with each dye channel using a ScanArray Express laser scanner (Applied BioSystems, Inc., Foster City, Calif.) at least three times under varying laser power and PMT gain settings to increase the dynamic range of expression measurement (4). Images were quantified using the softWorRx Tracker analysis software package (Applied Precision, Inc., Issaquah, Wash.). Spot-specific mean signals were corrected for local background by subtracting spot-specific median background intensities. The natural logarithm of the background-corrected signals from a single scan were adjusted by an additive constant so that all scans of the same array-by-dye combination would have a common median. The median of these adjusted-log-background-corrected signals across multiple scans was then computed for each spot to obtain one value for each combination of spot, array, and dye channel. These data for the two dye channels on any given array were normalized using LOWESS normalization to adjust for intensity-dependent dye bias (5, 26). Following LOWESS adjustment, the data from each channel were adjusted by an additive constant so that the median for any combination of array and dye would be the same for all array-by-dye combinations. The normalized values for triplicate spots were averaged within each array to produce one normalized measure of expression for each of the 632 probe sequences and each of the 12 RNA samples.

Data analysis A separate mixed linear model analysis was conducted for each probe sequence using the normalized data (25). Each mixed model included fixed effects for treatment (strain J vs. control), slide region (upper vs. lower), and dye (Alexa 555 vs. 647) as well as random effects for slide and slide-by-region interaction. A /-test for differential expression across treatments was conducted for each probe as part of our mixed linear model analysis. The 632 ^-values from these /-tests were converted to ^-values using the method of Storey and Tibshirani (20). These ^-values can be used to obtain approximate control of the False Discovery Rate at a specified value. For example, declaring probes with ^-values less than or equal to 0.05 to be differentially expressed produces a list significant results for which the False Discovery Rate is estimated to be approximately 5%. 84

Results Microarray studies Total RNA yields were between 9-17 \ig post purification. Approximately 9-12 p,g total RNA was included in each RT-PCR reaction. Fluorescent labeling efficiencies ranged from 27-79 base pair/dye molecule, which is within the manufacturer's expected range of fluorescent dye incorporation rates (http://probes.invitrogen.com/resources/calc/ basedyeratio.html). A total of five arrays were a part of this study. One complete array was dropped from the statistical analysis based on the low quality of the array (scratches and smears). A total of 30 genes showed significant transcript differences between strain 232 and J at jpO.OI (Table 4.1 and Fig. 4.1). The ^-values were less than 0.1434. Eight genes were up-regulated relative to strain J, and six of those were characterized as either conserved or unique hypothetical genes. Conserved hypothetical indicates a sequence with similar homology found in another organism but with unknown function. Twenty-two genes were up-regulated relative to strain 232 and many of these genes are involved in the translation process.

Log2 Fold Change

Figure 4.1. Volcano plot of transcriptional differences in M. hyopneumoniae strains 232 and J. Data represent individual gene responses plotted as Logz fold change vs. -Log p-value. Points above -Log/?-value = 2.0 are significantly up- or down-regulated at /XO.01. 85

Table 4.1. Transcriptional differences between M. hyopneumoniae strains 232 and J p<0.01. ID Gene1 Description p-value ig-value FC Lower levels of mRNA transcripts in J strain mhp016 CH Conserved Hypothetical 0.0063 0.1222 1.43 mhp394 CH Conserved Hypothetical 0.0041 0.0992 1.36 mhp441 sgaH Hexulose-6-Phosphate Synthase 0.0051 0.1074 1.57 mhp446 CH Conserved Hypothetical 0.0035 0.0892 1.67 mhp529 UH Unique Hypothetical 0.0069 0.1235 5.81 mhp530 CH Conserved Hypothetical 0.0015 0.0601 20.6 mhp535 UH Unique Hypothetical 0.0008 0.0531 14.3 mhp686 pr2 Multidrug Resistance Protein Homolog 0.0011 0.0531 3.03 Higher levels of mRNA transcripts in J strain mhp058 rpsB 3OS Ribosomal Protein S2 0.0017 0.0631 2.23 mhp059 Elongation Factor TS (EF-TS) 0.0045 0.1034 1.73 mhp083 fusA GTP-Binding Protein Chain Elongation Factor EF-G 0.0012 0.0531 2.08 mhpl21 recR Recombination Protein RecR 0.0001 0.0182 1.45 mhp135 rpsT 30S Ribosomal Protein S20 0.0055 0.1113 2.43 mhp157 nrdl Ribonucleotide Reductase 0.0068 0.1235 1.42 mhp158 nrdE Ribonucleoside-Diphosphate Reductase Alpha Chain 0.0003 0.0389 1.85 mhp164 CH Conserved Hypothetical 0.0029 0.0768 3.05 mhp165 CH Conserved Hypothetical 0.0002 0.0258 1.31 mhp166 oppF Oligopeptide Transport System Permease 0.0008 0.0531 1.53 mhp169 oppB Oligopeptide Transport System Permease 0.0011 0.0531 1.54 mhp180 alaS Alanyl-tRNA Synthetase 0.0018 0.0631 1.64 mhpl81 CH Conserved Hypothetical 0.0083 0.1377 1.14 mhp182 Protein PI02 0.0024 0.0727 2.27 mhp 197 rpS17 Ribosomal Protein S17 0.0049 0.1067 1.64 mhp198 rpL14 Ribosomal Protein L14 0.0006 0.0531 2.59 mhp199 rpL24 Ribosomal Protein L24 0.0021 0.0665 2.05 mhp204 rpL18 Ribosomal Protein LI8 0.0089 0.1434 1.72 mhp257 rpL35 Ribosomal Protein L35 0.0012 0.0531 1.40 mhp258 rpL20 Ribosomal Protein L20 0.0000 0.0141 2.79 mhp592 UH Unique Hypothetical 0.0028 0.0768 1.82 mhp672 rplM 50S Ribosomal Subunit Protein LI3 0.0074 0.1279 2.20 1 UH= unique hypothetical; CH=conserved hypothetical 2 (+) indicates that gene is up-regulated in strain 232 relative to J; (-) indicates that gene is up-regulated in strain J relative to 232. 86

Discussion The molecular basis of pathogenesis in M. hyopneumoniae is not understood. Aside from the participation of P97 in adherence to swine cilia, little is known about the cell-cell interactions that result in ciliated epithelial cell death and alteration of the host immune system. In lieu of a genetic system for the construction of defined mutations, insight into pathogenic mechanisms is often gained by comparing strains that differ in phenotype. For instance, spontaneous mutations arising in Mycoplasma pneumoniae were instrumental in defining the molecular interactions that direct the PI adhesin to its membrane location on the attachment organelle (I, 8, 11, 19). Isolation of mutants of M. hyopneumoniae, however, has not been possible because of its poor growth on agar surfaces, a prerequisite for mutant screening protocols. Thus, studies that identify colony-associated phenotypes or depend on multiple clonal isolates are simply not possible with M. hyopneumoniae. To compensate for these deficiencies, organisms that have been repeatedly cultured in vitro have been compared to low passage strains. The former lose their cilium adherence activity and disease potential during in vitro passage (30, 31). Other studies have compared different strains to gain at least some knowledge concerning M. hyopneumoniae's mechanisms of pathogenesis (23, 24, 29, 31). These studies have been difficult to interpret because of the variable protein patterns, which may be related more to surface proteolytic processing than to differences in gene expression (3). In addition, these comparative studies have not been extensive because there are few field strains available to study; M. hyopneumoniae is one of the more fastidious of the animal pathogens. One approach to identifying factors related to virulence in organisms refractory to genetic manipulation is to compare the transcriptional profiles of virulent and avirulent strains on a global scale to identify genes that may contribute to virulence. Based on the assumption that virulent organisms express factors that are either missing or down-regulated in avirulent strains, global expression profiling can be performed using microarrays representing the complete gene repertoire of an organism. The strains used in this study are well characterized regarding their disease potential and have been used in numerous investigations. Strain 232 is a virulent isolate that has been used repeatedly in pig challenge studies. It grows well in culture compared to recently isolated field strains that are sometimes difficult to culture at all. It is maintained at low passage both in pure culture and as swine lung homogenate, and it reproducibly causes lung lesions when 87 used to infect mycoplasma-free pigs (22, 27, 28). Strain J is the type strain, is in high passage, and cannot colonize pigs or cause disease (31). It is nonadherent to swine cilia. Transcriptional differences between strains 232 and J show a preponderance of transcripts related to translational processes up-regulated in avirulent strain J. This includes nine ribosomal protein genes and two elongation factor genes, fusA and tsf This is somewhat surprising and suggests that laboratory adaptation to artificial growth media such as had occurred with strain J involves increased levels of translation-associated gene products. This can occur either through increased transcription rates or decreased transcript turnover resulting in increased steady state levels. Not all translation-related proteins vary between the strains, however, so how these specific genes contribute in a meaningful way to the strain J phenotype is not clear. Since ribosomal RNA genes were not included on the array or in the primer set for cDNA generation, we were unable to determine if the structural RNAs were also up-regulated in the avirulent strain J. Eight genes up-regulated in strain 232 relative to strain J include six hypothetical genes of unknown function. Since none of the genes identified in this study are lipoproteins or have identifiable functional motifs, additional studies will be required to establish the function and virulence relationship of these genes. Also up-regulated in strain 232 is mhp182, the gene encoding PI 02, and mhp686, Pr2, a multidrug resistance protein homolog. The role of PI 02 has not been elucidated, but Pr2, more correctly described as an ABC type multidrug/protein/lipid transporter, might be involved in increased protein translocation. This could infer that increased protein secretion is important to colonization and disease and less important to growth in broth in the laboratory. This would not be surprising since the organism has little competition for nutrients in vitro in contrast to the highly competitive environment of the host mucosal surfaces. In summary, these studies show that strain 232 and J vary in transcriptional activity in at least 30 genes. For strain J, the variation is related to translational processes, but in the virulent strain 232, variation is mostly found in hypothetical genes of unknown function. Future studies should focus on the function of the genes up-regulated in strain 232 and determination if they are involved in pathogenesis.

Acknowledgements We thank Nancy Upchurch and Barb Erickson for assistance with mycoplasma cultures. Monica Perez contributed to the construction of the microarray. Mike Carruthers assisted 88 with spot finding. Funding for this project was provided in part by the Iowa Healthy Livestock Advisory Council.

References 1. Baseman, J. B., R. M. Cole, D. C. Krause, and D. K. Leith. 1982. Molecular basis for cytoadsorption of Mycoplasma pneumoniae. J. Bacteriol. 151:1514-1522. 2. DeBey, M. C., and R. F. Ross. 1994. Ciliostasis and loss of cilia induced by Mycoplasma hyopneumoniae in porcine tracheal organ cultures. Infect. Immun. 62:5312-5318. 3. Djordjevic, S. P., S. J. Cordwell, M. A. Djordjevic, J. Wilton, and F. C. Minion. 2004. Proteolytic processing of the Mycoplasma hyopneumoniae cilium adhesin. Infect. Immun. 72:2791-2802. 4. Dudley, A. M., J. Aach, M. A. Steffen, and G. M. Church. 2002. Measuring absolute expression with microarrays with a calibrated reference sample and an extended signal intensity range. Proc. Natl. Acad. Sci. USA 99:7554-7559. 5. Dudoit, S., Y. H. Yang, M. J. Callow, and T. P. Speed. 2000. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. http://www.stat.Berkelev.edu/users/terrv/zarrav/Html/DaDersindex.html. 6. Friis, N. F. 1975. Some recommendations concerning primary isolation of Mycoplasma hyopneumoniae and Mycoplasma flocculare, a survey. Nord. Veterinaermed. 27:337-339. 7. Goodwin, R. F. W., A. P. Pomeroy, and P. Whittestone. 1965. Production of enzootic pneumonia in pigs with a mycoplasma. Vet. Rec. 77:1247-1249. 8. Hansen, E. J., R. M. Wilson, and J. B. Baseman. 1979. Isolation of mutants of Mycoplasma pneumoniae defective in hemadsorption. Infect. Immun. 23:903-906. 9. Hsu, T., S. Artiushin, and F. C. Minion. 1997. Cloning and functional analysis of the P97 swine cilium adhesin gene of Mycoplasma hyopneumoniae. J. Bacteriol. 179:1317-1323. 10. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infec. Immun. 66:4762-4766. 11. Krause, D. C. 1996. Mycoplasma pneumoniae cytadherence: Unravelling the tie that binds. Mol. Microbiol. 20:247-253. 12. Mare, C. J., and W. P. Switzer. 1965. New Specieis: Mycoplasma hyopneumoniae, a causitive agent of virus pig pneumonia. Vet. Med. Small Anim. Clin. 60:841-846. 89

13. Minion, F. C., C. Adams, and T. Hsu. 2000. RI region of P97 mediates adherence of Mycoplasma hyopneumoniae to swine cilia. Infect. Immun. 68:3056-3060. 14. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 15. Muller, R. D., and P. B. Abbot. 1986. Estimating the cost of respiratory disease in hogs. Anim. Health Nutr. February:30-35. 16. Opriessnig, T., E. L. Thacker, S. Yu, M. Fenaux, X. J. Meng, and P. G. Halbur. 2004. Experimental reproduction of postweaning multisystemic wasting syndrome in pigs by dual infection with Mycoplasma hyopneumoniae and porcine circovirus type 2. Vet. Pathol. 41:624-640. 17. Park, S.-C., S. Yibchok-Anun, H. Cheng, T. F. Young, E. L. Thacker, F. C. Minion, R. F. Ross, and W. H. Hsu. 2002. Mycoplasma hyopneumoniae increases intracellular calcium release in porcine ciliated tracheal cells. Infect. Immun. 70:2502-2506. 18. Ross, R. F. 1992. Mycoplasmal disease, p. 537-551. In A. D. Léman, B. E. Straw, W. L. Mengeling, S. D'Allaire, and D. J. Taylor (ed.), Diseases of Swine. Iowa State University Press, Ames. 19. Stevens, M. K., and D. C. Krause. 1991. Localization of the Mycoplasma pneumoniae cytadherence-accessory proteins HMW1 and HMW4 in the cytoskeletonlike triton shell. J. Bacteriol. 173:1041-1050. 20. Storey, J. D., and R. Tibshirani. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA 100:9440-9445. 21. Straw, B. E., V. K. Tuovinen, and M. Bigras-Poulin. 1989. Estimation of the cost of pneumonia in swine herds. J. Am. Vet. Med. Assn. 195:1702-1706. 22. Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 23. Vicca, J., T. Stakenborg, D. Maes, P. Butaye, J. Peeters, A. de Kruif, and F. Haesebrouck. 2003. Evaluation of virulence of Mycoplasma hyopneumoniae field isolates. Vet. Microbiol. 97:177-190. 24. Whittlestone, P. 1979. Porcine mycoplasmas, p. 133-176. In J. G. Tully and R. F. Whitcomb (ed.), The Mycoplasmas. Volume II. Human and Animal Mycoplasmas. Academic Press, New York. 90

25. Wolfmger, R. D., G. Gibson, E. D. Wolfmger, L. Bennett, H. Hamadeh, P. Bushel, C. Afshari, and R. S. Paules. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comp. Biol. 8:625-637. 26. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed. 2002. Normalization for cDNA microarray data: a robust composite method for adressing single and multiple slide systematic variation. Nucleic Acids Res. 30:el5. 27. Young, T. F., Y. W. Chiang, and R. F. Ross. 1990. Evaluation of local and systemic humoral immune responses to Mycoplasma hyopneumoniae. Proc. Int. Pig Vet. Soc. 11:97. 28. Young, T. F., and R. F. Ross. 1987. Assessment of antibody response of swine infected with Mycoplasma hyopneumoniae by immunoblotting. Am. J. Vet. Res. 48:651-6. 29. Zhang, Q., T. F. Young, and R. F. Ross. 1995. Identification and characterization of a Mycoplasma hyopneumoniae adhesin. Infect. Immun. 63:1013-1019. 30. Zielinski, G. C., and R. F. Ross. 1993. Adherence of Mycoplasma hyopneumoniae to porcine ciliated respiratory tract cells. Am. J. Vet. Res. 54:1262-1269. 31. Zielinski, G. C., and R. F. Ross. 1990. Effect of growth in cell cultures and strain on virulence of Mycoplasma hyopneumoniae for swine. Am. J. Vet. Res. 51:344-348. 91

CHAPTER 5. TRANSCRIPTIONAL VARIATION BETWEEN HIGH AND LOW ADHERENT VARIANTS OF MYCOPLASMA HYOPNEUMONIAE

A paper to be submitted to Infection and Immunity

Melissa L. Madsen,1 Dan Nettleton,2 Eileen L. Thacker,1 and F. Chris Minion1

Departments of Veterinary Microbiology and Preventive Medicine1 and Statistics2, Iowa State University, Ames, IA;

Abstract Mycoplasma hyopneumoniae attaches to swine cilia in the respiratory tract resulting in colonization and disease. The overall attachment process is poorly understood, but one component, P97, has been defined as the major cilium adhesin. Studies with a nonadherent, high passage strain of M. hyopneumoniae, strain J, showed that P97 must be working in concert with other mycoplasma components. To study this system further, a high adherent and a low adherent variant of the virulent strain 232 were compared by transcriptional profiling using a microarray containing 91 % of the open reading frames of M. hyopneumoniae. Only when compared against each other and not to the parent strain were transcriptional differences observed at /?<0.05. Genes up-regulated in the high adherent variant included the heat shock protein gene dnaJ, the signal peptidase gene lepA, and the gene for lipoprotein P65. Genes up-regulated in the low adherent variant included translation components and pdhD, the gene for dihydrolipoamide dehydrogenase. Both variants had genes up-regulated with unassigned functions that will require additional analysis.

Introduction Mycoplasma hyopneumoniae causes enzootic pneumonia in swine and is a major component of the porcine respiratory disease complex. The disease is endemic worldwide and results in significant annual loss for the swine industry. It has been estimated that as many as 80% of swine at slaughter have lesions consistent with mycoplasma pneumonia (35). Vaccination programs limit the disease, but do not prevent colonization of the 92 organism. Chronic, low-level disease serves as a source of infection for naïve or other unprotected swine and has been shown to enhance disease due to other pathogens (1, 5, 38) In addition to the pneumonia it causes, M. hyopneumoniae predisposes the host to other pathogens. For instance, studies by Thacker et al. have shown that M. hyopneumoniae colonization significantly enhances the disease due to porcine reproductive and respiratory syndrome virus (38). Enhancement of porcine circovirus 2 disease has also been shown to occur as a result of M. hyopneumoniae infection (30). How M. hyopneumoniae effects this change in disease potential of other pathogens is not understood, but M. hyopneumoniae has been shown to adversely affect the pig immune system (4, 20, 26, 39) and perhaps the host's ability to respond to other infectious agents as well. Adherence to the ciliated epithelia is necessary to cause disease associated with M. hyopneumoniae. Previous studies have shown that strains of M. hyopneumoniae that adhere to swine cilia induce pneumonia in pigs (45, 48). Once attached to cilia, M. hyopneumoniae causes cilia clumping, cell death and reduces the mucociliary apparatus function (6). The mechanism involved in cilium clumping and epithelial cell death has not been determined, but it is thought that the interaction of M. hyopneumoniae with the epithelial cell membrane induces the release of intracellular stores of calcium (31). The loss of mucociliary function can contribute to susceptibility to other pathogens and to the severity of M. hyopneumoniae infections. Thus, it is clear that attachment alone is not sufficient to cause disease. Other factors must play a role in the virulence of M. hyopneumoniae. How the organism is able to affect host immune system function is not understood, but it may relate to gene products of the mycoplasma that interact with host immune effector cells resulting in abherent cellular responses. The lack of genetic systems to study pathogenesis in M. hyopneumoniae is a detriment to solving these issues. Disease begins with colonization of the respiratory tract by M. hyopneumoniae and binding of M. hyopneumoniae to the cilia resulting in ciliostasis. Two adherent colony variants of M. hyopneumoniae strain 232 have been identified and characterized as high- adherent variant 91-3 and low-adherent variant 60-3 (T. F. Young, Q. Zhang, B. Z. Erickson, and R. F. Ross, Abstr. 10th Congr. Int. Org. Mycoplasmol. abstr. P260, 1994). Preliminary reports using microliter swine cilia adherence assays (46) indicated that the variants and parent strain 232 differ significantly in adherence (R. F. Ross, personal communication). The investigators' results also included pig infection studies that concluded that in vivo passage of the low-adherent variant resulted in its reversion to a moderate to high adherent state. In a tissue culture assay, Ross et al. showed that the high adherent variant 91-3 and non-adherent strain J mimicked adherence in vivo (43). The adherent variant 91-3 showed significant 93 binding, while the non-adherent strain J failed to bind cilia. In addition, the adherence phenotype can be lost during in vitro passage (R. F. Ross, personal communication). Thus, in the absence of cell surfaces, adherence is not required for growth and is not maintained. Several studies have sought to identify components involved in adherence to the cilia of the swine lung (17, 45, 47). The P97 adhesin has been well characterized and the epitope necessary for binding to the cilia defined (15, 27). The culmination of these studies indicates that P97 is the main adhesin for M. hyopneumoniae, but it also appears that other proteins may be involved in adherence to swine cilia. This study examined the relative mRNA transcript levels of the high-adherent variant 91-3 and the low-adherent variant 60-3 compared to the parent strain 232. The results show differences between 91-3 and 60-3 that could give clues to accessory proteins and activities associated with the adherence phenotype.

