“Doubt No More”: Two Sizes Fit All

In this commentary, we highlight a landmark study by Salgame et al. (1) that reaches back to the foundational era of functional T cell subsets: the late 1980s and early 1990s. Present-day understanding holds that CD4⁺ T cells can differentiate down several pathways to generate effector cells with distinct functional properties. At last count, perhaps five types of effector are classically taught, which include Th1, Th2, Th17, T follicular helper, and T regulatory cells, and that count is not distinguishing between various regulatory subtypes. We teach that these distinct forms carry out a specialized tuning of immune responses, allowing adaptation to different challenges posed by pathogens with variable lifestyles, which require different “modules” of effector function to eliminate. Although today’s textbooks present these ideas rather blandly, along with the requisite inducing cytokines and transcriptional programs, they leave out a great deal about how this framework came into being, and they pay little attention to its experimental basis—especially the contemporary dialogue and controversies that surrounded its evolution. So it is occasionally worth reviewing the evolution of our knowledge, even if only to remind us that establishing what are now dogmas in immunology is almost never as straightforward as presented in current textbooks. For this story, it suits us to set the stage properly.

A minimalistic summary of Salgame et al. (1) would simply state that it provided definitive confirmation that humans follow the Th1/Th2 paradigm recently established in mice a few years before. However, that would not properly credit this Pillars of Immunology article, or the many preceding “pillaresque” studies on which it is based. Moreover, it would omit a description (or more an insinuation) that discovery of “Th1/Th2” can in some ways be viewed as an inadvertent stepchild of a much more interesting (and lengthy) saga in the history of immunology.

To fully appreciate this Pillars article, it requires us to go back a ways. Arguably, the most important accomplishment of the last century, within immunology anyway, was defining the molecular basis of T cell recognition of Ag. The 1908 Nobel Prize to Ehrlich and Mechnikov documents the appreciation even then that immunity could be mediated by cells, and not just by “Ab.” Although genetics itself was sparked into existence only in 1900 by De Vries, Correns, and von Tschermak (2, 3), the long march to understanding T cell Ag recognition might be traced to 1909, when Tyzzer (4) reported a genetic component in the rejection of transplanted tumors in Japanese waltzing mice. And in 1916, Tyzzer had recruited C.C. Little into the search for the genetic basis of tumor rejection between different strains of inbred mice (4, 5). With Little’s support, Snell and colleagues (6) eventually identified the “H-2” locus, prompting interest in identifying the relevant surface proteins mediating (7) and allowing assignment of specific “haplotypes” to an increasing number of inbred mouse strains (8). Those genetic tools allowed Zinkernagel and Doherty (9) to discover “H-2 restricted” T cell recognition. Following evidence of Ag processing (10) and discovery of peptides as elemental antigenic ingredients (11, 12), the actual structure of human HLA-A2 with bound peptides reported in 1987 provided the essential epiphany in understanding H-2–restricted T cell Ag recognition (13, 14). With the mystery of T cell recognition solved, what was next?

The drive to understand T cell Ag recognition during this long saga had spawned progress in two other particularly relevant areas. First, it is probably easy to forget the importance of the development of T cell cloning technology (15–17). But the ability to clone T cells of defined Ag specificity and maintain them in vitro almost definitely was not a trivial accomplishment, and it was essential for discovering T cell heterogeneity. Second, enumeration and cloning of the immune system’s soluble mediators (i.e., cytokines), such as IFN-α (18), IL-2 (19), and eventually those relevant to our dichotomy, IFN-γ (20) and IL-4 (21, 22), was also essential. Each of these advances was relatively new in the mid-1980s, when T cell heterogeneity began to raise its multiple heads, but it is important to state that this phase of discovery would not have been possible without both of these advances, particularly because the former was substantially facilitated by the latter.

By the mid-1980s, evidence for heterogeneity among CD4⁺ T cells had accumulated from several studies (23–27) and was based largely on their abilities to provide help for B cells. In 1985, Kim Bottomly analyzed a series of murine I-a–restricted L3T4a⁺ (i.e., CD4⁺) clones for characteristics of B cell help (27). She described “type 1” and “type 2” T cell clones, differing in support for B cell expansion in vitro, with type 2 being superior in this capacity. Then, in 1986, Coffman, Mosmann, and coworkers (28) similarly reported two types of murine CD4⁺ T cell clones now defined on the basis of cytokine production, particularly a reciprocal production of IFN-γ by Th1 clones and IL-4 by Th2 clones, and this prior Pillars article is often cited (or blamed) as the origin of the Th1/Th2 dichotomy. That same year, Locksley reported a related functional “dichotomy” in defense against the intracellular parasite Leishmania major, with successful defense being associated with the IFN-γ production (29). His group later documented CD4⁺ T cell involvement and reciprocal susceptibility associated with Th2 and IL-4 production (30–33). With nearly simultaneous work by Mosmann and colleagues (34, 35) that related Th1 responses to delayed-type hypersensitivity, a strong case for a functional dichotomy had been made, at least for the CD4⁺ T cell biology in mice (36).

