Travels in time: Assessing the functional complexity of T cells

When the host encounters a pathogen, the ensuing immune response involves a complex set of cellular responses distributed across many different types of cells. In T and B lymphocytes of the adaptive immune system, these responses include irreversible differentiation events that generate functionally specialized subpopulations of cells (1). Understanding how pathogens and vaccines influence the number, type, and efficacy of specific differentiation states in the T-cell compartment is a major goal in immunology. The study by Han et al. in PNAS (2) interrogates the functional response of individual T cells over time using a nanofluidic platform. Their experiments reveal that the sequence of cytokines released over time by individual activated T cells is highly diverse but tightly programmed. Their work suggests that understanding the complexity of the T-cell response may not only be a matter of cataloging the possible phenotypes present in a cross-section of the T-cell population but will also need to involve understanding how those functions change over time.

Diversity Training

During a prototypical immune response to, say, a virus, naïve T cells that happen to express T-cell receptors specific for epitopes from viral antigens can become activated, proliferate, and gain effector functions (3). This initial burst of proliferation yields a large number of effector cells equipped with the ability to secrete cytokines that wake up the rest of the immune system and the cytotoxic machinery needed to destroy host cells that are harboring virus. After the virus is controlled, this army of effector cells contracts through the voluntary demise of most of its number, leaving a smaller population of T cells that forms the memory T-cell pool. This population of memory lymphocytes is the reservoir of immune protection that insures the host against future encounters with the same enemy.

Although this train of events is well characterized, it is clear that the population of memory T cells generated after an infection is highly diverse and contains many qualitatively different subclasses of T cells (4, 5). Characterizing the differentiation states that exist in the pool of memory T cells has been a major preoccupation of immunology research for the last 25 y. Immunologists equipped with flow cytometers, monoclonal antibodies, and the advantage of studying a tissue that exists in single-cell suspension have been able to categorize the T-cell compartment with increasing sophistication. Indeed, some of the seminal discoveries in T-cell immunology have been those that have

The present work by Han et al. provides some crucial information about the polyfunctional T-cell response.

shed light on the heterogeneity of this memory pool. For instance, variations in the expression of isoforms of CD45 were able to distinguish T cells that had never been called to arms (naïve T cells) from their battle-hardened counterparts (memory T cells) (6). Subsequently, Sallusto et al. (7) showed that the memory lymphocyte pool could be further subdivided by the expression of homing receptors like CCR7 into those that tend to confine themselves to the lymph nodes (central memory) and others that vigilantly patrol the borders of the peripheral tissues (effector memory).

In addition to heterogeneity in the phenotype of T cells, differences in their functional profile—the mixture of cytokines that they secrete after stimulation via their T-cell receptor—can subdivide the T-cell compartment into distinct fractions. Fundamental observations by Mosmann and Coffman (8) showed that CD4 T cells could be subdivided not only on the basis of surface markers but also by virtue of distinctive patterns of cytokines secreted after stimulation. This finding led to the understanding that naïve CD4 responding to antigenic stimulation in specific inflammatory environments could develop into to a range of lineages such as T helper cell 1 (Th1), Th2, or Th17 cells, each state being characterized by the capacity to secrete a distinctive combination of cytokine (9).

Correlates of Immunity

As the profusion of these functional subtypes of T cells continues to grow (to the general bewilderment of the nonimmunologist), it is important to note that surveying this complexity is more than just “immunologic stamp-collecting.” One of the central goals of this effort is to understand the relationship between the qualities of a T-cell response and the protection it confers to the host (10). Identifying correlates of protective T-cell immunity has been a particularly nettlesome problem, because protection does not seem to track neatly with any known T-cell state recognized to date. For instance, many vaccines elicit populations of antigen-specific CD4 T cells of the Th1 type that secrete INF-γ, TNF-α, or IL-2. However, although Th1 responses are well suited to combating intracellular pathogens, animal studies suggest that different vaccine-induced Th1 cell responses can vary widely in their ability to confer immunity to vaccinated animals. Work from Seder and colleagues (11) showed that vaccine-specific Th1 CD4 T cells are heterogeneous in the number of cytokines that are secreted by individual cells. A high frequency of individual Th1-type CD4 cells that secrete not just some but several cytokines simultaneously is the hallmark of a vaccine-induced T-cell response that does a better job of immunologic protection. The immunologic benefits of this trait, known as polyfunctionality, have subsequently been confirmed in HIV infection and vaccination (12–14). However, this work left many questions unanswered: are polyfunctional T cells a stable subset of T cells? If so, what controls their differentiation? Additionally, what are the hallmarks of this population of cells?

