CO-Opting the Host HO-1 Pathway in Tuberculosis and Malaria

Abstract

Infection with Mycobacterium tuberculosis and Plasmodium species results in upregulation of the host heme oxygenase-1 pathway. In tuberculosis infection, this leads to upregulation of the bacterial “dormancy regulon,” whereas in malaria, it enhances the efficiency with which sporozoites develop into exoerythrocytic stages. Here we discuss these findings as well as some of the interesting questions they raise.

Tuberculosis (TB) and malaria are two of the world’s most important infectious diseases; approximately 3.7 million deaths per year and a staggering burden of morbidity are attributable to these diseases alone. Moreover, drug resistance has made both infections more difficult to treat, and neither is preventable with an efficacious vaccine.

Two papers in this issue of Cell Host & Microbe reveal a previously unknown role for the host enzyme heme oxygenase (HO) in these infections (Epiphanio et al., 2008; Shiloh et al., 2008). HO catalyzes degradation of heme to biliverdin, carbon monoxide (CO), and free iron, and expression of HO-1, the inducible isoform of the enzyme, is upregulated by stimuli that include oxidative stress, hypoxia, and diverse cytokines. CO and biliverdin have potent anti-inflammatory, antioxidant, and antiapoptotic effects and have been studied for their contributions to modulating outcomes in organ transplantation, acute lung injury, and endotoxic shock (reviewed in Otterbein et al., 2003). Investigating the role of HO-1 in the pathogenesis of infectious diseases, Epiphanio et al. report that HO-1 contributes to the establishment of malaria infection by decreasing the host inflammatory response to pre-erythrocytic stages, while Shiloh et al. report that HO-1 and CO induce the “dormancy regulon” of M. tuberculosis and may contribute to latency in tuberculosis infection.

Malaria infection has two distinct stages in the mammalian host: a clinically silent pre-erythrocytic stage that is required for establishment of infection, and an erythrocytic stage, responsible for all of the clinical symptoms of malaria. Previously, Mota and her colleagues, using a rodent model of malaria, showed that HO-1 modulates the inflammatory response induced by erythrocytic stages and can protect mice from experimental cerebral malaria (Pamplona et al., 2007). Their present study focuses on pre-erythrocytic stages and reveals that sporozoite infection in the liver induces expression of host HO-1, which contributes to establishment of liver stage infection (Epiphanio et al., 2008). Abolishing HO-1 activity by gene deletion or siRNA results in a significant decrease in the liver parasite burden. Importantly, they provide evidence that the mechanism by which HO-1 is acting involves its role as a modulator of the innate immune response. They find that HO-1 induction is associated with decreased levels of the chemokines MIP-1α and MCP-1, significantly less neutrophil and macrophage infiltration into the infected tissue, and lower levels of the pro-inflammatory mediators TNF-α and IL-12, all of which likely protect the developing exoerythrocytic forms from immune-mediated destruction.

Like malaria, TB can be considered to have two stages: active disease and latent infection. During latent infection, there are no symptoms or signs of infection, and latent TB can only be detected by assaying immune responses to mycobacterial antigens. The importance of latent infection is that it does not necessarily remain latent: approximately 10% of individuals reactivate and become ill and infectious to others. While the host and microbial determinants of latent TB are subjects of ongoing investigation, one potential microbial determinant, a set of genes termed the “dormancy regulon,” is regulated by decreased oxygen tension (Park et al., 2003; Sherman et al., 2001) and host-derived nitric oxide (NO) (Voskuil et al., 2003), stimuli that are likely associated with latent infection. The current study by Shiloh et al. (2008) and a recently published paper by Adrie Steyn and colleagues (Kumar et al., 2008) add host-generated CO to the list of stimuli that induce the M. tuberculosis dormancy regulon. The dormancy regulon is activated by the DosS/T-DosR two-component system, where DosS and DosT are sensor kinases and DosR is the cognate response regulator (Roberts et al., 2004). On the surface, it might not be surprising that another factor that can alter the redox state of heme-containing proteins, such as DosS and DosT, can regulate the dormancy regulon, but it may surprise some that CO is generated in response to infection and can contribute to the outcome. The Cox and Steyn groups found that HO-1 is induced in cultured macrophages infected with M. tuberculosis and is expressed in lungs of infected mice. They then found that physiological concentrations of CO rapidly induced expression of the genes in the M. tuberculosis dormancy regulon in a manner nearly indistinguishable from that of NO, although unlike NO, CO had no direct effects on bacterial growth. Further investigation revealed two findings of note. First, while indistinguishable in their sensitivity to NO, the absence of dosT enhanced, while absence of dosS diminished, gene induction by CO in the paper by Shiloh et al., while the other group reported that dosT is unresponsive to CO (at an unspecified concentration of CO). This suggests that these two sensor kinases may make distinct contributions to regulation of the dormancy regulon at various stages of mycobacterial infection. Second, HO-1-dependent induction of the M. tuberculosis dormancy regulon in macrophages exhibited an absolute dependence on the presence of the inducible nitric oxide synthase NOS2, even though CO was sufficient for gene induction in vitro. This suggests a complex relationship between these signaling systems that may optimize the dormancy response in vivo.

