At the Interface of Microbiology and Immunology

It is my great pleasure to welcome you to IMMUNOLOGY 2019, the annual meeting of The American Association of Immunologists (AAI). I am honored to be the president of AAI. This is an outstanding association, with incredible and dedicated staff. AAI is really all about the members, and I thank all the members who have volunteered their time over the past year to make AAI the best that it can be. Thanks to all of you for coming to the meeting and attending my presentation tonight. I hope you all enjoy the meeting because we have a lot of excitement planned for you!

I was raised in a small town, Eureka, on the coast in far northern California. While growing up, I was always interested in science and “inventions,” but I had no idea that being a scientist was actually a profession. I had never met a scientist, and I thought that if I was good at science, the only path was to be a medical doctor (which was not of interest to me). Although there was only one public high school in Eureka, I had the good luck to have some excellent teachers. In particular, during my senior year I was taught organic chemistry and biochemistry by a wonderful but stern teacher, Mr. Herron. I had never even heard of organic or biochemistry before this, and Mr. Herron made it so interesting that I decided I would study biochemistry in college.

I attended the University of California at Davis for my bachelor’s degree. When I arrived, I was still very naive about careers in science. My advisor was Dr. Irwin Segal in the Biochemistry Department. I realized that not only did he teach, but he also directed a laboratory and did research (again, I was very naive!). This was a revelation to me, and I decided then that my goal would be to direct my own laboratory and research. However, I didn’t have any time for laboratory-based research until my junior year of college because I was working nearly full-time. I did end up doing research in a microbiology laboratory, and that experience plus the microbiology coursework inspired me to pursue microbiology research in graduate school.

I was accepted into the Microbiology and Immunology Program at the University of California, Berkeley, and joined the laboratory of Dr. Dennis Ohman, investigating the opportunistic pathogen Pseudomonas aeruginosa. Specifically, I studied the genetic regulation of alginate production, a polysaccharide produced by Pseudomonas in the airways of cystic fibrosis patients (1, 2). Although I was taking immunology classes at Berkeley, to be honest, I found the subject difficult to understand at the time. In my first postdoc, in Dr. Magdalene So’s laboratory at Scripps, I began a project to develop Salmonella as a vaccine delivery vehicle for malaria Ags (3). This is when my relative lack of immunology knowledge became apparent! I wondered how I could study pathogens without understanding how the host interacted with them. I chose to undertake a second postdoc in a laboratory where microbiology and immunology intersected. I wanted to continue with a bacterial pathogen, and tuberculosis (TB) seemed like a good place to learn immunology. I joined the laboratory of Dr. Barry Bloom (AAI president, 1985–1986) at the Albert Einstein College of Medicine. My projects revolved around using various knockout mouse models to define the critical factors in the immune response to Mycobacterium tuberculosis (4–10). I was so fortunate to have Barry as a mentor, and I immersed myself in the field of immunology of infectious disease. I decided that I wanted to focus on TB when I started my own laboratory. I joined the University of Pittsburgh School of Medicine in 1994 as an assistant professor and continued to study the immunology of TB using mouse models.

TB kills more people worldwide than any other single microbe, with 10 million cases of active TB in 2018 and 1.5 million deaths. This is despite drugs that can cure TB! M. tuberculosis is transmitted primarily by the respiratory route and infects the lungs. The majority of humans can control the infection but remain infected, perhaps for life, and are at risk for reactivation of disease years to decades later. Approximately 10% of those infected will develop active, symptomatic, and transmissible TB, either within a year or two postinfection or as a result of reactivation. Despite more than a century of study of this disease, we still do not have a full understanding of the immunologic mechanisms that protect against infection or prevent the disease nor those that exacerbate disease. In my work in murine model systems, we identified several immune factors or molecules that were important in control of TB. These included IFN-γ, TNF, IL-12, CD4 and CD8 T cells, and CD40. We also identified mechanisms by which these factors or cells acted, including demonstrating that γ-δ T cells were early contributors of IL-17 in the lungs of M. tuberculosis–infected mice (findings that were published in The Journal of Immunology!), and various mechanisms by which macrophages control M. tuberculosis (11–27).

Around 1998, I became frustrated with the mouse models of TB. Despite many years of effort, we were unable to develop a mouse model of latent TB, which is a major interest of mine. In addition, the pathology in murine TB is substantially different from that seen in humans. In human lungs, a granuloma is formed in response to M. tuberculosis infection. This granuloma is an organized structure composed of immune cells, including macrophages, T cells, B cells, and neutrophils. Mice do not form organized granulomas, and I was convinced that studying granulomas was critical to understanding TB at the tissue level. Also, mice do not control M. tuberculosis bacterial burden well, with a relatively high number of bacteria in the lungs. Despite this high bacterial burden, mice can live up to a year with the infection. I went in search of another model system that better approximated human TB.