Materials and Methods Mycoplasma strains and culture conditions Pathogenic M. hyopneumoniae strain 232, a derivative of strain 11, was used in this study (23). Variants 91-3 and 60-3 are from single colonies picked from M. hyopneumoniae strain 232 plated onto solid media (R. F. Ross, personal communication). Cultures used were passed fewer than 15 times in vitro in Friis media as previously described (11). One hundred twenty-five ml cultures were grown at 37°C to mid-log phase as determined by medium color change and optical density. Cells were pelleted by centrifugation at 24,000 x g, and 1 ml of RNAlater (Ambion) was added to pellet. Pellets were stored at -70°C until the total RNA was isolated.

Microarray The M. hyopneumoniae microarray consists of polymerase chain reaction (PCR) products (probes) spotted to Coming UltraGAPS™ glass substrates (Coming, Inc., Big Flats, N.Y.). Ninety-one percent (632/698) of the open reading frames (ORFs) in the M. hyopneumoniae genome are represented on the array as PCR products of 125-350 base pair in length. Each product is a unique sequence even within paralogous families as described by Minion et al. (29). The 66 missing ORFs are due to an inability to design suitable primers (12 ORFs) or failed PCR reactions due to incorrect product size, multiple bands or no band (54 ORFs). Only ORFs greater than 125 base pairs are represented on the array and no tRNAs or rRNAs were included. The primer design, array construction and validation have been described (Chapter 2). 94

Experimental design Each array consists of 632 ORF-specific PCR products spotted in triplicate using a nonadjacent, well-spaced format. Each slide contains two complete arrays, one at each end. Six independent RNA samples from variant 91-3 cells were paired with six independent RNA samples from control strain 232 cells for hybridization on six two-color microarrays. The six arrays were spotted on a total of three slides. Five independent RNA samples from variant 60-3 cells were paired with five independent RNA samples from control strain 232 cells for hybridization on five two-color microarrays. The five arrays were spotted on a total of three slides. Each slide was divided into two regions (upper and lower), and each region contained the full array of spots. For three of the arrays, the control sample was labeled with Alexa 555 dye and compared to either variant 91-3 or 60-3 sample labeled with Alexa 647 dye (Molecular Probes, Inc., Eugene, Ore.). The dye assignment to control and treated samples was reversed for the other two arrays. The slides were hybridized under identical conditions as described below.

RNA isolation RNA was isolated from frozen cell pellets using the Versagene™ RNA Purification System (Centra Systems, Minneapolis, Minn.) according to the manufacturer's protocol. The optional step of DNase treatment was performed on column according to the manufacturer's recommendation. With a column cut-off of 150 base pairs, 5S rRNA and tRNAs were removed from the samples, limiting interference in downstream manipulations. Samples were quantified and checked for purity using the Nanodrop® ND-1000 Spectrophotometer (Nanodrop, Wilmington, Del.). If necessary, samples were concentrated using Microcon YM-30 micro-concentrators (Millipore, Billerica, Mass.) for optimal cDNA generation.

Target generation and hybridization Targets were generated from total RNA extracted from cell pellets as described above. Fluorescently labeled cDNA targets were generated and purified using the SuperScript™ Indirect cDNA Labeling System (Invitrogen Corp., Carlsbad, Calif.) with a set of 129 ORF- specific hexamer oligonucleotide primers designed as previously described (Chapter 2). Targets were labeled with Alexa Fluor™ 555 Reactive Dye or Alexa Fluor™ 647 Reactive Dye (Molecular Probes, Inc.). Following purification of the labeled cDNA, samples were dried in a vacuum centrifuge and then resuspended in 10 ^1 Pronto! cDNA/long oligo hybridization solution (Coming). Targets were denatured at 95°C for 5 min and centrifuged at 95

13,000 xg for 2 min at room temperature. Targets from a control strain 232 and either a 91-3 or 60-3 adherence variant culture were then combined, pipetted to an array, and covered with a 22 x 22 mm HybriSlip™ (Schleicher & Schuell, Keene, N. H.). Slides were placed in a Corning hybridization chamber and incubated in a 42°C water bath for 12-16 h. Slides were washed according to Coming's UltraGAPS™ protocol and dried by centrifugation.

Image acquisition and normalization Each array was scanned with each dye channel using a ScanArray Express laser scanner (Applied BioSystems, Inc., Foster City, Calif.) at least three times under varying laser power and PMT gain settings to increase the dynamic range of expression measurement (8). Images were quantified using the softWorRx Tracker analysis software package (Applied Precision, Inc., Issaquah, Wash.). Spot-specific mean signals were corrected for local background by subtracting spot-specific median background intensities. The natural logarithm of the background-corrected signals from a single scan were adjusted by an additive constant so that all scans of the same array-by-dye combination would have a common median. The median of these adjusted-log-background-corrected signals across multiple scans was then computed for each spot to obtain one value for each combination of spot, array, and dye channel. These data for the two dye channels on any given array were normalized using LOWESS normalization to adjust for intensity-dependent dye bias (9, 41). Following LOWESS adjustment, the data from each channel were adjusted by an additive constant so that the median for any combination of array and dye would be the same for all array-by-dye combinations. The normalized values for triplicate spots were averaged within each array to produce one normalized measure of expression for each of the 632 probe sequences and each of the 10 control and 10 adherent variant RNA samples.

Data analysis A separate mixed linear model analysis was conducted for each probe sequence using the normalized data (40). Each mixed model included fixed effects for treatment (adherent variant versus control), slide region (upper versus lower), and dye (Alexa 555 versus 647) as well as random effects for slide and slide-by-region interaction. A /-test for differential expression across treatments was conducted for each probe as part of our mixed linear model analyses. The 632 ^-values from these /-tests were converted to q-values using the method of Storey and Tibshirani (37). These q-values can be used to obtain approximate control of the False Discovery Rate at a specified value. For example, declaring probes with ^-values less than or equal to 0.05 to be differentially expressed produces a list significant results for 96 which the False Discovery Rate is estimated to be approximately 5%. Along with ^-values, estimates of fold-change were computed for each probe by taking the inverse natural log of the mean treatment difference estimated as part of our mixed linear model analysis.

Results Microarray studies Microarrays were used to compare mRNA steady state levels between strain 232 and two adherence variants isolated from its clonal population. Total RNA yields in the preparations were between 9-17 |ig post purification. Approximately 9-12 jug total RNA was included in each RT-PCR reaction to generate cDNA for preparation of fluorescently labeled targets. Following labeling reactions, fluorescent labeling efficiencies ranged from 18-47 base pair/dye molecule, which is within the manufacturer's expected range of fluorescent dye incorporation rates (http://probes.invitrogen.com/ resources/calc/basedyeratio.html). This provided significant signal intensities for spot finding and analysis. Five arrays were analyzed for variant 91-3; one complete array was dropped from the statistical analysis based on the quality of the array (scratches and smears). Four arrays were analyzed for variant 60- 3. The fifth array was dropped from the statistical analysis due to poor label incorporation and low probe signals when scanned. Since each array contained triplicate spots for each ORF, the analysis included 12 and 9 values for adherent variant 91-3 and 60-3, respectively, and 21 control 232 values. Statistical analysis was performed using a mixed linear model design. When each variant (91-3 and 60-3) was compared to parent strain 232, no significant transcriptional differences were identified (p<0.05). Variants 91-3 and 60-3 were then compared directly to one another to determine if any transcription differences between the two variants would yield insight into mechanisms of adherence. ANOVA was used to determine differences between signal intensities of variants 91-3 and 60-3. Estimated mean differences were significant for 34 genes at p<0.05 (Table 5.1) with a Rvalue of 0.2882. Eighteen of the 34 genes (53%) differentially transcribed are classified as hypothetical. 97

Table 5.1. Variation in signal intensities between variants 60-3 and 91-3 at /?<0.05.* Fold ID Gene Description p-value Change Genes with increased signal intensities in 60-3 mhp025 CH Conserved Hypothetical 0.0471 2.08 mhp083 fusA GTP-Binding Protein Chain Elongation 0.0135 4.69 Factor EF-G mhpl 10 smpB SSRA-Binding Protein 0.0095 2.39 mhp200 rpL5 Ribosomal Protein L5 0.0456 2.17 mhp215 CH Conserved Hypothetical 0.0114 1.60 mhp321 UH Unique Hypothetical 0.0433 2.12 mhp340 CH Conserved Hypothetical 0.0149 2.19 mhp359 CH Conserved Hypothetical 0.0309 2.02 mhp504 pdhD Dihydrolipoamide Dehydrogenase 0.0260 2.74 mhp548 Phosphoglucose Isomerase B 0.0266 1.54 mhp608 thrS Threonyl-tRNA Synthetase 0.0193 1.78 mhp639 CH Conserved Hypothetical 0.0113 3.50 mhp650 CH Conserved Hypothetical 0.0177 2.50 mhp670 CH Conserved Hypothetical 0.0053 2.89 mhp694 valS Valyl-tRNA Synthetase 0.0393 1.50 Genes with increased signal intensities in 91-3 mhp046 UH Unique Hypothetical 0.0352 4.80 mhp073 dnaJ Heat-Shock Protein 0.0318 2.18 mhp078 lepA 30 kDa GTP-Binding Protein LepA, 0.0388 3.30 Signal Peptidase mhp091 CH Conserved Hypothetical 0.0334 5.55 mhp092 CH Conserved Hypothetical 0.0095 22.18 mhp094 rpsP 30S Ribosomal Protein SI6 0.0474 10.18 mhp095 trmD tRNA (Guanine-N1 )-Methyltransferase 0.0257 42.10 mhpl 00 pyrG CTP Synthetase 0.0381 5.00 mhplOl CH Conserved Hypothetical 0.0301 19.10 mhpl 32 CH Conserved Hypothetical 0.0161 2.93 mhpl75 ftsH Cell Division Protein 0.0307 6.78 mhp247 CH Conserved Hypothetical 0.0282 2.22 mhp304 gW GTP-Binding Protein 0.0228 3.80 mhp530 CH Conserved Hypothetical 0.0113 6.69 mhp542 CH Conserved Hypothetical 0.0470 5.82 mhp583 UH Unique Hypothetical 0.0423 3.64 mhp588 nagA N-Acetylglucosamine-6-Phosphate 0.0242 3.34 Deacetylase mhp677 p65 Surface Lipoprotein 0.0351 10.64 mhp693 CH Conserved Hypothetical 0.0090 7.84 * UH = unique hypothetical; CH = conserved hypothetical. 98

2.5

0.5

6 •5 •4 •3 •2 1 0 1 2 3 Log2 Fold Change

Figure 5.1. Volcano plot of transcriptional differences in M. hyopneumoniae high- adherent variant 91-3 and low-adherent variant 60-3. Data represent individual gene responses plotted as Logz fold change vs. -Log p-value. Points above -Log p-value = 1.3 are significantly up- or down-regulated relative to variant 91-3 at p<0.05.

Discussion Mycoplasmas adhere tightly to host epithelial cells and should be considered surface parasites because of their need for macromolecular precursors. This is due to their limited genome capacity and lack of biosynthetic pathways for amino acids, purines and pyrimidines, and membrane components, i.e., phospholipids and cholesterol (32-34). A guiding principle has been that nonadherent mycoplasmas do not cause cell damage or disease (18). Mycoplasmas must adhere tightly to their host's mucosal surfaces to initiate and perpetuate disease. 99

The interactions of mycoplasmas with eukaryotic cells have been classified into various categories including hemagglutination (28) and hemadsorption of red blood cells (22), attachment to tissue culture cells (13, 47) and attachment to organ expiants (12, 44) and epithelial surfaces (25). The most significant advances in unraveling adherence mechanisms in mycoplasmas have occurred through isolation of mutants lacking adherence activity. For instance, Mycoplasma pneumoniae spontaneous hemadsorption-negative mutants were used to identify components of the adherence apparatus (22). In that species, the adherence protein PI is organized into a specialized attachment organelle by interacting with multiple proteins that serve to organize the attachment organelle structure (21). Mutations in any of the components involved resulted in a hemadsorption-negative phenotype and were easily recognizable. By comparing protein gel patterns of the mutants with the wild type strain, it was possible to identify these components. Mycoplasma hyopneumoniae adherence has proved difficult to study because it has no attachment organelle; the adhesin is distributed around the cell surface, and it is difficult to study individuals in a population. This species is more fastidious and is difficult to grow on agar surfaces. Despite these limitations, it was possible to identify cell populations with high and low adherence phenotypes. Parent strain 232 was plated onto agar media following dispersion of clumps by passing through a 26-gauge needle, and after 10 days of incubation, 94 well-isolated pinpoint colonies were picked to broth. Eighty-one colonies grew and were tested in the swine cilium adherence assay (46). One high and one low adherent variant were identified and designated variant 91 and 60, respectively. Single colonies were picked from each of the variants and the adherence phenotype was consistent. The designation 91-3 and 60-3 were indicative of the number given to the second set of colony isolates. These variants were studied further by immunoblot analysis, and pig challenge studies. By one-dimensional electrophoresis, only one protein band difference was observed between the two variants, a 106-kDa protein that has not been identified (R. F. Ross, personal communication). In challenge studies, the mean onset of coughing was different (11.6 days for variant 91, 25 days for variant 60) and both caused lung lesions. Pigs receiving the low adherent variant had fewer lung lobes with pneumonia per pig and lower lesion scores than either the high adherent variant or the parent strain 232 (R. F. Ross, personal communication). Upon reisolation from challenged pigs, the low adherent variant had improved its adherence to swine cilia to a moderate to strong phenotype, so the variant 60-3 is fully competent to regain adherence and cause disease. Because of the lack of genetic tools to generate more precise mutants in this mycoplasma species, these two variants represent the best models for studying adherence mechanisms in M. hyopneumoniae. 100

Microarrays were used to identify transcriptional differences between the two phenotypic variants. Like most bacteria, it is thought that mycoplasmas control gene product concentrations by regulating transcription. We hypothesized that the low adherent phenotype was due to changes in transcript levels of one or more proteins involved in adherence. When the transcriptome profiles of the adherence variants were compared to their parent strain 232 by microarray, however no transcriptional differences were observed at £><0.05. This can be explained if one considers strain 232 as a mixed population of cells arising from small genetic changes occurring during DNA replication. This has been referred to as quasispecies, a population of related, nonidentical genotypes (2). This population would consist of varying phenotypes, and any value arising from an analysis using this population would therefore represent a mean. Not surprisingly, transcriptional profiles of the adherence variants were not significantly different from this control mean. Direct comparison of transcriptional profiles between variants 91-3 and 60-3, however, did show significant differences at/K0.05. Genes that had higher signal intensities and are consequently thought to be up-regulated in variant 60-3 included eight of fifteen (53%) hypothetical genes (Table 5.1). The gene for elongation factor FusA was up-regulated as well as two metabolic genes, pdhD (dihydolipoamide dehydrogenase) and pgiB (phosphoglucose isomerase B). In addition, other genes up-regulated in variant 60-3 seem to be responsive to environmental stress. During amino acid starvation, valS (mhp694), a valine synthase, is transiently derepressed (14). Mutations in pgiB (mhp548) cause a decrease in growth in the presence of glucose in Escherichia coli (10) suggesting that it has an important role in glycolysis. PdhD is also involved in glycolysis, and its gene is up-regulated in 60-3 relative to 91-3. This molecule has been previously characterized as having a lipoyl-binding domain, which is normally associated with pdhC (24). thrS (mhp608) is autoregulated negatively at the translational level in E. coli (36). Perhaps thrS is regulated transcriptionally in M. hyopneumoniae. How these genes with assigned functions contribute to the low adherent phenotype is not known. A greater number of genes had higher signal intensities in variant 91-3 relative to 60-3. Forty-seven percent of these were hypothetical genes, a number that is similar to the overall estimation across the genome (29). Among the genes with assigned functions were the heat shock gene dnaJ (mhp073) and the gene coding for the membrane lipoprotein P65 (mhp677)(19). Lipoproteins play important roles in mycoplasma physiology and pathogenesis, often undergoing phase switching and size variation (42). For mycoplasmas containing these gene families, the membrane surface is in constant flux, changing in random fashion and providing a variable antigenic profile to the host immune system. It is unlikely 101 that M. hyopneumoniae has the capacity to vary its surface structure in this way; no lipoprotein gene families were found in the genome sequence (29). Fifty-three putative lipoproteins were found in the genome and none have an assigned function. In some mycoplasmas, however, lipoproteins participate in cell adherence processes (19). Whether P65 can assist in the binding of M. hyopneumoniae to swine cilia is unknown, but this would explain the increased adherence associated with this variant. Other genes up-regulated in 91- 3 are trmD and rpsP, which are regulated at the translational level in E. coli (3). Our data suggest a different mode of regulation in mycoplasmas, but the purpose for that regulation is not clear. Two gene products associated with cell division and DNA replication, FtsH (mhpl75) and PyrG (mhpl00), are up-regulated in variant 91-3 as well. How these two gene products might be involved in adherence is not understood. Finally, lepA is up-regulated in variant 91-3. This signal peptidase may play a critical role in processing of proteins involved in adherence as shown previously by Djordjevic et al. (7). Perhaps the proteolytic processing associated with the increased expression of this peptidase results in the exposure of additional binding epitopes in the high adherent variant. Further studies will be required to gain a better understanding of the adherence process in M. hyopneumoniae. These data in conjunction with other studies (16) indicate that P97 probably does not act alone; additional adhesins or supportive proteins probably participate in yet undefined ways. Unlike M. pneumoniae where proper placement of the P1 adhesin is critical for adherence (21), M. hyopneumoniae lacks an attachment organelle and the proteins involved in adhesin organization. The M. hyopneumoniae adhesin, however, is processed on the mycoplasma cell surface, so variation in processing may account for differences in the adherence phenotype (7). The hypothetical genes identified this study may play yet undefined roles in adherence as well. Further study with these variants may provide additional evidence for the contribution of specific proteins to the adherence process.

Acknowledgements We thank Nancy Upchurch and Barb Erickson for assistance with mycoplasma cultures. Monica Perez contributed to the construction of the microarray. Mike Carruthers assisted with spot finding. We are indebted to Richard F. Ross and Theresa F. Young for contributing the high and low adherent variants for this study. Funding for this project was provided in part by the Iowa Healthy Livestock Advisory Council. 102

References 1. Amass, S. F., L. K. Clark, W. G. van Alstine, T. L. Bowersock, D. A. Murphy, K. E. Knox, and S. R. Albregts. 1994. Interaction of Mycoplasma hyopneumoniae and Pasteurella multocida infections in swine. J Am Vet Med Assoc 204:102-107. 2. Baccam, P., R. J. Thompson, Y. Li, W. O. Sparks, M. Belshan, K. S. Dorman, Y. Wannemuehler, J. L. Oaks, J. L. Cornette, and S. Carpenter. 2003. Subpopulations of equine infectious anemia virus Rev coexist in vivo and differ in phenotype. J. Virol. 77:12122-12131. 3. Bjork, G. R. 1996. Stable RNA modification, p. 868-869. In F. C. Neidhardt (ed.), Esherichia coli and Salmonella cellular and molecular biology. ASM Press, Washington, D C. 4. Caruso, J. P., and R. F. Ross. 1990. Effects of Mycoplasma hyponeumoniae and Actinobacillus (Haemophilus) pleuropneumoniae infections on alveolar macrophage functions in swine. Am. J. Vet. Res. 51:227-231. 5. Ciprian, A., C. Pijoan, T. Cruz, J. Camacho, J. Tortora, G. Colmenares, R. R. Lopez, and M. de la Garza. 1988. Mycoplasma hyopneumoniae increases the susceptibility of pigs to experimental Pasteurella multocida pneumonia. Can. J. Vet. Res. 52:434-438. 6. DeBey, M. C., and R. F. Ross. 1994. Ciliostasis and loss of cilia induced by Mycoplasma hyopneumoniae in porcine tracheal organ cultures. Infect. Immun. 62:5312-5318. 7. Djordjevic, S. P., S. J. Cordwell, M. A. Djordejevic, J. Wilton, and F. C. Minon. 2004. Proteolytic processing of the Mycoplasma hyopneumoniae cilium adhesin. Infectious Immunology 72:2791-2802. 8. Dudley, A. M., J. Aach, M. A. Steffen, and G. M. Church. 2002. Measuring absolute expression with microarrays with a calibrated reference sample and an extended signal intensity range. Proc. Natl. Acad. Sci. USA 99:7554-7559. 9. Dudoit, S., Y. H. Yang, M. J. Callow, and T. P. Speed. 2000. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. http://www.stat.Berkeley.edu/users/terry/zarray/Html/papersindex.html. 10. Fraenkel, D. G. 1996. Glycolysis, p. 189-198. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, D. M. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and E. Umbarger (ed.), Escherichia coli and Salmonella. Cellular and molecular biology. ASM Press, Washington, D C. 103