But did this framework have anything to do with human immunity? Although difficult to document by citations, there has always been a healthy skepticism regarding the coherence of mouse immunology for its relevance for or application to humans. As early as 1988, Sergio Romagnani had suggested, then confirmed, the role of IL-4 production by human T cell clones as a basis for provision of help to B cell IgE production (37, 38). However, his initial evaluation of a large panel of 690 human T cell clones did not appear to support a “clear-cut” dichotomy of IFN-γ and IL-4 production as described in mice (39), lending credence to a chorus of contrarians that somehow things were different in the human, but this analysis had examined T cell clones induced by PHA or alloreactivity. In subsequent analyses reported in 1991, a series of human T cell clones derived from infiltrating cells in vernal conjunctivitis were found to be of the Th2 type, producing copious IL-4, but not IFN-γ (40). Contemporary work by Kapsenberg and coworkers (41) and Romagnani and colleagues (42) reported that most allergen-specific human T cell clones from atopic donors produced IL-4, but not IFN-γ, whereas clones specific for bacterial components derived from the same donors produced IFN-γ, but rarely IL-4. By August of 1991, there was greater enthusiasm for extending this paradigm into human immunity, encapsulated in a review by Romagnani (43) with the titular exhortation “doubt no more.”

By this time, opposing functions of IFN-γ and IL-4 had been established in the mouse for defense against Leishmania (33), as well as defense against filariasis in humans, where these cytokines exerted opposing actions on IgE production in response to infection (44). However, the study did not demonstrate that the opposing dichotomy was exerted by distinct CD4⁺ T cell subsets at the clonal level. Indeed, this specific point is where two back-to-back publications in Science in 1991 by Modlin and Bloom (1, 45) added to the multilayered strata of the Th1/Th2 dichotomy. These two studies examined human infection by Mycobacterium leprae, the causative agent of leprosy (1, 45). Leprosy in humans was known to manifest variably across a spectrum, with a resistant “tuberculoid” form at one end and a susceptible multibacillary lepromatous form at the other (46, 47). Tuberculoid leprosy was characterized by a predominantly CD4⁺ T cell infiltrate, and lepromatous leprosy by what was then considered a suppressor population. First, Modlin and colleagues’ study (45) directly examined the cytokines expressed in skin biopsy specimens taken from lesions of individuals with leprosy classified as either tuberculoid or lepromatous. This was performed at a time when cytokines could be measured by PCR of cDNA generated by reverse transcription of extracted mRNA (RT-PCR). Their survey included TNF-α, GM-CSF, IL-2, IFN-γ, IL-4, IL-5, IL-10, and others, a fairly extensive panel for the time. Although the data preceded the development of more quantitative RT-PCR methods, the PCR products were examined directly by electrophoreses and robustly demonstrated the presence of IFN-γ expressed in most tuberculoid biopsies, but not in lepromatous biopsies. Conversely, expression of IL-4 and IL-10 was observed in lepromatous lesions. These results established an association of distinct cytokine patterns with resistant or susceptible forms, such as the mouse pattern for L. major reported by Locksley (29).

Second, Bloom and colleagues’ study (1), our present Pillars article, generated a panel of human CD4⁺ T cell clones with specificity to M. leprae from individuals across the spectrum of leprosy, from their familial contacts, or with specificity to unrelated Ags such as tetanus toxoid or flu. The data in the single table comprising this paper reported the clone derivation, specificity, and production of the cytokines IFN-γ, IL-4, IL-5, IL-6, IL-2, GM-CSF, and TNF-α. All CD4⁺ T cell clones from Ag-responsive M. Leprae–infected individuals produced IFN-γ, as well as IL-2 and GM-CSF, but not IL-4 or IL-5. In contrast, CD4⁺ T cell clones specific for tetanus toxoid were predominantly producers of IL-4 and IL-5, but not IFN-γ. By showing essentially a nonoverlapping pattern of cytokine production at the clonal level, this short study cemented the relevance of the Th1/Th2 framework in the setting of human pathogen responses. With the inclusion of humans, the field as a whole would soon enthusiastically move on to the next phase, the examination of how these CD4⁺ T cell subsets develop and perform, a task that would consume the best part of the next decade.

Of course, this dichotomy was not to remain static, and it is fair to say that the original paradigm might have been a bit overused at times before its subsequent expansion. For example, by 2003, one commentary eloquently argued that it was no longer possible to fit all CD4⁺ T cells into the “procrustean paradigm” that two sizes fit all (48). But with the self-correcting action of science, eventually more differentiated states were recognized, beginning with rejuvenated enthusiasm for “regulatory” cells brought about by Powrie, Coffman, and colleagues (49, 50). But that is a story for another Pillars.