Sometimes, you get lucky. In 1999, Buddy Capuano was in the office next to mine, and Buddy is a nonprimate research veterinarian. He sold me on the idea of trying to develop a monkey model of TB. Another stroke of luck was a National Institutes of Health (NIH) request for applications on lung infections in HIV-positive persons. TB is the major cause of death in HIV-positive persons worldwide, so this seemed like a great opportunity. Buddy taught me everything I needed to know about performing research with macaques, and the NIH awarded me a grant to develop a model of latent TB. This was a leap of faith for me—abandoning the tried-and-true (and accepted) mouse models and forging ahead with a difficult large animal model system. In fact, I wondered at the time whether this was a potentially career-killing move. We started with cynomolgus macaques and chose to use a low-dose infection. This resulted in a model in which half of the animals developed active TB and half developed latent infection, which provided an excellent platform for the study of the full spectrum of infection outcomes seen in humans. We characterized the model system and found that the pathology was remarkably similar to that in humans, with organized granulomas. I was so excited that we had derived a good macaque model, and we eagerly wrote and submitted our first paper. I still remember the one-sentence review we got back: “We don’t need a nonhuman primate model of TB.” This was devastating. Yet, we persevered (28).

Many long years of model development, immunology, and microbiology studies later, we are now at a point where the nonhuman primate models of TB are accepted and used by others in the field. We have developed several novel tools and technologies for studying TB in macaques. I would like to highlight some of what we have learned.

We have used three different nonhuman primate species for our TB research: cynomolgus macaques, rhesus macaques, and the common marmoset (29, 30). Marmosets are extremely susceptible to TB (31), and so we have not pursued additional studies at this time with this species. Cynomolgus macaques, as noted above, develop the full spectrum of infection outcomes and manifestations similar to humans. Rhesus macaques are more susceptible, and nearly all rhesus macaques will present with active TB within several months of low-dose infection (30). We have developed quantitative and sensitive measures of bacterial burden, both at the individual granuloma level and in the whole animal. This is critical as an outcome measure for interventional studies.

About 12 y ago, we began using positron emission tomography with computed tomography (PET CT) imaging to serially track infection in macaques (32), which revolutionized our understanding of the infection process. In collaboration with Dr. Sarah Fortune at Harvard, we generated and used “bar-coded” strains of M. tuberculosis, in which each bacterium has a unique DNA barcode (33). This, in conjunction with PET CT imaging, enables us to track each bacillus through the entire infection. We also have sophisticated immunological assays to study the immune responses in each individual granuloma (34). Through a series of studies using these tools, we determined that each individual bacillus that enters the lungs forms an individual granuloma (33, 35). The bacillus replicates up to ∼10⁵ bacteria in a granuloma by ∼4 wk of infection. However, at that point, each granuloma in an animal’s lung is independent and has its own trajectory. Some granulomas begin killing the bacteria by 8 to 10 wk and can even sterilize. A subset of granulomas appears unable to kill the bacteria; in that case, the bacteria escape from the granuloma and seed a new granuloma, a process called dissemination that is associated with development of active TB. These events happen simultaneously in the same lung lobes and indicate that true control of infection must occur at all sites of infection; only one failing granuloma in a lung can lead to active TB. We call this the “one bad apple” hypothesis (36). We are actively studying the characteristics of granulomas that control the infection and those that do not use multiparameter flow cytometry and single-cell RNA sequencing to identify protective mechanisms in TB. Our data indicate that a combination of pro- and anti-inflammatory factors (e.g., TNF and IL-10), rather than a strong TH1 response, are associated with sterilization of granulomas (34). Achieving this balance—not too little and not too much inflammation—seems to be critical in the control of TB. In addition to our wet-laboratory work, we have collaborated with Dr. Denise Kirschner at the University of Michigan for the past 20 y to fill in the gaps with computational modeling of the granuloma (26, 34, 37–51).