11. Friis, N. F. 1975. Some recommendations concerning primary isolation of Mycoplasma hyopneumoniae and Mycoplasma flocculare, a survey. Nord. Veterinaermed. 27:337-339. 12. Gabridge, M. G., H. Gunderson, S. L. Schaeffer, and Y. D. Barden-Stahl. 1978. Ciliated respiratory epithelial monolayers: New model for Mycoplasma pneumoniae infection. Infect. Immun. 21:333-336. 13. Gabridge, M. G., D. Taylor-Robinson, H. A. Davies, and R. R. Dourmashkin. 1979. Interaction of Mycoplasma pneumoniae with human lung fibroblasts: characterization of the in vitro model. Infect. Immun. 25:446-454. 14. Grunberg-Manago, M. 1996. Regulation of the expression of aminoacyl-tRNA synthetases and translation factors, p. 1432-1457. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, D. M. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and E. Umbarger (ed.), Escherichia coli and Salmonella. Cellular and molecular biology. ASM Press, Washington, D C. 15. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infect. Immun. 66:4762-4766. 16. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infec. Immun. 66:4762-4766. 17. Hsu, T., and F. C. Minon. 1998. Molecular analysis of the P97 cilium adhesin operon of Mycoplasma hyopneumoniae. Gene 214:13-23. 18. Kahane, I., and S. Horowitz. 1993. Adherence of mycoplasma to cell surfaces, p. 225- 241. In S. Rottem and I. Kahane (ed.), Subcellular Biochemistry. Mycoplasma Cell Membranes, vol. 20. Plenum Press, New York. 19. Kim, M. F., M. B. Heidari, S. J. Stull, M. A. Mcintosh, and K. S. Wise. 1990. Identification and mapping of an immunogenic region of Mycoplasma hyopneumoniae p65 surface lipoprotein expressed in Escherichia coli from a cloned genomic fragment. Infect Immun 58:2637-2643. 20. Kishima, M., and R. F. Ross. 1985. Suppressive effect of nonviable Mycoplasma hyopneumoniae on phytohemagglutinin-induced transformation of swine lymphocytes. Am. J. Vet. Res. 46:2366-2368. 21. Krause, D. C. 1996. Mycoplasma pneumoniae cytadherence: Unravelling the tie that binds. Mol Microbiol 20:247-253. 22. Krause, D. C., D. K. Leith, R. M. Wilson, and J. B. Baseman. 1982. Identification of Mycoplasma pneumoniae proteins associated with hemadsorption and virulence. Infect. Immun. 35:809-817. 104

23. Mare, C. J., and W. P. Switzer. 1965. New Specieis: Mycoplasma hyopneumoniae, a causitive agent of virus pig pneumonia. Vet. Med. Small Anim. Clin. 60:841-846. 24. Matic, J. N., J. L. Wilton, R. J. Towers, A. L. Scarman, F. C. Minion, M. Walker, and S. P. Djordjevic. 2003. The pyruvate dehydrogenase complex of Mycoplasma hyopneumoniae contains a novel lipoyl arrangement. Gene 319:99-106. 25. Mebus, C. A., and N. R. Underdahl. 1977. Scanning electron microscopy of trachea and bronchi from gnotobiotic pigs inoculated with Mycoplasma hyopneumoniae. Am. J. Vet. Res. 43:1249-1254. 26. Messier, S., R. F. Ross, and P. S. Paul. 1990. Humoral and cellular immune responses of pigs inoculated with Mycoplasma hyopneumoniae. Am. J. Vet. Res. 51:52-58. 27. Minion, F. C., C. Adams, and T. Hsu. 2000. R1 region of P97 mediates adherence of Mycoplasma hyopneumoniae to swine cilia. Infect Immun 68:3056-3060. 28. Minion, F. C., G. H. Cassell, S. Pnini, and I. Kahane. 1984. Multiphasic interactions of Mycoplasma pulmonis with erythrocytes defined by adherence and hemagglutination. Infect. Immun. 44:394-400. 29. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 30. Opriessnig, T., E. L. Thacker, S. Yu, M. Fenaux, X. J. Meng, and P. G. Halbur. 2004. Experimental reproduction of postweaning multisystemic wasting syndrome in pigs by dual infection with Mycoplasma hyopneumoniae and porcine circovirus type 2. Vet. Pathol. 41:624-640. 31. Park, S. C., S. Yibchok-Anun, H. Cheng, T. F. Young, E. L. Thacker, F. C. Minion, R. F. Ross, and W. H. Hsu. 2002. Mycoplasma hyopneumoniae increases intracellular calcium release in porcine ciliated tracheal cells. Infect. Immun. 70:2502-2506. 32. Pollack, J. D. 2003. Central carbohydrate pathways: metabolic flexibility and the extra role of some "housekeeping" enzymes. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, London. 33. Razin, S. 1985. Mycoplasma adherence, p. 161-202. In M. F. Barile, S. Razin, J. G. Tully, and R. F. Whitcomb (ed.), The mycoplasmas, vol. IV. Academic Press, Orlando, FL. 34. Razin, S., D. Yogev, and Y. Naot. 1998. Molecular biology and pathogenicity of mycoplasmas. Microbiol. Mol. Biol. Rev. 62:1094-1156. 105

35. Ross, R. F. 1992. Mycoplasmal disease, p. 537-551. In A. D. Léman, B. E. Straw, W. L. Mengeling, S. D'Allaire, and D. J. Taylor (ed.), Diseases of Swine. Iowa State University Press, Ames. 36 Springer, M., M. Graffe, J. S. Butler, and M. Grunberg-Manago. 1986. Genetic definition of the translational operator of the threonine-tRNA ligase gene in Escherichia coli. Proc. Natl. Acad. Sci. USA 83:4384-4388. 37. Storey, J. D., and R. Tibshirani. 2003. Statistical significance for genomewide studies. Proc. Natl. Acad. Sci. USA 100:9440-9445. 38 Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 39 Thanawongnuwech, R., and E. L. Thacker. 2003. Interleukin-10, interleukin-12, and interferon-gamma levels in the respiratory tract following Mycoplasma hyopneumoniae and PRRSV infection in pigs. Viral Immunol. 16:357-367. 40 Wolfinger, R. D., G. Gibson, E. D. Wolfmger, L. Bennett, H. Hamadeh, P. Bushel, C. Afshari, and R. S. Paules. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comp. Biol. 8:625-637. 41 Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed. 2002. Normalization for cDNA microarray data: a robust composite method for adressing single and multiple slide systematic variation. Nucleic Acids Res. 30:el5. 42 Yogev, D., G. F. Browning, and K. S. Wise. 2002. Genetic mechanisms of surface variation, p. 417-443. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 43 Young, T. F., E. L. Thacker, B. Z. Erickson, and R. F. Ross. 2000. A tissue culture system to study respiratory ciliary epithelial adherence of selected swine mycoplasmas. Vet. Microbiol. 71:269-279. 44 Young, T. F., E. L. Thacker, B. Z. Erickson, and R. F. Ross. 2000. A tissue culture system to study respiratory ciliary epithelial adherence of selected swine mycoplasmas. Vet. Microbiiol. 71:269-279. 45. Zhang, Q., T. F. Young, and R. F. Ross. 1995. Identification and characterization of a Mycoplasma hyopneumoniae adhesin. Infect Immun 63:1013-1019. 46. Zhang, Q., T. F. Young, and R. F. Ross. 1994. Microliter plate adherence assay and receptor analogs for Mycoplasma hyopneumoniae. Infect. Immun. 62:1616-1622. 47. Zielinski, G. C., and R. F. Ross. 1993. Adherence of Mycoplasma hyopneumoniae to porcine ciliated respiratory tract cells. Am. J. Vet. Res. 54:1262-1269. 106

Zielinski, G. C., and R. F. Ross. 1990. Effect of growth in cell cultures and strain on virulence of Mycoplasma hyopneumoniae for swine. Am. J. Vet. Res. 51:344-348. 107

CHAPTER 6. GENETIC ANALYSIS OF FIELD ISOLATES BY MICROARRAY

A paper to be submitted to Infection and Immunity

Melissa L. Madsen,1 Dan Nettleton,2 Erin L. Strait,1 Eileen L. Thacker,1 and F. Chris Minion1

Departments of Veterinary Microbiology and Preventive Medicine1 and Statistics2, Iowa State University, Ames, IA

Abstract Mycoplasma hyopneumoniae is the causative agent of porcine enzootic pneumonia and a major factor in the porcine respiratory disease complex. A clear understanding of the mechanisms of pathogenesis does not exist although it is clear that M. hyopneumoniae adheres to porcine ciliated epithelium by action of a protein called P97. Previous studies have shown that variation within different isolates of M. hyopneumoniae exists in not only the gene encoding the cilium adhesin P97 but also in other genes, however, the extent of genetic variation among field isolates is not known. Since M. hyopneumoniae is a worldwide problem, it is reasonable to expect that a wide range of genetic variability may exist given all of the different breed and housing conditions. This variation may impact the overall virulence of a single isolate or strain. Using microarray technology, this study examined potential variation of eight field isolates in comparison to strain 232 on which the array was based. Genomic DNA was obtained, amplified with TempliPhi™, and labeled indirectly with Alexa dyes. Post genomic hybridization, the arrays were scanned and data analyzed using a mixed linear statistical model. Results indicate that genetic variation could be detected in six of the eight field isolates but in different loci suggesting that variation occurs throughout the genome. Thirty-six of forty-six loci were hypothetical genes.

Introduction Genetic variation is thought to occur among bacterial species as a survival mechanism in both adverse environmental and host niches. A natural consequence of evolutionary and environmental pressures, genetic variation in pathogenic species results in changing 108 phenotypes. Genetic variation can occur by three general mechanisms, local nucleotide sequence changes, intragenomic recombination resulting in reshuffling of genome sequences, and acquisition of foreign DNA (1). It can also occur vertically and horizontally. Vertical transmission refers to passage of genetic material to siblings through cell division and its accompanying replication mistakes (point mutations, inversions and spontaneous deletions). Horizontal transmission involves the acquisition of new genetic material. This could occur among closely related species by transformation of native DNA, transduction by phages or by conjugation mechanisms, giving rise to organisms with subtle changes in phenotype. Alternatively, it could occur between dissimilar species by similar mechanisms that result in dramatic changes in phenotype. For example, pathogenicity islands are thought to arise by uptake and insertion of large DNA segments that encode large blocks of genes related to a virulence phenotype (7, 18). While it is clear that numerous species of bacteria have acquired large segments of DNA (19), there is no evidence for gene acquisition in mycoplasmas. Mycoplasmas are cell wall-less bacteria that are thought to be the smallest organisms capable of self-replication. Their genome sizes range from 580 kilobases to over 1,700 kilobases (15). Their small genomes do not restrict their ability to generate high rates of diversity, however. This type of variation is not a consequence of environmental signals, but rather occurs through random events. There are numerous examples of both small sequence changes and recombination to introduce genetic variation in mycoplasmas (28). This variation is usually expressed by the generation of new chimeric surface molecules with high rates of antigenic diversity. The mechanisms by which this occurs include slipped strand mispairing during DNA replication and recombination between homologous sequences. Genetic variation can result in phase switching when it occurs within homopolymeric tracts of adenine in promoter regions (29) or in structural gene sequences (31), or by DNA inversion (12, 20). The generation of chimeric genes by intragenic recombination also occurs (11). There is no evidence that any of these mechanisms are operative in Mycoplasma hyopneumoniae, however. Analysis of the M. hyopneumoniae genome sequence failed to identify families of lipoprotein genes that could undergo phase switching through mechanisms employed by other mycoplasmas for surface variation (14). Mycoplasma hyopneumoniae is the primary agent of porcine pneumonia (16). There is increasing evidence that M. hyopneumoniae has a predisposing influence on other infectious agents (21, 22, 27). Genetic variation is known to occur in M. hyopneumoniae (2, 3, 8, 24), but there are few studies examining the extent of variation within field isolates at the molecular level (23). Phenotypic variation does occur within M. hyopneumoniae as described by Young et al. within the context of protein immunoblotting (30) and in some cases within 109 specific genomic regions (24), but no studies have been reported that examine genetic differences in field strains of M. hyopneumoniae within genes on a global basis. This is due to the difficulty in isolating and cloning M. hyopneumoniae from field samples and adequate tools have not been available until recently (14). The studies reported here examine genetic variation in M. hyopneumoniae on a genome- wide basis using microarray technology. The arrays were based upon the genome sequence of strain 232. Our results with eight field strains show that microarrays can be used to examine genetic diversity and that six of the strains of M. hyopneumoniae vary in at least one genetic loci. Two of the strains could not be differentiated from strain 232 using this approach.

Materials and Methods Mycoplasma strains and culture conditions Pathogenic M. hyopneumoniae strain 232, a derivative of strain 11, was used in this study (13). Eight field isolates designated 29024, 1879, 27533, 29726A, 27644, 426, 27604B, and 27008 were cultured from case studies from the Iowa State University Veterinary Diagnostic Laboratory (Ames, Iowa). All M. hyopneumoniae strains were grown in Friis media as previously described (6) and are from in vitro passage less than 15. Cultures consisted of 125 ml of Friis media in 250 ml Erlenmeyer flasks incubated at 37°C with slow agitation until the culture reached mid log phase as indicated by color change and turbidity. Mycoplasmas were pelleted by centrifugation at 24,000 x g, and the cell pellets were stored at -70°C until the chromosomal DNA was isolated.

Microarray The M. hyopneumoniae microarray consists of polymerase chain reaction (PCR) products (probes) spotted to Coming UltraGAPS™ glass substrates (Coming, Inc., Big Flats, N.Y.). Ninety-one percent (632/698) of the total open reading frames (ORFs) of strain 232A are represented on the array as PCR products of 125-350 base pairs in length. Each product is a unique sequence even within paralogous families as described by Minion et al. (14). The 66 missing ORFs are due to an inability to design suitable primers (12 ORFs) or failed PCR reactions due to incorrect product size, multiple bands or no product (54 ORFs). Only ORFs greater than 125 base pairs are represented on the array and no tRNA or ribosomal RNA sequences were included. The primer design, array construction and validation have been described (Chapter 2). Each slide was divided into two regions (upper and lower), and each 110 region contained the full array of spots, printed in triplicate in a noncontiguous well-spaced format. This design allowed two independent hybridizations simultaneously to reduce variation due to slide interactions.

Experimental design TempliPhi™ amplified DNA samples from field isolates were compared to control strain 232 using a two-color experimental microarray design. Independent samples from one isolate labeled with one dye were paired with control samples labeled with the alternate dye; the samples were mixed and hybridized to the microarray. For isolate 29726A, three independent DNA samples were paired with three independent DNA samples from control 232. For two of the three arrays, the control sample was labeled with Alexa 555 dye and compared to the field isolate sample labeled with Alexa 647 dye (Molecular Probes, Inc., Eugene, Ore.). The dye assignment to control and treated samples was reversed for the third array (dye swap). The arrays were hybridized under identical conditions as described below. This procedure was repeated for isolates 1879 and 27533 for a total of four arrays each including two dye swaps, isolates 27644 and 27604B for a total of five arrays each including two dye swaps, and isolates 426 and 27008 for a total of six arrays each including three dye swaps.

DNA isolation DNA was isolated from frozen cell pellets as follows. The cells were first resuspended in 1 ml of TNE buffer (10 mM Tris, 140 mM sodium chloride, 1 mM ethylenediamine tetraacetic acid, pH 8.0), and Proteinase K was added to a final concentration of 70 (ig/ml. The suspension was incubated at 50°C for 5 min, and then sodium dodecyl sulfate was added to a final concentration of 0.1% and incubation was continued at 50°C for 4 h. The suspension was then extracted with an equal volume of 25:24:1 phenol:chloroform:isoamyl alcohol three times, and the DNA was precipitated by the addition of one tenth volume of 3 M sodium acetate and bringing the solution to 70% ethanol as described (17). The DNA pellets were dissolved in nuclease-free water, and samples were quantified and checked for purity using the Nanodrop® ND-1000 Spectrophotometer (Nanodrop, Wilmington, Del.).

TempliPhi™ reactions Field isolate samples yielded low amounts of genomic DNA compared to strain 232 due to their fastidious growth and lack of adaptation to growth media. To overcome the issue of limited quantities of DNA, genomic samples were amplified using the TempliPhi™ 100 Reaction Kit (Amersham, Biosciences, Piscataway, N.J.) according to the manufacturer's Ill protocol. A total of five reactions were combined for each field isolate and strain 232, yielding approximately 5-8 (ig total DNA in each preparation which was subjected to mechanical shearing.

Nebulization The DNA was mechanically sheared prior to labeling to ensure an optimized fragment size for efficient labeling and hybridization. Each amplified sample was added to the modified nebulizer (product #4100, MEDEX, Carlsbad, Calif.) containing 2 ml of sterile 50% glycerol. The nebulizer was modified by removing the plastic cuff, trimming the edge and inverting it during reassembly. The samples were sheared using a 10 psi nitrogen stream for 15 min. The fragment size of less than 1,000 base pairs was optimal for efficient labeling and signal strength. This was confirmed by gel electrophoresis on 1.5% agarose gel.

Target generation and hybridization Targets were generated and purified from mechanically sheared DNA samples using the BioPrime® Plus Array CGH Indirect Genomic Labeling System (Invitrogen Corp., Carlsbad, Calif). A set of 129 open reading frame-specific hexamer oligonucleotide primers (Chapter 2) was used to generate amino-allyl modified DNA targets. These targets were then labeled with either Alexa Fluor™ 555 Reactive Dye or Alexa Fluor™ 647 Reactive Dye (Molecular Probes, Inc.) according to the experimental design. Following purification of the fluorescently labeled cDNA per manufacturer's instructions, samples were dried in a vacuum centrifuge and then resuspended in 10 pi Pronto! cDNA/long oligo hybridization solution (Coming). Targets were denatured at 95°C for 5 min and centrifuged at 13,000 x g for 2 min at room temperature. Labeled targets from one 232 control and one field isolate were then combined, pipetted to an array, and covered with a 22 x 22 mm HybriSlip™ (Schleicher & Schuell, Keene, N. H.). Slides were placed in a Coming hybridization chamber and incubated in a 42°C water bath for 12-16 h. Slides were washed according to Coming's UltraGAPS™ protocol and dried by centrifugation.

Data acquisition and normalization Each array was scanned with each dye channel using a ScanArray Express laser scanner (Applied BioSystems, Inc., Foster City, Calif.) under varying laser power and PMT gain settings to increase the dynamic range of measurement (4). Images were analyzed for spots and signal intensities quantified using the softWorRx Tracker software package (Applied Precision, Inc., Issaquah, Wash.). Spot-specific mean signals were corrected for local 112 background by subtracting spot-specific median background intensities. The natural logarithm of the background-corrected signals from a single scan were adjusted by an additive constant so that all scans of the same array-by-dye combination would have a common median. The median of these adjusted-log-background-corrected signals across multiple scans was then computed for each spot to obtain one value for each combination of spot, array, and dye channel. These data for the two dye channels on any given array were normalized using LOWESS normalization to adjust for intensity-dependent dye bias (5, 26). Following LOWESS adjustment, the data from each channel were adjusted by an additive constant so that the median for any combination of array and dye would be the same for all array-by-dye combinations. The normalized values for triplicate spots were averaged within each array to produce one normalized value for each of the 632 probe sequences.

Data analysis A separate mixed linear model analysis was conducted for each probe sequence using the normalized data (25). Each mixed model included fixed effects for field isolate differences (isolate versus control), slide region (upper versus lower), and dye (Alexa 555 versus 647) as well as random effects for slide and slide-by-region interaction. A /-test for differential expression across treatments was conducted for each probe as part of the analysis. A />value was reported for each probe tested.

Results TempliPhi™ reactions Genomic DNA from all mycoplasma isolates were subjected to amplification by TempliPhi™ because of low chromosomal DNA yields in several of the field isolates. Figure 6.1 shows the results of an amplification reaction before and after mechanical shearing and resolved on a 1.5% agarose gel. The optimal time of nebulization (15 min) and nitrogen stream pressure (10 psi) was determined empirically by taking samples at various time points during the shearing process and analyzing them by electrophoresis. All amplified DNA preparations were then sheared using that time point and nitrogen pressure, and the fragment size was confirmed by electrophoresis as shown in Figure 6.1 prior to labeling. To determine if there were any effects on signal strength due to the TempliPhi™ reaction, strain 232 nonamplified DNA was compared to TempliPhi™ amplified DNA from the same batch of chromosomal DNA in a dye swap experiment. Equal amounts of DNA were sheared and added to the labeling reactions. The chromosomal DNA sample was mixed with an 113

amplified DNA sample (alternate dye label) and hybridized to one array; a dye swap mixture was hybridized to the array at the opposite end of the same substrate. The probe signal intensities were quantified, and these values after background subtraction were compared using correlation analysis. Pearson's correlation coefficient was R=0.9803 (Fig. 6.2).

Figure 6.1. Analysis of TempliPhi™ amplified and nebulized mycoplasma chromosomal DNA. Left, molecular weight markers (arrow indicates 1,000 base pairs); Right is TempliPhi™ amplified DNA; Center is nebulized, amplified DNA.