To date, most of the work on the anti-inflammatory effects of HO-1 activity has focused on noninfectious inflammatory conditions, such as cardiac ischemia/reperfusion injury and allograft rejection, where HO-1 through its byproducts, CO and biliverdin, modulates immune responses and reduces tissue injury (reviewed in Otterbein et al., 2003). Given its ability to downmodulate immune responses, it stands to reason that HO-1 has a role in infectious processes, and further study will determine if HO-1 induction is a more general phenomenon associated with pathogens as they trigger an inflammatory response. The consequences of HO-1 induction, however, are likely to be pathogen specific, as illustrated here by its divergent downstream effects in TB and malaria. In addition to determining whether this is a general phenomenon in host-pathogen interactions, it will be important to determine the trigger(s) for host HO-1 induction. Do Mycobacteria and Plasmodia induce HO-1 directly to better ensure their survival in the host, or is HO-1 induction a response to reactive oxygen or nitrogen or proin-flammatory cytokines that helps to restore homeostasis and limit inflammatory damage? Evidence that CO can induce the dormancy regulon of M. tuberculosis suggests that the bacteria employ CO to ensure that they regulate their gene expression for optimal survival, even when cytokine production and immune cell recruitment subside. While the results provide strong evidence for CO induction of the M. tuberculosis dormancy regulon and for establishment of liver stage malaria infection in rodent models, it remains to be determined whether HO-1 and CO contribute to the course of tuberculosis and malaria in humans.

Since the human HO-1 gene possesses functional polymorphisms in its promoter region that are known to regulate transcription (reviewed in Exner et al., 2004), it will be important to determine whether differential activity of HO-1 alters the outcome of TB or malaria infection. For example, are certain HO-1 alleles differentially associated with prolonged latency versus progressive primary infection in tuberculosis, or with uncomplicated versus severe malaria infection? Recently, a small study from a malaria-endemic region of Myanmar found an increased incidence of cerebral malaria in individuals with HO-1 alleles associated with increased HO-1 expression (Takeda et al., 2005), which, given the current findings of Epiphanio et al., may reflect susceptibility to a large inoculum of sporozoites during initial infection. To date, epidemiological studies have demonstrated that polymorphisms that increase HO-1 expression are associated with protection from disease states such as restenosis after angioplasty and organ rejection after kidney transplantation (Exner et al., 2004). Since infections such as TB and malaria are much more likely to exert selective pressure on the human population than are inflammatory conditions resulting from medical interventions, it will be interesting to know whether TB and malaria have played a role in the selection of HO-1 alleles in different human populations.

In addition to further studies of polymorphisms in the human HO-1 gene, assays of carboxyhemoglobin, or CO concentrations in exhaled breath, can be readily performed under field conditions and may yield valuable information on the contributions of CO to infectious disease outcomes. Distinct environments have large differences in CO concentration, and it would be important to determine whether CO generated by wood stoves in poorly ventilated homes, or generated by motor vehicles in increasingly dense urban environments, has an impact on malaria infection or contributes to maintaining TB as a latent infection. Since little is known of the specific environmental or genetic factors that determine outcomes in either malaria or TB, these new findings on CO suggest a starting point for epidemiologic and translational studies.