Having a model of latent TB allowed us to study reactivation, which is very difficult to study in humans (52, 53). Early on, we demonstrated that depletion of CD4 T cells caused reactivation of latent TB in ∼50% of cynomolgus macaques (54) and that SIV infection (as a model of HIV) resulted in 100% reactivation but with variable timing (3 to 11 mo post-SIV) (55). TNF-neutralizing drugs are used in humans for a variety of inflammatory conditions, and these drugs increase the risk of reactivation of latent TB. Our early studies showed that anti-TNF Abs caused reactivation in cynomolgus macaques (56). We set out to try to understand the risk factors for reactivation using anti-TNF Abs. We treated 25 latently infected cynomolgus macaques with anti-TNF Ab for 8 wk and assessed them by PET CT scan prior to and during anti-TNF treatment. Half of the monkeys reactivated latent TB, whereas the other half did not. By examining the PET CT scans prior to anti-TNF treatment, we found that a higher level of inflammation and/or the presence of an extrapulmonary site of infection could accurately predict reactivation risk with 92% sensitivity and specificity. We applied this metric to latent control monkeys without TNF neutralization and found that those predicted to be at higher risk of reactivation had a granuloma with higher bacterial burden than any of the granulomas in the low-risk animals (57).

Developing an effective vaccine against TB is a major priority. We wanted to determine first whether concurrent infection with M. tuberculosis was protective against subsequent infection. The data in the human literature, although primarily epidemiologic, indicate that latent infection is protective against active TB due to secondary exposure. However, we wanted to investigate this in detail at the tissue level. In collaboration with Sarah Fortune, we used one bar-coded library of M. tuberculosis as primary infection, and 4 mo later we rechallenged the animals with a different bar-coded library. Compared with naive macaques, the macaques with primary infection were robustly protected against reinfection (58). There were significantly fewer granulomas due to the second infection, and the majority of those new granulomas were sterile by 4 wk. The extent of disease due to the primary infection was not associated with protection. Overall, we found a 10,000-fold reduction in bacterial burden in the reinfected animals compared with naive macaques. This reduction was because of a combination of fewer granulomas being established and both restricted growth and increased killing in those few granulomas. Thus, concomitant immunity exists in TB and provides a model of sterilizing immunity in which protective immune mechanisms can be assessed. We are currently determining the importance of T cell subsets in this protection against reinfection.

We next turned to studying vaccines. We have tested several vaccines in macaque models, without demonstrating robust protection (59, 60). In collaboration with Patricia Darrah, Mario Roederer, and Robert Seder at the Vaccine Research Center of the National Institute of Allergy and Infectious Diseases/NIH, we tested various routes of bacillus Calmette–Guérin (BCG) for protection (61). BCG is an attenuated Mycobacterium bovis strain that is the only approved TB vaccine. It is administered via the intradermal (ID) route to infants in most countries around the world, although the United States actually never routinely vaccinated with BCG. We compared the standard dose of BCG (5 × 10⁵) administered ID with a higher dose (5 × 10⁷) delivered either ID, via aerosol, or i.v. using the very susceptible rhesus macaque model. Both high-dose ID and aerosol routes led to increased Mycobacterium-specific CD4 T cells in the airways, but the i.v. route resulted in significantly higher CD4 and CD8 T cells in the airways. There were also higher levels of Mycobacterium-specific Abs in the i.v.-vaccinated animals. Upon challenge, the standard-dose ID-vaccinated animals had a very minor degree of protection compared with unvaccinated animals. To our surprise, the i.v.-vaccinated macaques showed exceptionally robust protection against M. tuberculosis challenge, with 6 of 10 macaques showing by PET CT no signs of infection and no M. tuberculosis bacilli anywhere at 12 wk postchallenge; three others had <50 M. tuberculosis bacilli in the lungs, all contained within a single granuloma. Overall, the BCG i.v. route resulted in 90% protection and a 100,000-fold reduction in overall bacterial burden. In contrast, the high-dose ID or aerosol routes resulted in no difference in protection compared with the standard-dose BCG ID route. Although there is much work to do to determine the mechanisms responsible for this remarkable level of protection, the results give us new hope that sterilizing protection by vaccination is possible against TB.

I would like to close with a couple of take-home messages. Above all: trust yourself. Be brave. Embrace complexity. Do what you need to do to answer the question, even if it is hard and scary. Finally, work with good people and people you like!

I have so many people who have been crucial to my journey: my past mentors, especially Dr. Barry Bloom and Dr. Olja Finn; all my former and current trainees; my outstanding collaborators; my colleagues; and friends. Finally, I am so grateful to my family. My parents; my sisters; my incredibly supportive husband, Mike Cascio; and my children, Annie, Ray, and Rose. Without this support, I could not have accomplished a fraction of what we have done and what we will continue to do. I would also like to thank the NIH and the Bill & Melinda Gates Foundation for support of my research over the years.

Thank you to my fellow AAI councillors, Michele Hogan, the outstanding AAI staff, and all of you for making AAI the best organization. Enjoy the meeting!