Microarray studies Data from each of the field isolate replicates was used in the statistical analysis. Some amount of error was attributed to the dye, but the effect was not significant in the experimental model as analyzed by ANOVA (data not shown). Statistical analysis indicated that 47 genes for the combined field isolates had significant differences from the control strain 232 at ap-value <0.05. The results are presented in Table 6.1. The field strains differed in the number of genes that varied with the control strain 232. The strain with the most variation was strain 1879 with 25 loci. Two of the strains, 27533 and 27604, showed no variation at/K0.05. Only a single locus, mhp024, showed differences in more than one strain, strains 426 and 1879. The majority of the genes showing differences (35/47) were hypothetical with no known function. 114

R= 0.9803

1

1 O

12 3 4

TempliPhi™ Amplified DNA

Figure 6.2. Comparison of mean log signal intensity values of genomic vs TempliPhi™ amplified chromosomal DNA. Data represent Log mean intensity values following background subtraction. Correlation analysis was performed and Pearson's coefficient was R=0.9803. 115

Table 6.1. Genetic variation of M. hyopneumoniae field isolates compared to strain 232 with /?<0.05*. Field Isolate ID Gene Description p-value 426 mhp024 UH Unique Hypothetical 0.0264 mhp025 CH Putative ABC Transporter ATP-Binding 0.0291 mhp435 CH Conserved Hypothetical 0.0262 1879 mhp024 UH Unique Hypothetical 0.0287 mhp027 UH Unique Hypothetical 0.0089 mhp041 recA RecA Protein 0.0250 mhp044 UH Unique Hypothetical 0.0016 mhpl57 nrdl Ribonucleotide Reductase 0.0491 mhpl58 nrdE Ribonucleoside-Diphosphate Reductase 0.0236 Alpha Chain mhp338 UH Unique Hypothetical 0.0118 mhp342 UH Unique Hypothetical 0.0123 mhp348 UH Unique Hypothetical 0.0014 mhp398 UH Unique Hypothetical 0.0269 mhp404 CH Conserved Hypothetical 0.0463 mhp446 CH Conserved Hypothetical 0.0060 mhp452 CH Conserved Hypothetical 0.0484 mhp522 UH Unique Hypothetical 0.0153 mhp526 CH Conserved Hypothetical 0.0094 mhp527 CH Conserved Hypothetical 0.0290 mhp529 UH Unique Hypothetical 0.0303 mhp530 CH Conserved Hypothetical 0.0393 mhp531 CH Conserved Hypothetical 0.0089 mhp535 UH Unique Hypothetical 0.0218 mhp538 CH Conserved Hypothetical 0.0292 mhp639 CH Conserved Hypothetical 0.0168 mhp650 CH Conserved Hypothetical 0.0248 mhp658 UH Unique Hypothetical 0.0114 mhp692 CH Conserved Hypothetical 0.0240 27008 mhp068 UH Unique Hypothetical 0.0059 mhp236 CH Putative Chromate Transport Protein 0.0021 * UH= unique hypothetical; CH=conserved hypothetical; strains 27533 and 27604 are not included because they had no differences at ^<0.05. 116

Table 6.1.Continued. Field Isolate ID Gene Description p-value 27644 mhp278 clpB ATP-Dependent Serine Proteinase, Heat 0.0253 Shock Protein mhp279 CH Conserved Hypothetical 0.0457 mhp331 CH Conserved Hypothetical 0.0059 mhp436 CH Conserved Hypothetical 0.0388 mhp473 CH Conserved Hypothetical 0.0277 mhp484 CH Conserved Hypothetical 0.0498 mhp488 Phosphoglycerate Kinase 0.0462 mhp497 asnS Asparaginyl-tRNA Synthetase 0.0236 mhp520 pepF Oligoendopeptidase F 0.0112 mhp660 CH Conserved Hypothetical 0.0234 27926 mhpl35 rpsT 30S Ribosomal Protein S20 0.0391 mhpl56 nrdF Ribonucleoside-Diphosphate Reductase 0.0257 Beta Chain mhpl60 UH Unique Hypothetical 0.0409 mhp229 gmk Guanylate Kinase 0.0486 mhp361 CH Conserved Hypothetical 0.0220 mhp619 CH Conserved Hypothetical 0.0399 29024 mhpl36 CH Conserved Hypothetical 0.0468

Discussion Mycoplasmas have shown a surprising ability to undergo rapid antigenic variation, which is primarily associated with surface lipoproteins (28). This variation arises from mutations at homopolymeric tracts of adenines and repeated sequences generated during DNA replication and by recombination within homologous sequences in the genome. This type of variation is random and is thought to support survival of these organisms during innate and adaptive host immune responses. This high rate of variation is associated with specific genes and gene families, but mycoplasmas are also thought to evolve at high rates from random mutations as a consequence of inadequate proofreading and DNA repair systems (33). Thus, it is not surprising that mycoplasmas show a significant level of genetic variation (9). Previous studies have shown that different stains of M. hyopneumoniae vary in their virulence potential (32). Genetic variation within M. hyopneumoniae is not well understood, 117 however. Mycoplasma hyopneumoniae does not contain the lipoprotein gene families found in other mycoplasma species that undergo high rates of diversity (14). It is possible that M hyopneumoniae, like M. pneumoniae, is more "stable" and generates most of its diversity through spontaneous mutations randomly distributed in the genome. In some instances, these changes can be observed using randomly amplified polymorphic DNA analysis (2, 23), but no studies have been reported in mycoplasmas that examine variation systematically throughout the genome. In unpublished studies, one locus was identified that displayed significant sequence variation in two of the field isolates used in this study, isolates 426 and 1879 (Strait et al., unpublished). This chromosomal region involved mhp024 and included both a deletion and sequence variation. The region was identified using a nested PCR test that failed to identify these two strains with the inner primer pair (10). It is significant that the present studies confirm the variation within mhp024 in those two strains since the regions containing this sequence variation are represented on the array (Table 1). Our data indicate that the M. hyopneumoniae microarray can identify genetic variability across the M. hyopneumoniae genome rapidly and accurately. Further confirmation of these changes will require sequencing the indicated genomic regions. These data can be used to improve diagnostics by screening more field isolates and eliminating the variable genes from consideration for PCR targets. Ideally, the PCR target should be homogenous across multiple field isolates. In addition, the arrays can be used to screen other mycoplasmal and bacterial species to enhance the specificity of the PCR target sequences for M. hyopneumoniae by eliminating those open reading frames that are cross-reactive. In summary this microarray is a powerful tool for genomic analysis, which has potential for improving diagnostics.

Acknowledgements We would like to thank Nancy Upchurch and Barb Erickson for assistance with mycoplasma cultures. Monica Perez contributed to the construction of the microarray. Mike Oneal assisted with the TempliPhi™ reactions. Mike Carruthers assisted with spot finding. Funding for this project was provided in part by the National Pork Board.

References 1. Arber, W. 2000. Genetic variation: molecular mechanisms and impact on microbial evolution. FEMS Microbiol. Lett. 24:1-7. 118

2. Artiushin, S., and F. C. Minion. 1996. Arbitrarily primed PCR analysis of Mycoplasma hyopneumoniae field isolates demonstrates genetic heterogeneity. Int. J. Syst. Bacteriol. 46:324-328. 3. Dubosson, C. R., C. Conzelmann, R. Miserez, P. Boerlin, J. Frey, W. Zimmermann, H. Hèani, and P. Kuhnert. 2004. Development of two real-time PCR assays for the detection of Mycoplasma hyopneumoniae in clinical samples. Vet. Microbiol. 102:55- 65. 4. Dudley, A. M., J. Aach, M. A. Steffen, and G. M. Church. 2002. Measuring absolute expression with microarrays with a calibrated reference sample and an extended signal intensity range. Proc. Natl. Acad. Sci. USA 99:7554-7559. 5. Dudoit, S., Y. H. Yang, M. J. Callow, and T. P. Speed. 2000. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. http://www.stat.Berkelev.edu/users/terrv/zarrav/Html/papersindex.html. 6. Friis, N. F. 1975. Some recommendations concerning primary isolation of Mycoplasma hyopneumoniae and Mycoplasma jlocculare, a survey. Nord. Veterinaermed. 27:337-339. 7. Hacker, J., and J. B. Kaper. 2000. Pathogenicity islands and the evolution of microbes. Annu. Rev. Microbiol. 54:641-679. 8. Hsu, T., and F. C. Minion. 1998. Identification of the cilium binding epitope of the Mycoplasma hyopneumoniae P97 adhesin. Infect. Imun. 66:4762-4766. 9. Kokotovic, B., N. F. Friis, J. S. Jensen, and P. Ahrens. 1999. Amplified-fragment length polymorphism fingerprinting of Mycoplasma species. J. Clin. Microbiol. 37:3300-3307. 10. Kurth, K. T., H. Tsungda, E. Snook, E. L. Thacker, B. J. Thacker, and F. C. Minion. 2002. Use of a Mycoplasma hyopneumoniae nested polymerase chain reaction test to determine the optimal sampling sites in swine. J. Vet. Diagn. Invest. 14:463-469. 11. Lysnyansky, I., Y. Ron, K. sachse, and D. Yogev. 2001. Intrachromosomal recombination within the vsp locus of Mycoplasma bovis generates a chimeric variable surface lipoprotein antigen. Infect. Immun. 69:3703-3712. 12. Lysnyansky, I., Y. Ron, and D. Yogev. 2001. Juxtaposition of an active promoter to vsp genes via site-specific DNA inversions generates antigenic variation in Mycoplasma bovis. J. Bacteriol. 183:5698-5708. 13. Mare, C. J., and W. P. Switzer. 1965. New Specieis: Mycoplasma hyopneumoniae, a causitive agent of virus pig pneumonia. Vet. Med. Small Anim. Clin. 60:841-846. 119

14 Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 15 Razin, S., D. Yogev, and Y. Naot. 1998. Molecular biology and pathogenicity of mycoplasmas. Microbiol. Mol. Biol. Rev. 62:1094-1156. 16 Ross, R. F. 1992. Mycoplasmal disease, p. 537-551. In A. D. Léman, B. E. Straw, W. L. Mengeling, S. DAllaire, and D. J. Taylor (ed.), Diseases of Swine. Iowa State University Press, Ames. 17 Sambrook, J., and W. R. David. 2001. Purification of nucleic acids, p. A8.9-A8.l5. In J. Argentine (ed.), Molecular Cloning: a laboratory manual, Third ed, vol. 3. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. 18 Sanderson, K. E., M. McClelland, and S. Liu, -L. 1999. "Stable" genomes, p. 217- 234. In R. L. Charlebois (ed.), Organization of the prokaryotic genome. American Society for Microbiology, Washington, D C. 19 Schmidt, H., and M. Hensel. 2004. Pathogenicity islands in bacterial pathogenesis. Clin. Microbiol. Rev. 17:14-56. 20 Shen, X., J. Gumulak, H. Yu, C. T. French, N. Zou, and K. Dybvig. 2000. Gene rearrangements in the vsa locus of Mycoplasma pulmonis. J. Bacteriol. 182:2900- 2908. 21. Thacker, E. L., P. G. Halbur, R. F. Ross, R. Thanawongnuwech, and B. J. Thacker. 1999. Mycoplasma hyopneumoniae potentiation of porcine reproductive and respiratory syndrome virus-induced pneumonia. J. Clin. Microbiol. 37:620-627. 22 Thacker, E. L., B. J. Thacker, and B. H. Janke. 2001. Interaction between Mycoplasma hyopneumoniae and swine influenza virus. J. Clin. Microbiol. 39:2525- 2530. 23. Vicca, J., T. Stakenborg, D. Maes, P. Butaye, J. Peeters, A. de Kruif, and F. Haesebrouck. 2003. Evaluation of virulence of Mycoplasma hyopneumoniae field isolates. Vet. Microbiol. 97:177-190. 24. Wilton, J. L., A. L. Scarman, M. J. Walker, and S. P. Djordjevic. 1998. Reiterated repeat region variability in the ciliary adhesin gene of Mycoplasma hyopneumoniae. Microbiol. 144:1931-1943. 25. Wolfinger, R. D., G. Gibson, E. D. Wolfinger, L. Bennett, H. Hamadeh, P. Bushel, C. Afshari, and R. S. Paules. 2001. Assessing gene significance from cDNA microarray expression data via mixed models. J. Comp. Biol. 8:625-637. 120

26. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed. 2002. Normalization for cDNA microarray data: a robust composite method for adressing single and multiple slide systematic variation. Nucleic Acids Res. 30:el5. 27. Yazawa, S., M. Okada, M. Ono, S. Fujii, Y. Okuda, I. Shibata, and H. Kida. 2004. Experimental dual infection of pigs with an H1N1 swine influenza virus (A/Sw/Hok/2/81) and Mycoplasma hyopneumoniae. Vet. Microbiol. 98:221-218. 28. Yogev, D., G. F. Browning, and K. S. Wise. 2002. Genetic mechanisms of surface variation, p. 417-443. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 29. Yogev, D., R. Rosengarten, R. Watson-McKown, and K. S. Wise. 1991. Molecular basis of mycoplasma surface antigenic variation: a novel set of divergent genes undergo spontaneous mutation of periodic coding regions and 5' regulatory sequences. EMBO J. 10:4069-4079. 30. Young, T. F., and R. F. Ross. 1987. Assessment of antibody response of swine infected with Mycoplasma hyopneumoniae by immunoblotting. Am. J. Vet. Res. 48:651-6. 31. Zhang, Q. J., and K. S. Wise. 1996. Molecular basis of size and antigenic variation of a Mycoplasma hominis adhesin encoded by divergent vaa genes. Infect. Immun. 64:2737-2744. 32. Zielinski, G. C., and R. F. Ross. 1990. Effect of growth in cell cultures and strain on virulence of Mycoplasma hyopneumoniae for swine. Am. J. Vet. Res. 51:344-348. 33. Zou, N., and K. Dybvig. 2002. DNA Replication, repair and stress response, p. 303- 321. In S. Razin and R. Herrmann (ed.), Molecular Biology and Pathogenicity of Mycoplasmas. Kluwer Academic/Plenum Publishers, New York. 121

CHAPTER 7. GENERAL CONCLUSIONS General Discussion The construction and application of this microarray represents a significant advancement for the field of mycoplasmology. Weiner et al. were the first to report a global transcription profiling experiment in mycoplasmas (11), but their studies involved membrane based arrays, a technology that is falling into less favor and is thought to be inferior to glass substrates and fluorescent signals. The most important difference between membrane and glass substrate arrays is the ability to have a control sample hybridized at the same time as an experimental sample. This allows for direct comparison of the two samples and diminishes error associated with independent hybridizations. The smaller amount of target necessary to visual data is a benefit, especially when minimum samples are available. Smaller hybridization volumes increase the opportunity for probe:target interaction. Glass substrates can accommodate substantially more probes than a membrane. This is especially important when samples are at a premium and all possible gene products can be evaluated on a single substrate. Additionally, the exposure to radioactively labeled targets is completely omitted. This project required the design and construction of microarray s because they were not commercially available for mycoplasmas, particularly M. hyopneumoniae. The construction of the microarray was possible due to the completed genome sequence of M. hyopneumoniae (6). Glass substrates were used for the arrays because of the improvements in technology and detection systems. The construction process entailed a complex series of steps (Fig. 7.1), each of which required careful optimization through trial and error. The process was complicated by the lack of suitable components during the early construction phases, particularly in substrate quality and reproducibility, and PCR product spotting buffers. Post construction, poor mycoplasma RNA yields of initial isolation methods had to be overcome. Several isolation techniques were evaluated for quantity and quality of the RNA. Labeling efficiencies also hampered data acquisition. Whether these complications were due to the AT richness of the genome, especially long homopolymeric tracts of adenines, or problems with manufactured kits was never clearly determined. The bulkiness of the dyes could impede efficient labeling if strings of modified nucleotides were incorporated into the cDNA and competition for those locations made binding an issue. The addition of a second modified nucleotide to the Invitrogen kits seemed to resolve some of the labeling issues. The task of quality assurance also proved challenging. Finding a staining or hybridization method that would give a good evaluation of the array without wasting precious samples proved difficult. Once a suitable system for detecting the quality of the arrays could be ascertained, the studies could commence. 122

G#nc Specific Prime r Design Ï PCH M ir PCR Product Purification f Quantification X ^ &Dfytng J J f Rehydrating & X 1 Rcarc aying Jj

/ centemsien & \ ^ Rahfjitfa-atgd °nacu£te Army -3maî Printing Opt Tiization Hmidy and pm sin to > cqntt; Substrates .f" UV Cross- \ I i iriU1Inn J Optimization oye Selection V linking I l-c "ect vs r PmhyWd'zm:. _a'bn Hi» Opt-nizmon Uireti Laae:"c Laae "c T conditions Probe OptTiizalian Gencfotiorera t arget vW "urrtEaban Hyb'fdization ^ Opt-nizairon J Hyondizato WW Washing & V Drying y " empeisfcire

( Slide Scanning X) —, y

r Spot Finding v* Opt nizalton Saiware \ y fcva. jairan ,r Opt-nizatiori Database Soiware r Data Anal y$k ^i -4 fcva jation EvWsliqn

Figure 7.1. Steps for microarray construction and utilization. 123

Using the microarray to elucidate previously unknown transcriptional and genomic differences proved to be a powerful tool in the studies conducted. Numerous genes of interest have been identified in response to environmental stress, strain variation and field isolates. Additionally, a look at variation across the heat shock and iron deprivation studies has pointed to numerous genes that show both common and diverging responses under both conditions (Table 7.1). Thirty-two of the 53 overlapping genes are conserved hypothetical genes. A paralog of the PI 02 family was also identified in both studies. The remaining genes were involved in translation, stress response, and transport. Of these 53 genes, 6 were differentially transcribed, meaning up-regulated in one treatment group and down-regulated in the other. These genes could be useful in determining possible promoter sequences. One additional conclusion that should be drawn from the culmination of these studies is the identification of potential housekeeping genes. By examining those genes with highly conserved homology across the transcriptional studies with little to no transcriptional difference among treatments or variants, genes can be identified as housekeeping or reference genes in future analyses. Table 7.2 provides a list of 15 genes that could serve as putative housekeeping genes and serve as transcriptional controls in any further applications. For example, an accurate method of quantification using real-time PCR relies on a robust set of genes that do not show genetic variation or altered expression. These genes could provide those controls.

Recommendations for Future Research Now that a powerful molecular tool is available to investigate transcriptional and genetic differences, future experiments can be designed to examine these differences more intricately. In response to environmental stress, additional studies should evaluate the response of M. hyopneumoniae to cold shock and acid tolerance. Studies by Zielinski et al. have shown adherence is impaired significantly by a decrease in temperature (12). Using this global approach, when taken in concert with the heat shock study results, the additional information could outline a more complete picture of how M. hyopneumoniae responds to fluctuations in temperature. M. hyopneumoniae may also find itself in an altered pH environment following the influx of neutrophils into regions of lung containing mycoplasma antigen. It is likely that mycoplasmas experience a change in pH as the lung disease progresses. Initially, colonization occurs at the upper respiratory surface on ciliated cells, but eventually, these cells are lost, mycoplasmas are trapped in the lung surfactant and in fluids at deeper regions and lymphocytic infiltration ensues, causing additional tissue damage. In these inflammatory loci and areas of active lymphocyte activity, the microenvironment pH 124 drops (1,4, 5). These changes in the microenvironment have profound implications on immune cell function (3, 5, 9), but they could also cause changes in the pathogen as well. A comparison of acid tolerance and in vivo response could have correlated data and help define how the organism survives in the host respiratory tract. Another environmental factor that might be examined is M. hyopneumoniae''s response to carbon dioxide. Mycoplasmas are normally grown under conditions where CO2 concentrations are quite low. In the lung, one can expect concentrations of CO2 to reach 5% and under pneumonic conditions, possibly higher, i.e. 6%. A study of normal broth grown organisms and those cultured in 6% CO2 might give insight into transcriptional changes M. hyopneumoniae undergoes in vivo. Some bacteria have been shown to respond to hormones, possibly through a mechanism involving a structural mimic. For Escherichia coli 0157:H7 this appears to be in a quorum sensing system (7). There have been some studies in Mycoplasma pulmonis infections in mice, but these appear to affect the host more than the mycoplasma although that could not be ruled out (2, 8). The effect of hormones on the mycoplasma was not tested directly. It is not unusual to have high concentrations of epinephrine in the lung during infections. Perhaps M. hyopneumoniae can respond to epinephrine to facilitate its persistence in the lung. This kind of study would be simple to perform by adding epinephrine to cultures at concentrations that simulate the levels expected in the lung. Studies in other bacterial organisms are indicating that the ribosomal genes are contributing to variation in transcription and translation through conformational changes, acting as chaperones and other possible unknown functions (10). The microarray constructed for these studies did not include any of the ribosomal or transfer RNA genes. A subarray constructed with these genes may give insight into any possible regulatory effects the ribosomal genes may have on transcriptional variation. Having analyzed field isolates from across the country as well as inter-continental isolates could help determine the true genetic variation present in M. hyopneumoniae. This would be both a challenging and costly undertaking, but would provide relevant information especially for identifying conserved genes for diagnostic purposes. The array might also prove useful in epidemiologic studies. Since M. hyopneumoniae is endemic worldwide, if proven vaccine programs could be put in to practice using attenuated strains and the "genotype" established, then any "new" infections could be detected, characterized, and the origin of infection identified. 125

Table. 7.1. Genes with significant (p<0.05) transcript values in both heat-shocked and iron- depleted studies. Gene ID1 Gene2 Description p-value H p-value I H3 I

mhp005 CH 0.00329 0.01495 - -

mhp007 UH 0.00073 0.01435 - -

mhpO13 UH 0.00648 0.04918 - -

mhp023 CI I Putative ABC Transporter ATP- 0.00407 0.01478 - + Binding Protein

mhp030 go# Glu-tRNA(Gln) Amidotransferase, 0.00324 0.00626 - - Subunit B mhp060 fjh Signal Recognition Particle Protein 0.00860 0.04635 + + (fifty-four homolog)

mhp064 CH 0.00071 0.01959 + - mhp071 UH 0.00192 0.03527 + + mhp072 dnaK Chaperone Protein dnaK 0.00002 0.03556 + + mhp078 lepA 30 kDa GTP-binding protein LepA 0.00399 0.04766 + +

mhp085 rpsL 30S Ribosomal Protein S12 0.02290 0.00021 - -

mhp087 CH 0.02272 0.00312 - -

mhpl 17 hpt Hypoxanthine Phosphoribosyl 0.02385 0.00965 - - Transferase mhpl28 serS Seryl-tRNA synthetase (Serine-tRNA 0.00423 0.04630 + + Ligase) mhpl44 CH 0.00800 0.01198 + + mhpl46 rbsC ribose ABC transporter 0.03053 0.04079 + + mhpl 47 rbsA Ribose transport ATP-binding protein 0.00527 0.03613 + +

mhpl 50 CH 0.00112 0.02044 + + mhpl51 CH 0.00131 0.00689 + + mhpl52 MC myo-inositol catabolism 0.00413 0.00825 + +

mhpl67 oppD oligopeptide transport system 0.03350 0.00420 - - permease protein

mhp 169 oppB oligopeptide transport system 0.04435 0.00495 - - permease protein

mhpl 76 CH 0.00314 0.00093 - + mhp275 CH PI02 paralog 0.00234 0.00756 + + mhp278 clpB ATP-dependent serine proteinase, 0.00031 0.02153 + + heat shock protein

mhp 3 09 UH 0.04443 0.03562 - -

mhp315 UH 0.02183 0.02743 - - Shading indicates differentially regulated genes between heat-shock and iron depletion.2 CH = Conserved hypothetical; UH = unique hypothetical.3 H = Heat shocked genes;I = Iron deprived cells; (+) up-regulated; (-) down-regulated. 126

Table 7.1. Continued. Gene ID Gene' Description p-value H p-value I H mhp325 CH 0.04769 0.02067 mhp337 UH 0.00201 0.00618 mhp342 UH 0.01677 0.02216 mhp374 CH 0.00102 0.01007 mhp411 CH 0.04943 0.00789 mhp419 CH 0.00050 0.01501 mhp421 ruvA Branch Migration of Holliday 0.01921 0.04102 Structures; Repair mhp427 CH 0.03877 0.03104 mhp443 CH 0.02940 0.04119 mhp447 CH 0.01274 0.01353 + mhp461 CH 0.01567 0.04233 + mhp471 nadE NH(3)-Dependent NAD+ Synthetase 0.04455 0.01145 mhp476 alpD ATP Synthase Beta Chain 0.00551 0.01 109 mhp486 CH 0.00667 0.02187 mhp497 asnS Asparaginyl-tRNA Synthetase 0.00240 0.02807 + + mhp511 46kD Surface Antigen Precursor 0.00219 0.02293 + + mhp526 CH 0.00028 0.01669 mhp528 UH 0.02831 0.01647 + + mhp574 UH 0.01326 0.03069 + + mhp639 CH 0.01639 0.00010 mhp650 CH 0.01800 0.02905 mhp651 ushA 5-Nucleotidase 0.01649 0.00031 mhp653 CH 0.02766 0.02553 mhp672 rpIM 50S Ribosomal Subunit Protein LI3 0.00988 0.01345 mhp679 rpmG 50S Ribosomal Protein L33 0.01292 0.03545 mhp697 CH 0.00676 0.03037 127

Table 7.2. Putative housekeeping genes ID Gene Description mhp142 UH Unique Hypothetical mhp143 CH Conserved Hypothetical mhpl 61 UH Unique Hypothetical mhp236 CH Conserved Hypothetical mhp248 lip 3 lipase-esterase mhp293 UH Unique Hypothetical mhp345 CH Conserved Hypothetical mhp350 CH Conserved Hypothetical mhp357 CH Conserved Hypothetical mhp377 CH hypothetical lipoprotein mhp381 sn-glycerol-3-phosphate transport system permease mhp404 CH Conserved Hypothetical mhp405 CH Conserved Hypothetical mhp483 CH Conserved Hypothetical mhp583 UH Unique Hypothetical

Since the proposed studies outlined in this "future studies" section are all in vitro, the next major step is to examine samples taken from pig challenge studies. These studies would be challenging, but technology is now available to perform them. In addition, the natural host, the pig, provides some advantages not found in other model host-pathogen infection models. Mycoplasma hyopneumoniae cells can be isolated from bronchial alveolar lavage samples in sufficient numbers to use on the array. This kind of study could provide additional clues to the mycoplasma's response to the host environment. It would be instructive to correlate the results from these in vivo grown organisms with those exposed to heat shock or iron deprivation in vitro. Archived lung and bronchial alveolar lavage samples should be tested using the array. Laser capture microdissection could be used to collect M. hyopneumoniae cells that are still bound to the cilia from these samples and compare their transcriptional profiles with in vitro grown organisms or those isolated from alveolar lavage fluids. Would mycoplasmas attached to cilia have the same transcriptional profiles as those released into the lung? This is an interesting and provocative question. Likewise, an in vivo time course study could reflect changes M. hyopneumoniae undergoes to combat host immune responses. Knowing what genes and transcripts are involved in pathogen adaptation 128 and survival might prove important to developing new more effective vaccines by pinpointing better targets. Arrays are not the answer to all questions. While they can be used to interrogate samples for genetic and transcriptional variation, any alterations to phenotype at the translation level would not be detected. With the abundance of hypothetical genes showing significant differences between strains and isolates, a protein array could give valuable insight to post­ radiational modifications affecting virulence and pathogenicity.

References 1. Bidani, A., C. Z. Wang, S. J. Saggi, and T. A. Heming. 1988. Evidence for pH sensitivity of tumour necrosis factor-a release by alveolar macrophages. Lung 176:111-121. 2. Furr, P. M., and D. Taylor-Robinson. 1993. Factors influencing the ability of different mycoplasmas to colonize the genital tract of hormone-treated female mice. Int. J. Exp. Pathol. 74:97-101. 3. Geffner, J. R., A. S. Trevani, F. Minnuci, M. S. Palmero, N. Maugeri, and M. A. Isturiz. 1993. Extracellular pH modulates oxygen-dependent cytotoxic responses mediated by polymorphonuclear leucocytes and monocytes. Clin. Exp. Immunol. 91:164-169. 4. Kraus, M., and B. Wolf. 1996. Implications for acidic tumour microenvironment for neoplastic growth and cancer treatment: a computer analysis. Tumour Biol. 17:133- 154. 5. Lardner, A. 2001. The effects of extracellular pH on immune function. J. Leukoc. Biol. 69:522-530. 6. Minion, F. C., E. L. Lefkowitz, M. L. Madsen, B. J. Cleary, S. M. Swartzell, and G. G. Mahairas. 2004. The genome sequence of Mycoplasma hyopneumoniae strain 232, the agent of swine mycoplasmosis. J. Bacteriol. 186:7123-7133. 7. Sperandio, V., A. G. Torres, B. Jarvis, J. P. Nataro, and J. B. Kaper. 2003. Bacteria- host communication: the language of hormones. Proc. Natl. Acad. Sci. USA 100:8951-8956. 8. Taylor-Robinson, D., and P. M. Furr. 1990. Elimination of mycoplasmas from the murine genital tract by hormone treatment. Epidemiol. Infect. 105:163-168. 129

9. Trevani, A. S., G. Andonegui, M. Giordan, D. H. Lopez, R. Gamberale, F. Minucci, and J. R. Geffner. 1999. Extracellular acidification induces human neutrophil activation. J. Immunol. 162:4849-4855. 10. VanBogelen, R. A., and F. C. Neidhardt. 1990. Ribosomes as sensors of heat and cold shock in Escherichia coli. Proc Natl Acad Sci USA 87:5589-5593. 11. Weiner III, J., C.-U. Zimmerman, H. W. H. Gohlmann, and R. Herrmann. 2003. Transcription profiles of the bacterium Mycoplasma pneumoniae grown at different temperatures. Nucleic Acids Res. 31:6306-6320. 12. Zielinski, G. C., and R. F. Ross. 1993. Adherence of Mycoplasma hyopneumoniae to porcine ciliated respiratory tract cells. Am. J. Vet. Res. 54:1262-1269. 130

APPENDIX A. CIRCULAR REPRESENTATION OF THE MYCOPLASMA HYOPNEUMONIAE GENOME STRUCTURE 131

Mycoplasma hyopneumoniae

Functional Assignments

Amine aod biosynthesis Replication RNAs I I Cellular Processes I I Other Wk Biosynthesis of cofacrors, prosthetic groups and carriers • Transcription • Biosynthesis of polyamines• Central IntermediaryMetabolism • Unassigned I 1 Fattvaodand phospholipid metabolism H Translation O Biosynthesis of sugars H Energy metabolism m Purines,pyrrnidines, nudeosides and nucleotides H9 Transporrtand Binding B6 Cell envelope |__| Hypothetical Circular representation of the M. hyopneumoniae genome structure. The dnaA gene is at position one. Starting from the outside: circle one, replication arrows; circle two, the location of the putative coding sequences on the positive strand; circle three, the location of the putative coding sequences on the negative strand (the color code for the functional assignments are given in the figure); circle four, position and direction of transcription of the P97 (black) and PI02 (orange) paralogs; circle five, scale in bp; circle six, paralogous families of three or more members (colors indicate relationship and not functional assignment); circle seven, predicted lipoproteins. 132

APPENDIX B. PRIMERS USED IN THE MYCOPLASMA HYOPNEUMONIAE MICROARRAY 133

Table A.l. PCR primers for the M. hyopneumoniae microarray construction.

Gene ID Left Primer Sequence Right Primer Sequence mhpOOl TCGATTCGTGAGCTTGAAGG ATCAGATTGCAATATCGCGG mhp002 AATCAACAAACGCGACATCC TTTTCTGGCGAAACTATCAAC mhp003 TAAAAACTGGAACTCCCCCG TCGAAAAATATTTGGTGGCG mhp004 TGGTTCTAGCCCCAGAGAATG CATCAACGCCGTACGAACC mhp005 GATGGAGATTGTCTGGGTGC CTCTGAAGCAGCCGAAAAAG mhp006 TGCAGTTGGCAATCACAAAG ATGAATGTGGCACTTTTCCG mhp007 TGGGCCAATTTCCTGAAC TGAGTGGGGATTTTTAGCCG mhpOOS TTTAATCGACACAGCCGGAC TAAATCATCCATCGCTTCGC mhp009 CAGATTTTGGACTTGCTCGC AAACCATTGCGTGACAAAGC mhpOlO AGCTCGAAAATGAGCAGG CGCTGTTTTCACAATTCGG mhpO 11 AGCTCGAAAATGAGCAGG AGCGCTGTTTTCACAATTCG mhp012 AAAAATAGAAATCGCTCCCCC AAAAACTGAGCCCACCAAGC mhpO 13 GAATCCGCTCATTTTAGCGAC TAATTATCGCCGAGTGGGAC mhp014 GCTCTCCATTTGGATCATGG CAGGGACTTTCAATCTGGGG mhp015 CCGGGATTATTTTATACGGGC TCCCTAGTCCCACCATATTTTTC mhpO16 TC AA AGATT AAC AGTAG AGTCGTGG GCATCATCCTTATATTGCTCCC mhpO 17 TTTTAACAGCTGGGTTTGCG TTCCGCTTCCAAACTTGAAC mhpO 18 TCCGGTTACTAGAATGAAAGCAA CCGCTAATTAAAATTGGGACG mhp019 GCGAATGAAGAGAAAAACCCTT TTGATTTAGTTTTTCACCTTGAG mhp020 AAGCGAATGAAGAGAAAAACCC ACCGCCAGATAAATCTTTTAC mhp021 GCTCAACTCAAATGTTCAGGG TCGGCCTTGTAAATCGTTTG mhp022 GGGTTGAGGCTGTTAGCAAG TTTTGTGCCAGGCTTGAATC mhp023 GCTCAACTCAAATGTTCAGGG TCGGCCTTGTAAATCGTTTG mhp024 CGCTTCTCGGAGCAAAATC GTATCTCCAACCCGTACCCG mhp025 GGCCTTAGTGAATCTGGTGG AAACAACCACTCCCGATG mhp027 GAACCTCAGGAATTGGCTCC TTCTGCGTGGCGTTAGAATC mhp028 AATGTCGAAGGAAACTCGATG AAATCCAAAAGTACGGCTATC mhp029 TTCACCGAATTTACGCGG CAATCAATGGAACCCCAC mhp030 AGAAAATGGATCACTTCGTGC GATGGGGTAAGTGGAACCTG mhp031 TGGGGTATTATGGCATTCTGAC AAATAGCCGCTACCAAATCAAC mhp032 GCCACTTTTTGGTGAGGATG ATTGAGACAAACCATTGCGG mhp033 TGACTGATCTCACGTCAGCG AAAAGGAGGCTTTTCATCGC mhp034 TTGCCCAGGAATGAAAAATG GTTGCATCGATCACTTCGC mhp035 GCAATTAATGGCTTTGGACG AGGCAAGGACTTTGGGTCTC mhp036 TCAACATTTAGAGGCGGGAG TGCAGCACCTGTTGATGAAG mhp037 TGTACCCCCTGTTTCACCTG CCGGTTTTATTCCCCAAAAC mhp038 ATTGGTCATAAAAGCGAGCC TCCTGCTTGACATAATGGGC mhp039 TGATAAGCGGAAATGCGG TTGCCTCATTTTGTTCAATGG mhp040 GTTGATAAGCGGAAATGCGG TTGCCTCATTTTGTTCAATGG mhp041 AAAGCGCTCCGAAAAATCAC AGAATTCCAAGTTCCTCGCC mhp042 ATCATTCCGCGAGGGTAAC CGAAAGGAATGATTTGAGGG mhp043 GATAATATCGTTTCGCCGGG GACCTCCTCGACCTCCTTTG mhp044 CAAAAGACAAAGCCGAGTGG ATCGCCGAAACTTCCTCTTC mhp046 TTTCCAAATATTTATGCCAACGA AAATTGACCGTTCCGAAAGC mhp 047 GCAGGCATT AT GCAGGTTTC CGGAAAC ACCGT CT AAGATTTG mhp048 TCTTGCATTAGCGTTTTGGC AGCACAAACCGGGAAAATTG mhp049 TTAGGTGCAGGACTGGCG TTGTGCTTCAGGATTACGGG mhp050 GCAATCGATAATGCGCGTC AGCAATCAGACTTTGCTGGAG mhp052 AGCATGCAATTGCCTATCG CCAACCCGTGAGACTGAAAG mhp053 AAGCATGCAATTGCCTATCG CCAACCCGTGAGACTGAAAG mhp054 TTAGATCGAGGCATTGCTGC TTCTGCGCTACGAAGAATGG 134

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp056 GAATCAAGTCCTTATTCAGTGACAG AATTTGAGGAGGAAATCACACC mhp057 AAACCGCGGAATGTTTCTC AGCACAACCAAAATGTTCTC mhp058 TATTAGCGAGATTCCAGCGG TGTTCGGACAGCTTGTTCTTC mhp059 AATCAGATCGAATTGCCGC TTTTAAAAACGCGACGCAAG mhp060 GTTGGCCTTCAAGGTTCTGG TCTTCGTTGATTGAAAGCCG mhp061 ACAGGCATCTGGACACCTTG AAATACCCTGAGCTGTCTCCG mhp062 GTTGGATTCTCTGGTCGCTG ATCGAGCGCGAGTAGAACTG mhp064 TTGGGATAAGTGTGGCTGAC TCAAAATCGAATCCACACTGC mhp065 ACTTGCAGTTATTGGTGGGG GCGAAGCCACTGATTCTTTG mhp066 ACTTGCAGTTATTGGTGGGG GCGAAGCCACTGATTCTTTG mhp067 TTTTGATGAAGCCTCGCAAC TTCTTGTGGCCTTTCGATTG mhp069 CATATGAATTTTCGCCCGC AAGTATGCAATGCCTGAATGC mhp070 GATGCTGCCAAACCAGGTAG TTTTGACTTTTTGCTGCGTG mhp071 GTTAAACAAAATGACGGCGG CCAAAAACAACAAACTAACTGCAAG mhp072 ATTATCCGGTGGAACCTTCG AGTT CAACGTTAATCGGCCC mhp073 TGGTTTTGGTGGCTCACAAG TTTTTCCACGGCACTTTTTG mhp077 GCCGAAATTTAAATACAAAACCG T AAAACAAGGCC GACAACCA mhp078 GTAAATCGACTTTGGCCGAC TGCCTGAATCCCTTGACTTG mhp079 ACGATCTTTAGCGGCAACTG GGTGGCGGGATCAAATTAAC mhp080 AAAACGCATTCAGGAAGAAGC ATTGGTGGCATTTCAAGTCG mhp081 TTAGCATGCGATAAAAATGGC GGGATTAGGGTACTAGAATCGGC mhp083 CCGTTCGCAATGTTTTAGCC AGTGCGAGAGTTTGTTGTCG mhp084 TCCCGTTCGCAATGTTTTAG GCGT AGTGCGAG AGTTTGTTG mhp085 CAGCAAAAGCACGTATTTCAA TTCCTTTTTCATCAAGCTCTTTTC mhp086 TGTTCGTTTGGAGTCAAATTCA TTCCGATTTCAACGTGGTTT mhp087 TGTTCGTTTGGAGTCAAATTCA TCCGATTTCAACGTGGTTTT mhp088 CAAAACGAAACAATTGAACG AAATATAACTGAGTTCTCGTTGTGC mhp090 CCTAAACGCAAATCTTGAGGTC AAAATGGTGTTTTGCCACTC mhp091 CCGGAGAATCTGGTATTGGG ATTTTCGCGCAAAGGCAC mhp092 GGGGAGTTCAACGCTTTTTG TCCT GAGGAAATTAGTGCCG mhp093 GGTTCTGGCAGCGACATC TTTCAAGCCCCAAAAATC mhp094 AATTGTCGTAGCAGATGCCC TTGTGCGCCTATATGCAATC mhp095 TGCGAGAAAGCTTGCAAAAG CGCTGTTCCTGACGTCATTC mhp096 TTGAATCAAGCCAAATCCG GCGGCGAACTTTATTTGACC mhp097 AT AGCAAAGCAACCCGGAAG TTGGTCTAATGGCCTCATGG mhp098 GAATTGAAAAAGATCCAATTGTTG TTTGGTGAAATTCGATACGTT mhp099 TGCAGGTGATGACAGTAGCG AACCACCTCAAATCGGTGTC mhp 100 TGCAGGTGATGACAGTAGCG AACCACCTCAAATCGGTGTC mhplOl TCAATGAACTTGCAGCAACC TTGAAGAGGCTGTGCCAT AAG mhp102 GCTGAATTACAACAACTAAGCGG AAGCGAGGTAAAATCCAGCG mhp103 GACCACGAAAGGGTATCACG TCCTTAACCTCATCCAGGGC mhpl 04 CGAAGGTCAAATTTACGGGG TGAGCGAATATAGCAACCGC mhp105 GTCTTGGCAAATGAGCTTGC GACAATAGGAAGGTTGGCCTG mhp106 CAAAACATCCAGCTTTTGATCC GGGCGAAACCGAGTTTTAAG mhp107 TTAAGTATTGGCGGCTTTGG CAGATCAAGCCCGAACTTTG mhp108 GCAGACTCAGATGGAAACGG TTAGGCCCAAGAATTACTGC mhp 109 AAATTTCACGGTCTACGGGC GCCTTTTCCTGGGCATTTC mhp110 GGTCATGAACTTTTTCTTTCG AAAAATTTTGCGGCCTCC mhp 111 ATTGGTTTCCAAGACCCACC TTTTAAGTGAAGGCGCATCG mhp 112 AGCCTGAAAAATCCTGAGCC ACCAT CGCTCGTTTAATTCG mhp 113 AAAGCCGGCGAAATAATCC TATCGCAAGCCGAAGAAATC 135

Table A.1. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp 14 CATTCATAAAATGGACCCGC AAAACCCGGTATTGTGGAGC mhp 15 TAATGAAATTCGTCGGCGTG GTTCCTTGGGGATCAAGTCC mhp 16 CAACATGGATCAGCTTGTGC AGCGAT GT CAT CTT CA ACCG mhp 17 AGGTGTGCTTAAAGGTTCGC AGCGTCCGGAACTACAAAGC mhp 18 TTTTATGCGAGGATTCACGG AAACCGGAAAGGACAAAAGC mhp 19 AGAAACGCCATACAATTCGG TTGTGTGAGGACTGCATTACTTTAG mhp 20 AAGACTAATTTGGCCCGCAC AACACCTCTATTCTCCAGGCG mhp 21 TATGCCTTGATTTCCAACG TGC A AGCCGAGT AATTTT AAG mhp 22 TTACCGGTGATTTTGAACTAA AACTTGATTTGGATAATCGGT mhp 23 TAAATTGCACCCACAAAACG TTGGACACGGGAGAGAATTG mhp 24 TTGAATTCCGAGTTCTAAACTACC TTTCATAATCCCAGCGTAAAGC mhp 25 ATCTACCGTCGGCTTTCAGG TTTGTTACTCGTTGGCGGAC mhp 26 AAAGCAATTTTAGCCGGACC ATTTATCCATTTCAGCCCCG mhp 27 GCTGTTTTTATTCCGGGGG AACTTGCTGCATTTCTTGCC mhp 28 CGGAGGCAGAGACACAAAAG TTGAAAATCGCCGCAATAAC mhp 29 TCAAAAAGGCGTAATGACCG ATTTGAGGCATGTTCTCCCC mhp 30 AACGGTTATAAAGGTGTTGAGC TGTGATCCACCTGATTTCCG mhp 31 TTTGAGGATGTCCAAGTGGG CGCCGAGCTAGCAACAGAG mhp 32 AATTTTTGCCATCACATCGG GGATTCCATTCCGGAAAAAG mhp 33 TCTTCGACTGGATTTCGTGC GGTGGCAGTTAGGCGATTTAG mhp 34 GATTCCGCATAAAGTCTGGG TTTGTGAAGTGCCAGGTTCC mhp 35 GCTCGCGCAAGAAATAATG CTGGTGTTGTGTAAGTATTCGC mhp 36 TTCTGAGAAGGACGATGTTGC GTAAATGCGCATCTCCTGGG mhp 37 TAAGATTGCTTCGCAGGCAC ATTCGAATTGACCCTGAGCC mhp 38 TTTCTTTGGCCAAGGTGC TTCATAGCGAAACATTGGGC mhp 39 C ATTAAATGCGGCAATCCAG TTCTGCGATATTTCGATCCG mhp 40 TGTGCTGGGTTCCCTAAAATC CGTTAGTCGGGGCAAACTC mhp 41 TCCAGGGATTTTTGTCCC TGACTTTATCGCTTCTAAGTTCTC mhp 42 CAGAGCGTGGTATGGATTTAGG TTTAGCGCTGCACTTATCGG mhp 43 GAATATGCCGGAATTGGC AAATATTCCTCAGCCCGTTG mhp 44 TCAAATCGGCTTTATGCGTC CCGGCTTGCTGTCATTATTC mhp 45 GCCCAGATTCCATTAGGAGG CTTTTCCATCTGGGCCTTTC mhp 46 TTTATCACTTCGAGGGGCG GAAAGGGCACCT AAAATCGC mhp 47 TCGCTTTCGATCTCACAAAAG TTTTTCGGGATAACGCTGC mhp 48 GGCCTTAGGAGCGAAAGTTG CGCATGAGCAAAATGACCAG mhp 49 CAAGAAGACCCGATCCTGTG CCGATGATCACCAAAGGTTG mhp 50 TTTCACCGTTGGACATAATCG TCGATTGTTGCCATTGAACC mhp 51 GGCGAAGTGATTACCTATCAAG TCGATCCGATCAATTCAAGG mhp 52 TGCAATTGGGCTTAGTCGTC GATTTTGCAAGCGAAGTCCC mhp 53 GAGACCCAAATGGTT AGCCG TGATTGATCATTGCATTGGC mhp 54 TGCTGGTGGGCATTTAAGTC ATAAGCCCGGCAATATGAGC mhp 56 TTTATGGTTGGGGTTCATGC AATCAGGCGAATAACATCGG mhp 57 TAATACCCATCGGTTTGTCG TGGGAAATAATTGTGCCTGC mhp 58 CAGCCTTCAACTCCAAGCAC TCAAGCGCTTCCTCTGAGTC mhp 59 TTATCCAGGGAAGTTTTGCG TTTCCAAATTTCAAGATGCTG mhp 60 GCTTGGGTAATATTGTTGTTGCTC TCTTCAATTCTGGCAAATAAATC mhp 61 GGGATTAGGCTCATTACTCGC GAGGAGAAAACGGAGGTTTG mhp 62 TGTTGTCACAGCAAATCTAGTAAACC AACAATTCTTATTGAAACTATTCCG mhp 64 CAAGCCAAGTTTTTGATGGC CTCGGTGATATCAAAGGGGC mhp 65 AG ACCTACGA A A AGC AGGGC TTGATCCTCGATTGTCTGCC mhp 66 TATCTCTGATCGAATCGCGG CACCTTTTTAGTTCGGCAAAAG 136

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp 167 CAGCTGTCTTCCAGAATCCG AAGCTTGCACGGTTGTATCC mhp 168 TGTGGAAATTGATGGTTGCTC AAACATCGCGACCTTGTGAG mhp169 TAGTTTAAGGCCGCGTGTTG GAAAGTGCAATCGCTGGAAG mhp170 GGCGTTTTCTTGAGGTTAGAC CAGCTATTGACTCGGGTTGG mhpl71 TAATGGATGAGCCGACCAAC TTTTTCCGTGAAGTTGCCTG mhp 172 ACCTCAGATGTGCTTTTGCC ATTCCGGCGGTATCAAAAAG mhp173 TGCTGGCTGCCTTGATTTAG ATTACGGAAGGCAACTTGGG mhp 174 CCGGTTTTGATCGTGTCTATG AAATGGCTCGATCTTCACCC mhp175 TGCCTACCATGAAGCTGGTC TTGGGCCAAGTGAAGACATC mhpl76 TTTACGGGCAAATATCAACG AGCCCCACTAGTCGATTTCC mhp177 ATTGTCGGCCTTGGTAATCC TGTTTTGATTGCTGCCTGG mhpl78 TCAAGGGAAAACAACCCGTC TTGGAGTCAAATTGGAAGGC mhp179 AGCACTTGATCTTGGCATC TTGTTTTGTAGTCAGCCGCTC mhp180 GGCACCCACCTTTTATTTGC ACCGTAACGAATTTGGGTCC mhp 181 TCCCAAAAACTAAGCGGGG CATCAAACCCACCTTCAGGC mhp 182 TAAAAACCGTGATTGAGGGC GCTGTTCAAATGCTTGTCCC mhpl83 AGCAGCCAAAAAGATCCCTC GGCTGAACTTCATCTGGGC mhp184 ACAACCAAGTTTTGGATTTTT GCAGAAAGCGAGATGGCTAA mhp 186 CATCGTCAAATCGATGCTGC ATTCAAACCCCAGCAGGAAG mhp187 AGCAAATACGAAGCCAAAGC TTCATGTCCTAAACGCCCTG mhp 188 CACCAATTTTCGTAGGCGG CATTTGGTAGGTTTCGTGCC mhpl89 TCCAAAAACTAACCGTTCTGAG CATCAATTTTACAAGCGCC mhp190 TATGATCCAAACAGGTCGGC AAATCCGACGGTATTCTCCG mhp 191 CGATGAAGCAATTGCGAAAA TCCAACAAACTGGGGAAAAA mhp194 AATACGGTTATTCGCCGTGAC CTCCATTAAGTCGCCCACAG mhp195 TTCAGCTGCCCAAATTGAAG CGGTAATTTATGTCCGCCAAG mhp 196 TCGTTCCGAGTTATTCACCTT CATTTTCAGCACTCATTTCTTGC mhp197 GGAATTGTCATTAGAACCTCCCC TTCGACTAAACGGAAGTGTTTTG mhp198 CGTGGGTGTAATTCGCAATC CGGACCAAAAACCCTTGTTC mhp199 TGATAAAGGAAAGCAAGGCG CAACTCGGGTAGGAATTGCG mhp200 TAACTTCAAATCGCCTTCGC GAAAGTCCACGGAAATCACG mhp201 GTTCGCGCCTATACTCGTTG TCCTTGGTGTGCTAAAACTCG mhp202 ATTGCTGATATGCTCACCCG TTTGATGCACTTGTATAAACCCG mhp203 TCCTCGAAAGACAATTTGC TTTGTTTGTCGATCCCTTTG mhp204 ATCAAACCGTGAAGACAAGC TTTGCAAATGCTTTAACCCG mhp205 TTGTAAAAGGTGGTCGTCGC TCCGGCAAGTTCGACTACAG mhp206 GCCGGAAAAGGTAAACAAGC TTTTCTGCGCGGCTATTG mhp207 TCCAATCCATCGGCTATCAC TGTTGAAAACGACGGGGTAG mhp208 GCTCCTGGCTCAGGTAAAGG AACTGTTGAAGCGTTCGAGG mhp209 AATTATCACGGTTTTCCGGC TTTGCCTATCCCATGACCAG mhp210 CACGTTTTTAATGCCCAGGA TTTTTCCCGCGATATGACAG mhp211 GAATTGGAAAATCGCTGGC TGACCACGAACAGGTAAGCC mhp212 AATGTTACCGTCGGCATCG TAAATCATTTTCGAGGCGGG mhp213 AATTTCGCTGCTTGAGTCCC GAAAAACGGCGGATAGGTTC mhp214 TCGCCAAGTTTTTCGTTCAC AACAATCCGAGTATACCCCC mhp215 TTATCGGAAATGGGGGTTTG AACCCGGTAGTTTGGAAATC mhp216 TTGTGATGATAGGGCAAGTC CATACTAGGCGATTTGGGTG mhp217 TGGAGGAGGAACAGTTGCAG ATGAGACAACATCCGATGCG mhp218 TATTGGCGCTTCTAAAGGGG AATTTGCTTCGAATTGCGG mhp219 AAGTCCGAAATTAAGCGTTTT CCGGCGTTTGTCTTCGTAAT mhp220 ACGACGTTTGGGACGAATG AATAACTCCGCGGCACCTAC 137

Table A.1. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp221 TTTAAACGCACCTTCAACCG GACGGTGGCCATAACTTGAG mhp222 TTTCAGCGTTTTGATATTCAGG TCAAGTATTTCTCCTTCCTCG mhp223 TCCAATTTCAGGACCGCAG GAAAGAGTACTGCCGGTATTTGTG mhp224 CGAAAAAGAGCAAATAACAGGTG TCATCCTGGTGATCAAAGCC mhp225 AAAAAGCCTACTGTCCTTATTCAG TCAGCGCAAACTTGCAAAC mhp226 TGTCGATACTCCCGGCTTTC TGATCGGATGGATAAGCAAAC mhp227 ATGGTAGCATTGGGCGTTAG TTGCATCAGCATTTTTGG mhp228 TTCCGGAAATTATCGTGGTG TGGGGAGGATCTAGGAAAATG mhp229 GGAGAGACGAAATTAACGGAG ATTCAATGGAAGGAGGGAGG mhp230 TTCGTCCAAAAGTAGCCAAC AAAACTGGTTTTTCCGACCC mhp231 TCCCGAAGAGCAAATTAAGC TCCTTAATGCCACCATCAAC mhp232 ATCCAGGAAGCAGTTGGG GCAGCTTGAGATACAGAGCC mhp233 TTTGATGGCGGAGAAGGAG TTTCGATATTTTCGCGCTC mhp234 TGCCAAAAATCCATTTAGCC AATTTCTGGGCCTGCATTTC mhp235 TGGTGAGGTTCTCGGACTTG TTTCCCCTTTTTGGCCTC mhp236 AAAATTTTTCGGCCGGTG GGTTGAAGGCAAGTTAGGGC mhp237 GCCTTCGGTTGTGCAAATG GGTACAAGCGCTAGATTATAAGGG mhp238 CCCAGCACAAAAGTCCATC GCATCTTTTTCAAGTGAGCTG mhp239 CCGAAATCGGCCCTAATACC TGGAATGCCGGTAGATTCAC mhp240 TTCGGAAATGCTTGTCTATGAAA TTCCATATTTGTTATATAAGTGGCAA mhp241 TGGGCCTTATCGTCAGTCAG TCCGCGAACTAGGTCATTTC mhp242 CAGGTTTTATCGGTCTTGCG AAAATTCAACCATCGTCGGC mhp243 GCTTTAGCAGGCGGGTTTAG AATACCGGTGGGAATATGGG mhp244 AAGACCAAAGCAAACCCTGG TACCTGACCCGGACTGATTG mhp245 AGCGGGAAGACCACAAAAAC TCACCATGTTCACCCATCAC mhp246 GGACTTAACGGTAAAGCGGG TAATCTGCATTGTTCCGCC mhp247 ACGGTGCCAACTGGCTTATC AGGCAATTCCAATCGACCAG mhp248 GTCGGACATTCGATGGGAG GCTTGCTCAATTTGGTCC mhp249 ATGGCAAAGACCAAAGCAG TTAGCCATTACTACTGTTGCG mhp250 AAATCGACAAAAACGAGGGAG TGTAAATAGTTGGCGGTGCC mhp251 CGGGAATCTTGAAAACTGGG AATTTTGCGGAAAGCCATAG mhp252 ATCCCATGATACTGAAGGCG TTGAATGGAGGTAACGCTGG mhp253 TGGAGGAGCTGGAGTTCTTG TTAACAGTGCTTCCATCCGC mhp254 TGGCGAGTCAGCTTTTATGC TTGGTGAGTTCCCGACAAAG mhp255 GGAATTATGATTGCCCGTGG CCTGAACTGACTCGATTGGG mhp256 GCGGTCGTGAAATTACTAGGG GCGATCACGTTCAAGATGC mhp257 ATGGCCATGCTTATCGTTCA GTTGATTTTCTTGATTGACGTT mhp258 AGCAAAAGGCTATTGAGGGC TCACTTTGGTGATGAATTGCC mhp259 AGTGGAAAT AATCGCTCGCA TTTTTCCTGAGGCTGTTTTTAGTG mhp260 CAGGCGGGTTAAAAGTTTCAG TCGCATTAAAACGCGTGTATC mhp261 TTGGATTGATAGGGTTCAAATG TCGTCAAGTAAAATTGCGACG mhp262 CGGGATCAATAGGGTGTGTG TAACGGTGCAATCAAACTGG mhp263 AAGGGAGACGCAAGACAAGC AAGTCCCATTGTGGATGCAG mhp264 ATTCTCGCGTTGTTCTTTGG TTCAAGAGCCCTAACACCCC mhp265 TGCCCCTAACTTTGGTGAAG TAGCGCTTCGTAAAACTCGC mhp266 ACTTGCGGATGGAAAAGCAC ATCAACAATTGCTCCAGCTC mhp267 CTTTATGAGGTCCACGTGCC AATGCAACCATAAATTGCCG mhp268 ACCGCAGACAAAATACAACG TTGGCATCCTTTTGGCTAAC mhp269 CTGCCATAGGATCAAGTCGC ATGAGAATTCGCCGTATCCC mhp270 ATTCCGACTGGAATTCACCC AAGAATCTCAAAATCGGGGC mhp271 GAAAAAGCGGCAGTATCCG CCTTGACCTTTTTCATTGCG 138

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp273 GATTTTCCGCACGGGATAG GAAAACCAAGATAGTACTCCCATAC mhp274 TAAAAACCGTGATTGAGGGC GCTGTTCAAATGCTTGTCCC mhp275 TGATATTGTTGATCTAGTCGACGG GGGCTTAC ACCTTCTTT GGC mhp276 GGAAATACAAGCGTTCAGATTG GCAAT AAAACCGCGCATTG mhp277 TTCAGGAGCCTTTACTGGGG CGTTGCTGTTTTGCCAGTTC mhp278 GGAGACTACCAAAAAGCCGC TTAAACCGCAAAATCGCATC mhp279 AAGCGAAGTCGAACTGCTTG AATTGGGTCTTCCGCTTGAG mhp280 AATGTCTGAAACACAGCCCG CGGCTTGTTCATTTTCTTGC mhp281 AATTTTTCAAGGCCGAGCAG TGGCTCAGCATT AAATTGGC mhp282 ATCAGAAGTGGTGGCCTTTC GAGTTGTCCATCAGGAGCG mhp283 TCTTCATTAATTCAGCCAGGG CCCAGTTGCAAGACCAAAAC mhp284 TGCCAACGAAATTGAAAG TTCAAGGTATGCTTGTGTTACTTG mhp285 AAAAAGCGAGCCCGTT AT G TTCGGCTGTTTTTGCCAC mhp286 TGGCTATTGAGATGAAATCCAAA CGCGAATTAGTCCTGGATTATG mhp287 GAATATCGAATGTGCCTCGC TGT AATAACCGAAATTCGCCC mhp288 TTTAATCTGCAAACTGGGCG TCGTTGAGCTTCTCCTCCTG mhp289 AAGGTTTTCATGGGGCTG TGGGTCCCCAGCAGTTACTC mhp290 CGGGAAAAACAAATTTCCGT GAACATTCTCCATTATTATTATTTCTTG mhp291 AAATTTAAATACATCAAAATTGCCG AAATTCTTCTGCAAAATGTAAAAT mhp292 AACCTTTAGTCGTCGCTCCC AAATCAGGAGTGCCTGAGCC mhp293 CGAACATGGCAGCTATTC AATTTCGCCATTATCGCC mhp295 ACAAATATGGCCGGAAGAGG GCCTTCGATTTTCTTTTGGG mhp297 AGTTTTAATGCCTGGCGACC AAGTGCGGGAAAAGAGCAAG mhp302 TCAAGAGGCGATTTCGTATG TTTGAAGTAATTATGGGATCTTCG mhp303 TCAGCGGTATTGGTAGTCCG TCCTCCGATAGTTTTTCGACC mhp 3 04 AGCAGGTCTCATTGGACTTC TTATTACCCAAACCTTCGCC mhp305 GACATCGAAGTTCTCCAGCG AAGAAGAGCCACAAAACGGG mhp306 TTTTGCGGTTTAGTCTTGCTG CGGGGATTTCAAGACCTACC mhp307 TCGAGGAATTTGTACGTCGG GCCCGAAATTCTGTTGATTG mhp308 GGTTGTTTCTTGTCCAACGG CCGCTGTCTTTTGTGTTTTT mhp309 AATCGGAAGAGTAGTAATGACG TGCCACTAGAAGATAAGGCG mhp310 TGCCTCATTGGTTTTGCG GCGAG AG CG A AG A ATT AACG mhp311 GCTAACAACGTTGGCCAGTC CATAAAGGCCATATCAAAATGC mhp312 TTCGCTAGATATAAAGGAGCC AAAATAGAGGAGAACCTGACG mhp313 ATATGCGAGACAGCGAGC AACTTCTGCTATTGCTTGGGC mhp315 CTAGCAAAAACCGGCGATG TGCTAGTTGTGCTTGCATGTTC mhp316 TGCAACTCAACTACCGGGTG CCTTGAATCCCATAAGCAAGAC mhp317 CAAGTATCTCCATCCTTGAGTGG AAATATTTTTGGTAGCGGGC mhp318 GAAACCCAATCGGCACTAGG CAACTTGCGTTAATGAGGGC mhp319 TGAAAATCCTTTGGTCGATG CTAAGTTCCCAGTTGGCTCG mhp320 CGCTAAAAGAGATCGCCAAC CTCCCTCGGTAAAAACTCCG mhp321 TAATGCCGGAGCAAGAGTTC GCCAGCGGGAATTTGTTAG mhp322 GCAACACCTGAGGAACGAAG ATCAGGATCAGGCACAAAGG mhp323 TCTGGTGCAGCAATGTCG TTGGACTCAATCGCTATCAG mhp324 TGGATCATGAGAAACATACTGAGC TACATTCTGGGCATTTTGGC mhp325 ACGGCCGAATTACAAGATGA CAAATTCACTTTGGTCTATATCTTTTGC mhp326 TGCGTGAATAGGAAGAACGG ATTCCATCCAACCCCTTG mhp327 CCATTAACCGTAGAAACTGACGAG AACCACAGCAGCTATTGGAAG mhp328 TGGATGAAAATGGCACTG TCTCTTGAAAGCGAACGGAG mhp330 AAATTCTCGTGGCGATTGG GTAAAGGTTCGTTTTCCGCC mhp331 TATTTTAAAGCGCCGACAGG ATTT AAAAGATGCGCCACCC 139

Table A.1. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp334 ATTTTGGCCCTTACCCTTTG CATACTTTCGGCGAATGAGC mhp335 TCAACAACTCCCTTTACATACATC ATATGGCGGAACCGGAAG mhp337 CCCTAATTTCGGGAGTACTGA CCGATAGTAGCCTGATTTTTC mhp338 CGTGGGAAAAACAAATCTCCA G AACATTCTCC ATT ATTT CTTGA mhp339 GAAGCTAAAATGATCGGTCGG TATTGGCAGTTTCTCCAAGC mhp340 TGGGCAAGTTTTTACACCAG AATTTGCACGGAAAATGGC mhp341 A AACGG CA AGTTCGTTTC CGTTCTGGAGTAAGCCGC mhp342 AAATCTTGGCAAGTTGGACG ATAAGCCACCCAGGGTATCG mhp343 GTTGGAAAAGAAAGGGGTGC GACACTTCCAGCTGCTGTGC mhp344 TGCCTTTTGGGAGCAAAAAT TGATGGACAACTCTTTCAAGTTT mhp345 TGCCACGATAAAAGTCAAACAG GCATA AGCGCTTCA TTTTCC mhp346 TGTTGCTACCAATACCCAACTAGG ATTCAAGCCCTTGATGAGCC mhp347 TTGCATTTTGAACTTTGGGG TGATGCTTGCAATTCTGGG mhp348 TCGCGACTCATTTATTGTTGG TTGTCAGGGAAACTTACTTTTACG mhp349 CACTAACTAATAGACCTGGTACGGG TTTTGAACCAATCATCGTACCC mhp350 TAACCAACTTAACCGCCGC GGCGCTCTCAGAACGAAAG mhp351 TGTTCACTATAAGCGCAAAGG TCGCCTTCAGTAGTAGCTTTGTC mhp352 GGTCTTTGTCGTTGGACCTTC TTCACGTGCTAATGATCGC mhp353 GCTCACGAACCTGACATCAC TTTCCGGTTTTTCCGTTG mhp354 TTCGGAATATCATTAATAGGCCAC CACCTGTATTTGCACTTGCC mhp355 CCGTAGAATACCTGCCTCCG TTGACGCTTCCAGCAATACC mhp356 GGTTTGCTAATGCTCTTGGG TTTCTTCTTTTGGTTTGTTGATTGA mhp357 GCCGCTGCATCTGGAATAG ATCTTTGCCCACTCAGGGTC mhp358 CAACATAGGGGCACAATAATTAGC AATTTCCTTGCGGTTCCTTG mhp359 AAGAAAATCCCGTCAGTGGC ACCGATTGCCAAGACTAGCA mhp360 CAATCCAATTTCTAAGCAAACC TGAACGTTGAATACTATCCCCAC mhp361 AACTCTTCGAGTTGA ATAGCG GCTTTTAATGCATTTCGCAC mhp362 TTCCAAAGTTGCCCTTATTG CAGGGACCAAAATTTGCC mhp363 CAGGACCGGGTATTAAATGATG TTGCGTTAATGCTAATCTGGG mhp364 ATATGCTTGGATATGCCGGG TTCACGAGAAGGGATAATTGC mhp365 TTCCAGTTTCACCAGGTCATTT TTTGTTGTGCCTCTTTTTCGTT mhp366 TAACACAAGCCAGCGAAAGC GGCA A A CCGAAT GTTTCG AG mhp367 AAATCGGGTTAGGGCAAAAG CTGAGGCTCCTATTCCGTTG mhp368 CTTTCTTCAAACAGCAGGGG AACTGGAGGCAAGGCAAG mhp369 ATCGCTGCATGCCATCATAC AAACTAGGGGCAATGATCGG mhp370 TGGAACCGGATGTTTTATGC CAATAAGGTGCCCCAAGTCC mhp371 ACAATCGAAGACGATCCCG TTTGCTCCCCTTCTTGAGTG mhp372 TTTAAGTCAGACGGCTACCTC TGTTTTCGGATCTAGTGCTG mhp373 TGTAATCGACCCGATGCG CGATAACCCCAGTTGCCG mhp374 TTTGGTTATTCTAATTGGCG AACCACTTGGTTTT AGTTCGAC mhp375 TTTATTCGGGGACAAAACCA GCTTCTGGAGGTGTCTCTGC mhp376 AGCTTGAATCACGCCTGC AAACCTCATGTTGAATGGCG mhp377 CAACCTGGATATTTTCCTGCC CCGGGGAGATTTTTCATTTC mhp378 CTCCGCTTGAGTATTTTGCC TCCATCGCTGGCTTTTAATG mhp379 CCGCAACTTTTTAGCGTTCG GCTGCAAGACCAGCTTTGAC mhp380 TAAAGGCGAACTTGTCACCC TTAGCATTTACCTGGGCGTG mhp381 TCCTTGCACCGAAAATATGC ACTGTCGAATCGGTGAGGC mhp382 CTAACAGCATTAGCCGCCTG TGATCCTTG ACGGGAAACTG mhp383 TGGTTTT CG ATGATTAGCGG TCCATCAGTAAAAGCTTGTTGAA mhp384 ATTTGCCAAGCGACTAAACG GCCAGTGAAATTTTCGGGTAG mhp385 AGCAGTTTCACTGACCTCCG TTGCCCGAGTGATTCAGC 140

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp386 CAGCATCAAACTGACAAGATGC TTGCAACATCTTCAAAAGCG mhp387 TGTGGAAATGGAATGGGAAC AAGAACCGGAGTTAGCGCAG mhp388 GCCTTAATGACCGGGGT A AG TTGGCAAAAACACCCATAGC mhp389 CAAATGTTGGTCGGGATGTC GCTCTTGCTGCAACCTCAAG mhp390 ATCGAAAAAGGCGGAGGTAG GGACTCCAGAAAGGACTGCC mhp391 TGGAAAGTCAAAACAGTGCG TTTGGCAAATTCCCTAAGAC mhp392 CAAGTCAGCTTTCAGGGACG TGAATTAAACGGAATCGCCC mhp393 ATCTAGTTACAAATGCGCCG TGTCTGGAAAGATCGGTTTG mhp394 GAAAGGTTCAGCCGGAAGAC CATGAGCCGCCTTTTTGTAG mhp395 TATCGCGAGGATTGCCTATG TCCACGATTTAGGGATCGAAG mhp396 TTTGACATCCATTCAACTGAGG ACATAACCCTCAAAACGGCG mhp397 CAATCCATGCAAGTAAGGTCG TACAATTGCCCCGATTAGCC mhp398 CCAAAAACCGTTGCCTTACC AAATGCGGGTGGCTGTAAC mhp399 CTGACCAAAAACAATTAGCAA AAATTATATCAAAGCCGCCG mhp400 TGTT ATTGTGGAGCCAAGGG GTAAAACTGCGAGTGCCACC mhp401 AAGTTAATCTTGGACTTGTTTTTGTCA TCCTGGATTTGTCCAGGGAG mhp402 GCAACCCCTGATCAGAAACA AGC AGTTTT ATTTTATCGCCA mhp403 AAAATGCTCGAGAAAAAGCG TGAAGATCTTTCTGGTTTGG mhp404 GCAACCCTAACCCATTTGATG ACAACTGATTTTGCATCGGC mhp405 TCGAAATGAACTTGAAGGCG TCAGCGGTCATTTCGCTATC mhp406 TCAGGCTGTTCGGATTGAAG TTTGCTCGCGGATTTGAC mhp407 CTCAGCAATCGCTAAGGGTG GTGATGCGATAATGCCGATT mhp408 TTTAATCGAGGCTTGTAGGGC CTCCATAGGCATCAATTGGC mhp409 GCTTGAAATTGATGCACTCG TTGGCAAGACTAATGCAAGC mhp410 CCGGATTTGAGGCTAGATTG TAAAGCCCTTGGTACTCGCC mhp411 AACAGGACAGGAAAAGACCG CACCAGAGGAACCAAACGG mhp412 TTTGAATCTTTTTGTGCGGC GCTTCCTTTATCGACCTGCC mhp413 GGTTTCTTGTGGGGGAAATAG TGGGCGGGAAGATTATCAG mhp414 AGCAGCGAGAGCAAAAACAC AAATTGTAGTTCGCCATCCTTG mhp415 TTGAAATATTGGTTAAATTCCCACC TTTTT ATTTTTC A TACCGGCG mhp416 CTCCGAAAGCGTTTGCTTC GTGAGCCCCCAATTAACTCC mhp417 TGATTCTAACAAACTGAGCAAAAA GGCCGGAAAACTAAGACTAA mhp418 TTAGTTGTTGTCGGCGATCC TTTGAGCGTGCTAAAATGGC mhp419 GACAATCTTATTTACTCGTTGCCC AGAAGTTCGACAAAATCCGC mhp420 TTGATAAATCGGAAAATTAACCA TTTTTGCTCCCTTTTGGTTGT mhp421 AAAACGCTGGCAAAAATTCC GCATAATCAATTTGCCGTGG mhp422 TGTTCAGGGTCCTTTGCTTG ACGGTTTTGTAATGGCGTTG mhp423 TGCCGTAGCTCATTCTTCTG TAAGGCCG TTTTG TTTAGCG mhp424 CCTCCCAACTTATCGATGTC TTTGCTTTTACAACACCGGG mhp425 CCGCATCATCCGCTTTAG CATCGATCTGATAAAGTCCG mhp426 TAAGCAAGATTTCATCGGCG TTTCAAGATTGCCAGTTCCG mhp427 AGAAAAACGAGGGATCAGCG TTTCCTAGCTCAACGCCATC mhp428 TGAACAAAAACAATTAGCCTACA CGTTGAACGACAAACTGATAGACA mhp429 GCACTGAATCCGCACTTG TTTCCCTGGTTTTCCTCTAC mhp430 AAAACAAGGTCGAGGGCAAG GCTGGGAGTTCAATATCGAGG mhp431 TTATCTGCCGGACATGCTTC TTTCAGATGACCGGCTAACG mhp432 CATCTTGGTCTAAGTCTTGCTG AAGCGAAATTCAATAAGGAAGC mhp433 CGTTGTAGGACTTTCCGGTG ATATCCGGATTTGGGGTCAG mhp434 TTATTTTGGACCTCTGCCCC CATGGTAATCATTTTCGCGG mhp435 TCCAGAAAACAACCCTGCTC GGTTTTCGAGTTAAAAGCTGTGG mhp436 TTGATCCGAATGTCCTAGCC AGGCATTTGAAAGTGGTTGG 141

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp437 TTGATGGCTATTGACTTGTTCC AAATTAGCTCCAAAGAACGC mhp438 GGATATCTACCTTGTGCAGTCCC AAACTCGGGAATGCCGTC mhp439 ACCTGGGGAAATGTCTCAGC TTATCTGCATGGGTTGTGCC mhp440 CCAATGATCTTGAAGCCGAG ACGGGCTGAACGAATTATCC mhp441 AAGCAGGCGCTGATTTTACC TATCCTCAACTTCAACCCCG mhp442 CCATTTCTCTCCTTTTCACGG TTTTTGGCTAATTTTCCTTGC mhp443 GTTCGGCAATCCAAACTCTG CTTTGGAGACGAGCAAGTCG mhp444 CCGCGAGAATCAAAATGG TTGGAGACTTTGACTATTTGGG mhp445 AACCTCAAAACCAACCTCGC ATCCTGCCCGCCTAGTAATG mhp446 GAGGCCTGGTACCTAAACGC AATTCTGGGGGATTGACTGG mhp447 ATGAACCAAAAACCCTTCGG GAATTTGAGTTGCTCGCCTG mhp448 AAAACCAGGCCGAACTCAG AATATTGCTCGGGTTGACTG mhp449 TGATATTTTTGGTGCCCCTG GGGCAATATTTAAGGGCCTG mhp450 AAGTGGCGCTTATATTGCCC GGAAGATCGAGCTCATCACG mhp451 TGATAATGTCCAAGTTGGCG CTCCATTGCTGGATCATGTG mhp452 GAAATCCGGTTGTCATTTCG AATTTGCGATAGGCATCACG mhp453 CAAAAAGTAATCATCTCATCGCC GGGCACCAAGGAAATTAAAG mhp454 TTGATGCAATTACGGTGCC TCAGCAATTGGTTTCTGG mhp455 GGTTCAACGCGAGCTACTTC TTCTCCCTGTTGTCCGATTG mhp456 TGATATTGCAG AAAAGCTCCG AATCTGCTCTGCCAATTTACC mhp457 GTTGTTCACCCTGGTGTTGG TTGTAATCGCCGGTTTTGAG mhp458 TGGTTCTTCCTCATGGAACG TTCGGCTACTGCCTTTTCAG mhp459 GGTCTTGGGATTGTTATGCC CCCATTTGCATTGCTGTACC mhp460 TTATCTGCTCAACCTTTAGTGC TGGCTTTTGTTTGGCTCG mhp461 TAAGCAAGATTTCATCGGCG TTTCAAGATTGCCAGTTCCG mhp462 AATACGCCCGATGCAGTATG GAATCACGGATATTTTGGGC mhp464 GGATCGTCTGCACAAGATGAG TTCAGGTTCGTGAAAAACTGG mhp465 TGCTAAGCTGCGCAAACC CTTGAATTTCCAGGTCAGGG mhp466 TCCAGGAGCAAAAATCGCTA TTCGTTCTCAAGCCGTGTCT mhp467 AAAATTCATCGGTTATGTTCCG TCGCGCTGAAGGATCAAG mhp468 TTTTTGAGCCAAATGATCGG AAGTTTGGGCGT CTTGGG mhp469 AATCAAGCGCCAGGAATG TTCCTTGATCTCCAAGGATAAG mhp470 GCTTCGGGCACTAGTAAGGG TCAGCGGCAAAGGAGTATTG mhp471 GACTATCGGGAGGGGTTG CAAACTAGTCATTCTAAGGCGTGG mhp472 AAAGGTGCTCAAGATGCCC ATTTGACCATTTGCCCGC mhp473 TGCCGATACTAATGAAGCAG TTCACTCGGAAGTTTACCCG mhp474 GATTTACGCCAACTCAAACG AAGGACATCTCAAAACAGAGGC mhp475 TTAACGGCGCTTTTGTTCAC AACACGTCGGTTTTAAAGGG mhp476 CAATGATAGTGCCAATGGGG ACGGGAGAGCACAAATGATG mhp477 TCTCAATCGGTCAAAAACGG CAGCTGGGAAGGCTTCTTTC mhp478 CCCTCAAATCTCCGAAAAGC CTTGCGCCTTCAAAGTCATC mhp479 ATTTTGCCCCAATTCAACCT TCGATTTTCTTCATTGACTTGTT mhp480 TTCTTTTACGCACAATTCGC ATTAAGTGCCCGATCGCC mhp481 CCGAGTTCCGTGATTATTATG TGCCATCACATTTGAGGATTC mhp482 TGAGTTTCTTGCTGTTCGAATG TTTTCCGACCAATTCAGAGC mhp483 AATACCGGAAATCCACTGGC TTTTCCCCTTGAAGATGTGG mhp484 TGCCTCATTTTTCACCGC TAGACAAGAATTCCTGCGCC mhp485 GCCAGAGACTAATCGGAATGG TTGTTGTTGCTGACTGCGAG mhp486 GTCGTAGTTTTCTTGACCTTGG GATGACGGGCGCCATAAC mhp487 GGTCAGGGCCACAAACTACC TCCTAGCGCTAAAAACTGCG mhp488 GCTGAGTCCAAAAACGATGC ATTCCGCCACCGATTAAAAC 142

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp489 GCACAGGAAAATCCACAATTT TTCAGACCTTGGTTTGATATGC mhp490 TCTTTATAATTGGTGCGCCG TGCAAGACCCATTCCAAAAG mhp491 TTGTTTTGATTATTGCCGGA TTGGGCTATGACTTGGTTTAGC mhp492 AAAACCATTCCTGATTGGGG CGCAACTCCCCTGTCTTTC mhp493 CTTTGACAAAAGCTCACCGC CTTTGGATTAGACCCGAGCC mhp494 GGATCTGGAGGATCTGGGAC ACAGGCTCAATAACTGGGGC mhp495 CCCTAAAAGCGCCAATATGC CCGCC AGTAAT CCG AA AA AG mhp496 TTTAGCCAATGTTTTGCCAG CCGCCAACTTCGTCAGTATG mhp497 TCAGTCGGATATATTGCCCC GGTTCGATCGATGCCTTC mhp498 TCGCTTGGTTTTGAGCAG AGCCACTGGTTCATCGGC mhp499 TATTGCCGATGAGCCGAC TTTAGATCAGGGGGAGAACC mhp500 CAACTTCCAAGTGGCTGAGG TGCCAATATGAAATCCAAGG mhpSOl CAAAAGTAGCCGATCAAGGG CACTAATTGCGGCATCTTCG mhp502 ATTTATATACCCGCACCGGC TCAGGGGCATTTGTAGCATC mhp503 AAGCAATTGCAAGAGCGATG ATAAGCCCGGCGTCAGTATC mhp504 CAAGGTCTTGCTGAGGTTGG CAGAGGCAATTGCTTTCCC mhp505 TGCAAATCACAGTCAGGCAC TTCGGCCTTAATTGGGTAGG mhp506 CCGCAGTCAGAATTAGTGGC TCTTGAGTTGTTGATCCCCG mhp507 TTTTATTTTGAAAGGTGGACG GACCTTATATTCACGCAGTCGG mhp508 AAACCAAATTGAAGAATAAAGACCC ACACCGCGTTTTAATTTCCC mhp509 TCCAAGACCTGCAATGGTTT GCGCCATCAGAAATTGACAT mhp510 GCTCTTTCCTCTGGTATAGTGGC AAAAACCGCTCTGGTGAAC mhp511 AAGGTCTTTCACTTGCTGCG GATTGGATTCTTTGACCGGC mhp512 CCTTTCGATTGCAAAACAGC CCCTTGGAGGAAATTTGGAC mhp513 TTTTGGGCGCTCTGTTTATG AACCTTGGTTGATTACCCCG mhp514 TTTTCCTTTTGTTTTGCTTGTGA TGGTTTTGATGGCAGTTTTGA mhp515 ATTAACGCAATCGCCTTGTG CTGTCCAAAACCATTAGCACC mhp516 TTCCAACGAGAAAATGTTCAA ATGACAAAGTTGGCCCCATT mhp517 CGCAAATATGGTTTACACGAG TTTTCCATCGCACCTTTACC mhp518 GCAGCTATTATAAATTCCAAAGGTGAT TGACTTTTCATCAAGACTAGCAAA mhp519 TTGTATATGCCCAAAGATAAAAA AATTTTTGTTGCTTTTTCCGA mhp520 GGGGGTTCATATGGGCTTG TTACCTGACGTTGGACAGTCG mhp522 CCTTGGTCTTGTTGGCAGAC CTGAGCTGATCGGCTGTTTC mhp524 GGGCGGCATTTTCTTGTAGT GAGACCAAGAGAGGAATTTGAGAA mhp525 TGGAGTAGATCTCGAAAGTGCAT TGGGTCTTCTTGATTTTCTGG mhp526 GTCAGACTGGGATTTCTGGC TCCATAAGAAGCTGCCGTTG mhp527 ATGCCCCTATACCAAACAAG CTGTCTATTTTATCGGCCTG mhp528 TCTCCAGAATTCAGATCTAGAAAAA TTCTTGTTTTAATTCAGATAAAAGGG mhp529 ACTTGTGGGAATTATCGGGG GACTGTGTTTGTGGTGCTGC mhp530 AAACAGTTTCGAAAAGGGGC TTTGCTGAGCTCTTTGGGC mhp531 GGAATATCGAAAGAATCCCCG TTAAAGCGCAGCGAATAAGG mhp532 TTAATGTTTCGGCCCTAATG TTTCAATGATCGCCTCGG mhp533 TTCAGTCAGTTGTAAGTGGGC TAAACTGCCTTTCTTAGGGC mhp535 TTGAAGTTTGAAAGCAAGGC CAT C ACGGAAAT AACGGGC mhp538 TTTTCAGGAACACCAAACCG AACGCGGGAAAGAATAGGTG mhp539 AAGCAAGCTGAAGTCTCCCC TTCCGAAGCAGTGGATGAAC mhp540 TGGAAATTGTGGGTTATCGC TCGATTCCGCCTGTTACATC mhp541 GTAATGCAAGAATCAGCCCG TCCCATTTTTATAGGCCCCC mhp542 TTTTAATTTCGGGGTTTGGG CCGGTGTAATTAAAACGCCA mhp543 AGTTTGCCACAAGCAAATCC AAAAACGATCCCCTGCATC mhp544 GG A A ATCGGCGCT AAAATTC GTACCAAACCTTGTTGCCCC 143

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp545 ATTGAAATGCGACTTGGACG TAAAGTCCCAATTCCCCCAC mhp546 TGACTCAGGTTCAGGTTGGC ATACAGTAGCGCCTCCCCC mhp547 TGTCTGGTTTATTGGCCC TTTTATGGCGTGAACACAAG mhp548 GCCCTTTTGAAAACAAACCC AAAACGCTAAACCGACCACC mhp549 CACCGTGCTGATTATGATGC AATCAAAAGCGACTCTGGGC mhp550 TTCGATTCCAATAACTAGCG CAAGCCGGTCAAAAATTCC mhp551 TTTCTGCTCCACTTCCATTAC TTCTTGGCGTTTTTGGTGAG mhp552 TTTTGGCGCGGTACTTCC TTGCACCCATTTCGGATG mhp553 ATTTTTCTCGCCAAGTTCCG AACTTCATTTCCTGGCGGTC mhp554 CTGGA AGTTTT ACAGGCTGC TTATTTTTCCGCTTTGGG mhp555 CGAGACAATTTCCCAGGC AATCTCAAGCAGACTTTCAGG mhp556 TGAGTCCGAAGATGGTGAGG TAAGTCCGAATTTGCGAAGC mhp557 AGTTTTTGCTCTCTGGCGAC CCTAAATCCTGGGCTGCTTC mhp558 GAAGCCGGGACTGATTTAGG AGGGTCACCAAAACAAGCG mhp559 TGACAGGATAAGCAAAAGCG GCGATCTTGCAAGTGCTACC mhp560 GCAGAGCAAAATAACAAAATCCC TTAGCAAACGCAACCATAACG mhp561 TGACCTCAGTTTCAAAAGGC TCGCTATGAATTATTGGGTC mhp562 CCAGCTTGCTGAATCAGGTC ATTCCCGCTTCTTCTTCAGG mhp563 CACTAACTTTGAAAAATGGGCTT TTCATTTTTCCCCGTTTTGA mhp564 AATTCGTGGATCCCAAGATG ATTCCGAGCTGTCCAATCAG mhp565 CGGATTGACGATTTCTGATG ACTTTGCGACTTATCGACCG mhp566 ATCCCAGCGATTCCAAAGAT AAATTTCTGTGATGGGAAAAA mhp567 TTTTTAGCCTTACCGCACG TTTAACAGGTTGGCCGTCTC mhp568 ATGAGGCCCTAAATTCCGTC ATTTCGAGCGCCTGATTGTC mhp569 TGTCGAACCCGCAAAAATAC ACAGGAGCAACTGCACCAAC mhp570 ATAATGCTTTTTGGGAAAGG CGTGCAGAGGTGTGATTTTC mhp571 GCTTTTGCTGGTGCTCTCAC CATTTGTGCCGGTAATTTGG mhp572 GGTTCAGGACTAGGATCTTCAC AAGACTTGGGGCAACATCAG mhp574 GCGAGCTTGAAAACAAAAGTG AAAAGTTCCATTATTTCCTCCT mhp575 TCCAGTTTTCTTATAAGTTTTCATTTGG TTTCCCCAAAGCATTAAATTTGT mhp576 GTATTCAACGGTTGGGTTGC CAATTGTGAATTGTTCTACCG mhp577 CTCACAAAGCATTAGCGCAC TTGAAACTTGTTCAGCGGG mhp578 TTGAGCATGAGGTTCAGTTATTG AACGTTTGCTCAACCGCC mhp579 ACGGCAGAAGTGAGTATTCG TTTTGTTTTAATGCTAGACTCGCC mhp583 TGATCATGAAGTTCGTTTATCGTC TCATTGTTTGCTCAACCGC mhp584 TCGCCAGTTGATTATAGCCC CTAGCAAACTCCCAACGGC mhp585 CGGCTTGCAAATGAAAAAGA GCCACTATTTCCTGGCTCAA mhp588 ACCGGCTTTTATTGATTCGC AAACCTTGATTGTGTGCCCC mhp589 TAGCGCCAATGATTGTTTCG ATGACCAATTTCTTTCCCGC mhp590 TGGTTTTGCCAAAGACAACC CATAGGCACAAATCTGCGG mhp592 ACAAGCGCTCGAATCACG AACTTCAAGTCATGCGGCTG mhp595 ACTTAAAGATTTGGCGCCTG AGAACTGATGCTTGAACCGC mhp596 GGGCGAAATTTTAAGCCCTA CCCACTCAAAATCTAGTAGAAAAACG mhp597 AAATCGTGGGGAAGTTCAGG TTATTTGCGACAAGATCGCC mhp598 TCCCAAGGGAAAACAAGTGG AAAACCGOCTACCACCTGAC mhp601 CAAAGATCTTAGCCCCCGTC GATTATTGGCACATTTGCGG mhp602 GGCAGAGGCGCATATATTTTT GGCAAGAGTGGAAATATTTGTCTT mhp606 CTCGACCAGACGGAAACATC GCTGGGGAAGCTGATAAACC mhp607 TCTTCCGATTAGCCTTATTTATGC CAAAAACACACGTAAAATCACGA mhp608 ACCTTCACGCGCTTACTTTG TCTGCAGCCATCAAAATTGG mhp613 AATCGCCACAAATATCAATG TTTGCATTTTTGACAGAGCG 144

Table A.l. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp615 ACTTTTATTTGGGGCGCTTG TGGCCCATATACAGGTCCAG mhp616 ACCCGAATTACCTCGATTGG TCGACTCTGCTTAAGGCTGC mhp619 CGAAAGCCGGTTTGAGAGA ATAAAAGAGTTTAGAAATGGCATC mhp620 TTACTTACAACCAGCGCGAC AAACAGTTGACGGCTTTTCG mhp621 TTACTTACAACCAGCGCGAC AAACAGTTGACGGCTTTTCG mhp622 TGAGATGGCGATGCTCTTG GAATCAAGTCATTCCTGCGG mhp623 GCTGTCGCTGGACCACTTAC GAAATGGCTTCAGTTGCGAG mhp624 GAACTTTGATGTTCGCGGTG TCGGCCTTTATTGAGCACAC mhp625 CCCGTGAAGGTTTTGATGC GCCGCGAACGTAATAAAAAG mhp627 GGTCAGGCTTGAAAATTCCC AGCGCGTTTACAAAGTGAGC mhp628 CCCAGCTTCAAAAGTTGCC TCACCGGTAATTTCAATGCG mhp629 CACAAAACATGCCCTTGGC GTCGATTTCTCGCTGTGGAG mhp630 TGCCGCTTATCGAACTTTTG TTCATTGTCCCATGGAAACC mhp632 GAAAACGAGACAAGGCTACTCG GCTTTTGTTTGGTCAGGG mhp633 TAGTGCCGATTTTGGAGTCG TAGGATGGAGATTCACCCCG mhp634 CAGGAATGCGAGGTTTGATG AACA ATCTTCGGCAACAACG mhp635 CAGGAATGCGAGGTTTGATG AACAATCTTCGGCAACAACG mhp636 TCCAGTTGCTACCGAAGAGG TTTGAGAACTGAAGGGGCTG mhp637 TCCAGTTGCTACCGAAGAGG AATCGAAACTTCTGCTCCCG mhp639 AGGTCATCGAAAACGGGC ATAGTTTCAAGGGCTTGGCG mhp643 TTCATAAGATGAAACTTGCCG TTGCTTGCAATTTTGCGG mhp645 GCTTTTACATGGAGATGGAAGTAG TTTACTCCGGCATTAGGTTG mhp650 CATACAAACGACGAACACGG GCGGCAAGATCAAGAATGTG mhp651 TTCAGTTGCGATTGGAAACC TATCTGGAGTTGTCAGCCCG mhp652 TGCTCAGCTCTCGAAAATCC GTCCAGTCTGTTTTGTGGCG mhp653 TTAGGCAACTTTTCCGGTCC TTCGATGCTCTTGTTGCTCC mhp654 AGTTTTTCTGATGGCTCGAC TGGGTATTGTTTTCGTTGC mhp655 GCGGCCTTAAAACCCAAG TGCTTTTGCTATCATTGCCC mhp656 TAGCGGCCTTAAAACCCAAG TGCTTTTGCTATCATTGCCC mhp657 CGGGTCTCATGGTAGAGGAAC GAAAGTTGCTGGAGTTCGGC mhp658 CTATGTCACCAGAGCGAAAAG TCCTTTGTCGGAAAGTCCTG mhp659 CTATGTCACCAGAGCGAAAAG TTCAGCTGCAAATTGGAC mhp660 CCGACCTGGTTGACATACTG TTGATTGGAACTGTTGGG mhp661 GGCCGACCTGGTTGACATAC TATGCGAAAAATCCTTTGCC mhp662 TCCGGCAGTAAAACTTGAGG CCCAGGCAGC ATTTTT ACG mhp663 ATTGCTGATTCTTTCTGGGG TCATCAAGGTTGCAAGTGGG mhp664 AATAAGTTAGCCGAGCCCC TCCACTTGTTTGGATTGCTTG mhp665 GATGTTTCCCCTGGCTATGC TCCACTTGTTTGGATTGCTTG mhp666 GTTAAATGGGGCGCAAAAAC CGGAGTTAAAACTACCGCCC mhp667 A AGCC ACTTTTGGCTTTTG A CAATAGCTCGGTGATCTAACATGA mhp668 AAGCCACTTTTGGCTTTTGA CAATAGCTCGGTGATCTAACATGA mhp669 AATTATCAGGCCAACCGGAC AACTTCCGGGAT ATCAACGC mhp670 AATCATCAGTTGCCCGAG TAATCCTGGCCGGTAGTCTG mhp671 TGGAACTGGACGTAGGAAATC AAACTTTCAACAAGGCACGG mhp672 TGTCAAGCACCAAGAAGTCAAG TTCACGGGCTGTTCTCACTC mhp673 TATCCGCGATAAGTCAGGTG ACATTTGGTCAAAGTTGGCG mhp674 ATCGACATTCACTGCCATCC ATTTGGACTTCTTTTGGCGG mhp675 GTTGTCAGTCCCGATCATGG TAAAAACGCCGTGAGTAGCG mhp676 TACACTAGCGCCTTTTTCGG TGTTAGCGAATCGCCTAGTG mhp677 CCAGAATATGTTTGGTGGGG TTGATCAACTCGGGACATCG mhp678 TATGTTTGGTGGGGGAAGTG TTGATCAACTCGGGACATCG 145

Table A.1. (Continued)

Gene ID Left Primer Sequence Right Primer Sequence mhp679 GAGAAGGTTTAACTCTAAGATGCAC ATGCAAAGTGTGTTTGTTGC mhp680 GCATTGAAGGGCAAGTGAAG TTCCATAGCCTTGTTCGGTG mhp681 GCAGCCCGAAATAACTAGTCC TTTTTCG TGCGTTAGC A ACC mhp682 GCCAGGGAAGCAGCT AATG TCAGGGGCTAGGGAAGAAAG mhp683 AAAACGGAAGCTCCAGCAAC TCAATTTGGTTCGGATCAGG mhp684 TGGTCCCTAATTTTGTTGC CAGTGATAATAGAAACAAGCGG mhp685 GAGGCGGGATTTCTAACCC CCAGTGATAATAGAAACAAGCGG mhp686 TAGTAGTAACGCATTCATTCATATGCC AAAAACCCCCTTGAGAGCG mhp687 GTCGATGTGATCTCTTGGGG AGTGCGTGGTGGGTAGTTTG mhp688 TGATCACCAGTTTTGTCTCG AGACAAGCGACTAAACCGTC mhp689 GACGGACTGAAAGACGCATTA AAAAAGGAGTATAATATGAAAAACACA mhp690 TCAGCCGCAATTTAGGTCAG TTAGTTGCCAGACCCTGCTG mhp691 AATGGGTGTTGCTGGTTCAG CCGGAAATTGTCAGAAACCC mhp692 CCTTTCAATTATCGGGTTGC TTTTTGGTTGTTAGCGATTC mhp693 ACGCACAAATTGCGCTTC AACCCCCAAAATTCTTCTGC mhp694 TCAGATACGCTAAGGCAGGC GCGAAGAAAAATCCCCGTAG mhp695 CACTTCCGTTATTATCCCAAGC GCCGAAGGGGAAATTAAGAG mhp696 CACTTCCGTTATTATCCCAAGC GCCGAAGGGGAAATTAAGAG mhp697 TTTCCGATTGTTTCCTGAGC ACAATTTCCTGACGGTGTCG 146

APPENDIX C. PRINTING FILE FOR BIOROBOTICS MICROGRID PRINTING ROBOT 147

[Parameters] RunType=Micro Array BarcodesEnabled=0 HaltBarcodes=l CacheBarcodes= 1 ClearBarcodes=l WhereBarcodes= 1 ExpectedBarcode 1=[U nknown] ExpectedBarcode2= ReloadPins=l NumSourceP lates=2 WashAfterEveryVisit=0 WashAfterEverySample=l RequestLogInfo=0 SourcePlate=384standard.microplate ToolType=4x4 384 split.tool PromptPlates=0 PromptPlateslnterval=l SourceWiggle=0 SourceDibble=0 Remo veOneLid= 1 TargetDip=0 TargetWiggle=0 SoftTouchDistance=0.479999989271164 SoftTouchSpeed=4 DelayBefore=0 TargetDwell=0 DelayAfterSpotting=0 MultiStrikes=l SoftTouch=l TargetDepth=0.3400000035 76279 ILayout=0 PrespotRecycle=0 UseCamera=0 Microplates=l PreSpot=l SlideSize=25 NumSamples=48 LastPlate V isits=24 SpotsPerSource=29 Adapter?late=slidesvertical extra+1.adapter Adapter?Iate2=slidesvertical extra.adapter [PinGrid] XSpots=12 YSpots=12 Line0=37, 38, 39, 40, 41,42, 43, 44, 45, 46, 47, 48, Linel=25,26, 27,28, 29, 30, 31, 32, 33,34, 35, 36, Line2=13, 14, 15, 16, 17, 18, 19, 20,21,22, 23,24, Line3=l, 2, 3, 4. 5, 6, 7,8,9, 10, 11, 12, Line4=37, 38, 39, 40, 41,42, 43, 44, 45, 46, 47, 48, Line5=25,26, 27,28, 29, 30, 31, 32, 33,34, 35, 36, Line6=13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, Line7=l, 2, 3, 4, 5, 6, 7,8,9, 10, 11, 12, Line8=37, 38, 39,40,41,42,43,44,45,46, 47,48, 148

Line9=25, 26, 27, 28, 29, 30, 31,32, 33, 34, 35, 36, LinelO=13, 14, 15, 16,17,18, 19,20, 21, 22,23, 24, Linell= l,2, 3,4, 5, 6,7, 8, 9, 10, 11, 12, NthOrder=4 PatternFormat=2 ToolGridlndex=l, 1, Splice=0 Pitch=0.360000014305115 NumCopies=52 [SlideLayout] BottomMargin=10.0699996948242 CentreHoriz=l CentreVert=l TopMargin=l 0.0699996948242 LeftMargin=3.25999999046326 RightMargin=3.25999999046326 XSpacing=2.20000004768372 YSpacing=19 [Wash] lnterCycleDelay=5 LeftSeconds=5 RightSeconds=5 DrySeconds=2 V acuumEnabled= 1 Cycles=l MWSCycles=l [RunPrefs] PrintReplicatesFirst=0 SourceDipTop=0 SourceDipBottom=0 TargetDipTop=0 TargetDipBottom=0 VisionID= BioBankID=DEF AULT IOControl=0 Bathl2ZOffset=0 Bath3ZC)ffset=0 FeatureOffsetX=0 FeatureOffsetY=0 HumidStartRate=79 HumidRunRate=77 TargetHumidity= 102 MinHumidity=89 UseHumidityControl= 1 UseHumidityHEP A=0 PauseHumidity=0 EndHumidityOn=0 PrimeBath=l AskLoadTool=l WashAtStart=l WashAtEnd=l AskBioBank=0 AskTrays=l SourceDwellTime=l TargetDwellT ime= 1 Source Wiggles=0 SourceWiggleSize=0.25 SourceDips=l SourceDipSize=2 TargctWiggles=0 TargetWiggleSize=0.25 TargetDips=0 TargetDipSize=2 SourceDepth=0 TargetDepth=0 BathFillTime=0.5 BathDrainTime=8 RecirculateAction=1 RecirculateStatic=0 RecirculateWiggleSize=0.200000002980232 RecirculateDipSize=l BlotFirstTimes=l BlotFirstSlides=5 CacheBarcodes=1 HaltNoWater=l BathWiggles=4 BathWiggleS ize=0 HeaterOnTime=0 HeaterOffTime=0 HeaterCoolTime=0 SourceSoftTouchSpeed=4 TargetSoftT ouchSpeed=4 SourceSoftTouchEnab led= 1 TargetSoftT ouchEnab led=0 SourceSoftTouchDistance=l TargetSoftT ouchDistance=1 [PreSpotLayout] BottomMargin=5.94000005 722046 CentreHoriz=l CentreVert=l TopMargin=5.94000005722046 LeftMargin=3.40000009536743 RightMargin=3.40000009536743 XSpacing=3.20000004768372 YSpacing=4 ToolGridIndex=l, 2, 3, Splice=0 [LogFile] Operator name*=_ Library=_ Gridding_Number=_ Set_Number=_ Rep lica_Numb er=_ Date=_ Time=_ [PreSpotting] Pitch-0.28999999165535 SoftTouchDistance=0.479999989271164 SoftTouchSpeed=4 DelayBefore=0 TargetDwell=0 DelayAfterSpotting=0 MultiStrikes=l SoftTouch=l TargetDepth=0.340000003 576279 ISoftTouchDistance=l ISoftTouchSpeed= 1 lPitch=0 IDelayBefore=l ITargetDwell=l IDelayAfterSpotting=1 IMultiStrikes=l lSoftTouch=l ITargetDepth=l 151

APPENDIX D. HEXAMER OLIGONUCLEOTIDE SEQUENCES USED FOR GENERATING CDNA FOR LABELING EXPERIMENTS 152

Hexamer oligonucleotide primer sequences for cDNA generation*.

AACACA AACGRC AAGCGA AAGGYC AAGMCG AAGTAG AAGTGT AARTGC AATCGC AAWGCC ACGAGT ACTGGC ACTGTG AGACGA AGCAGA AGCGAG AGCSGC AGGAGC AGGGCY AGRAAC AGRTCG AGTACC AGTAGW AGTCCG AGTGGA ATACCS ATAGTG ATATAG ATCGGG ATCGTA ATCTGC ATGACA ATGCCC ATGCWG ATGCYA ATGTSG ATRGAA CAGAAG CAGGGG CATACC CATAGG CATGTT CCCGGC CCCKGT CCGGCG CCWGAA CGAAAA CGAGTA CGCATA CGCGAA CGCGGT CGGTGA CGGTTG CGSCAA CTGAAA CTGGCT CTGTAG CTSGGA CTTGTC GAAATA GAAGTG GAATAG GACTCA GAGGCT GAGTKA GATCSG GATKGA GCAAGG GCCATA GCCCCT GCCYGA GCGCAT GCTCAG GCTGGK GCTGTA GGAAAG GGAAGT GGCARC GGCCTY GGCTAG GGGACA GGGCTG GGMCAA GGTRTA GGTTCA GMCTGG GRGAAA GTAACA GTAAGC GTAGTT GTATCC GTCMGG GTCTGT GTGAGT GTGGCA GTKCAT GTRTAA GTTGGG GTTGGW KGAGAA KGTTGG RCAGAA SGGCCT TAGCGG TATGTG TATSGA TCCCGG TCGGCM TCGWGA TCTGKA TCWTGT TCYGCG TGAMCG TGCMCT TGGCTT TGGKTC TGGTCG TGTAAW TGTAGA TGTATC TGTSCC TGYCGA TTCTGG TTGCGT TTGTCT TTGTTG WGGACA WGTGCA WTGGAC * A=adenine, T=thymidine, G=guanidine; C=cytosine; Y=C or T; R=G or A; K=G or T; W=A or T. 153

APPENDIX E. OPEN READING FAMES MISSING FROM THE MYCOPLASMA HYOPNEUMONIAE MICROARRAY 154

Open reading frames missing from the M. hyopneumoniae microarray. Gene ID Gene Description nlength mhp026 CH Putative ABC Transporter ATP-Binding Protein 2325 mhp045 UH Unique Hypothetical 1206 mhp051 atpF ATP Synthase B Chain 588 mhp055 atpD ATP Synthase Beta Chain 1413 mhp063 rpoD RNA Polymerase Sigma Factor 2034 mhp068 UH Unique Hypothetical 177 mhp074 cmk Cytidine Monophosphate Kinase 687 mhp075 CH Putative GTP-Binding Protein 1299 mhp076 UH Unique Hypothetical 288 mhp082 tnp Transposase 1656 mhp089 UH Unique Hypothetical 579 mhpl55 UH Unique Hypothetical 156 mhp!63 UH Unique Hypothetical 210 mhpl85 UH Unique Hypothetical 177 mhpl92 UH Unique Hypothetical 738 mhp!93 rpL22 Ribosomal Protein L22 609 mhp272 CH Hypothetical protein P102 2835 mhp294 CH GTP Binding Protein Motif 816 mhp296 UH Unique Hypothetical 456 mhp298 deoA Pyrimidine-Nucleoside Phosphorylase 1350 mhp299 nox NADH Oxidase (NOXASE) 1380 mhp300 CH Conserved Hypothetical 1347 mhp314 UH Unique Hypothetical 189 mhp329 CH Conserved Hypothetical 396 mhp332 UH Unique Hypothetical 993 mhp336 CH Unique Hypothetical 1470 mhp463 CH Conserved Hypothetical 603 mhp521 CH Conserved Hypothetical 1395 mhp523 CH Conserved Hypothetical 1032 mhp534 CH Conserved Hypothetical 3645 mhp536 UH Unique Hypothetical 246 mhp537 CH Conserved Hypothetical 1164 mhp573 CH Conserved Hypothetical 453 mhp580 UH Unique Hypothetical 858 mhp581 UH Unique Hypothetical 441 mhp582 UH Unique Hypothetical 210 mhp586 CH Putative Serine Proteinase 660 mhp587 pdhD Dihydrolipoamide Dehydrogenase 1458 mhp591 nagB Glucosamine-6-Phosphate Isomerase 762 mhp593 rpsD Ribosomal Protein S4 615 mhp594 CH Conserved Hypothetical 432 mhp599 dnaE DNA Polymerase 111 Alpha Chain 2 2955 mhp600 rbfA Ribosomal Binding Factor A 327 mhp603 nusA N Utilization Substance Protein A 1842 mhp604 UH Unique Hypothetical 225 mhp605 CH Alkaline Amylopullulanas 1998 155

(Continued) mhp609 trpS Tryptophanyl-tRNA Synthetase 981 mhp610 CH Conserved Hypothetical 354 mhpôl 1 CH Hypothetical ATP Binding Protein 3198 mhp612 UH Unique Hypothetical 753 mhp614 CH Conserved Hypothetical 300 mhp617 CH Conserved Hypothetical 1806 mhp618 CH Conserved Hypothetical 1482 mhp626 CH Conserved Hypothetical 909 mhp631 CH Conserved Hypothetical 2082 mhp638 rplJ 50S Ribosomal Subunit Protein LIO 549 mhp640 CH Conserved Hypothetical Membrane Lipoprotein 2466 mhp641 CH Putative Methylase 1647 mhp642 UH Unique Hypothetical 174 mhp644 pr2 Conserved Hypothetical 1632 mhp646 CH Conserved Hypothetical 1587 mhp647 UH Unique Hypothetical 537 mhp648 CH Conserved Hypothetical 1608 mhp649 CH Conserved Hypothetical 1617 mhp698 UH Unique Hypothetical 273 mhp699 rpmH 50S Ribosomal Protein L34 141 * CH = Conserved Hypothetical; UH = Unique Hypothetical 156

Acknowledgements

I would like to thank Dr. Chris Minion for giving me the opportunity to work, study and learn in his laboratory. The experience has given me a greater appreciation of scientific research and the pride that stems from completing such projects. You provided great guidance without providing all the answers. The opportunity to work with challenging new technology and to study a significant pathogen has been very rewarding.

Additionally, my POS committee members, Drs. Nettleton, Phillips, Thacker and Wurtele provided critical input in achieving my goals. Thank you for your patience and guidance during my studies here at Iowa State University.

The Minion laboratory was vital to my success. Bobbie Cleary helped make my first years easier. I thank her for being a good friend, as well as, her great lab skills. Josh Pitzer made sure the lab ran smoothly and was helpful when troubleshooting. Monica Perez was an excellent lab assistant and was always willing to lend a hand in any way possible. I cannot forget the numerous undergraduate lab assistants that also made my job easier: Shawna Greene, A1 Green, Jenn Rardin, Mike Carruthers and Mike Oneal. Thank you all.

The Thacker laboratory was instrumental in providing cultures and technical expertise. I would especially like to thank Barb Erickson and Nancy Upchurch for all their assistance. Also, I want to thank Erin Strait, a fellow graduate student all too familiar with the difficulties of working with Mycoplasma hyopneumoniae.

Brad Van De Woestyne was extremely helpful and patient with all of my questions regarding statistics. Thank you for your time and devotion.

I want to thank my numerous friends who have supported me during the years I have been in school with their encouragement. Thank you to my parents, Medea and Carl Madsen, who never doubted I could accomplish what I set my mind to and never let me give up. Finally, my brother, Will "Punky" Snow, who provided comic relief over the years and reminded me to make the best of my life.