Trajectories in Time

The present work by Han et al. (2) provides some crucial information about the polyfunctional T-cell response. This work was enabled by the impressive technical expertise of the authors. Instead of using a flow cytometer to capture the functional state of individual cells at a particular moment in time, they studied the pattern of cytokine secretion over time by placing individual, live cells in nanowells, each in its personal bath of 125 pL of medium. This allowed the cytokine content of the medium in each well to be sampled repetitively over a 16-h period following T-cell stimulation and assayed for the presence of each of three Th1 cytokines (TNF-α, IL-2, and IFN-γ), much as a chef might repeatedly taste a pot soup while it cooks. This approach was combined with microscopy to confirm the phenotypic identity and viability of each cell.

A sophisticated analysis of this morass of data reveals some startling findings. The authors find that the point at which each cell commences secretion of cytokines after stimulation varies considerably, but that once set into motion, each cell tends to produce cytokines one a time, according to one of a large variety of particular sequences. Polyfunctionality does not seem to be the preserve of a particular subset of the cells, but rather is a transient characteristic of many different cells that had previously and would subsequently produce only one cytokine. Although this might sound like a rather chaotic arrangement, careful analysis of the sequence in which cytokines are produced demonstrates two key facts: (i) the range of cytokine sequences is not random but is instead limited to a finite, albeit numerous, set of patterns; and (ii) the gamut of cytokine sequences is different in T cells with different differentiation states (naïve, central, and effector memory cells).

What Does This Mean?

This work has two major implications. First, it suggests that polyfunctionality does not seem to be a persistent identity but instead is a condition that some T cells pass through transiently. Because flow cytometry captures the representation of polyfunctionality at a static moment in time, a time-dependent analysis of the T-cell response may therefore be a more informative way of capturing T-cell characteristics that correlate with immunologic protection. More importantly, the findings of Han et al. suggest that a polyfunctional T-cell response may be generated at the level of a population of cells, each taking turns at being polyfunctional, rather than by an elite group of specialized cells that are always polyfunctional. In other words, if polyfunctionality is the ideal T-cell state with which an organism could respond to a pathogen, then that state is represented by different members of a population of T cells at different times. This makes perfect sense in terms of how cellular networks should function and ought to resonate conceptually with immunologists who tend to think in terms of a multicellular response to pathogens. However, a major preoccupation of many immunologists has been identifying the characteristics of T cells that confer immunologic protection. The study by Han et al suggests that the search for a single differentiation state that correlates with the ability to confer protective immunity may need to be adapted to encompass a group of cell states, each with the potential to provide one of a number of functions at any given time.

The second implication is that the T-cell response seems to be divided up into a huge number of subpopulations that vary in the sequence of the cytokine(s) that they secrete. The skeptical view would hold that these sequences are nothing more than stochastic events in the life of a T cell: observe enough T cells for enough time and you will see all possible permutations. However, that seems not to be the case. As large as the number of cytokine sequences observed by Han et al. was, it was still just a small fraction of all possible cytokine sequences that could be generated. In other words, these cytokine sequences are deterministic and suggest the existence of regulatory mechanisms that specify their trajectory. Consistent with this, Han et al. found that the distribution of cytokine secretion sequences was different depending on the differentiation state of the cell (naïve, central memory, or effector memory). Han et al. are not alone in finding that there are many more functionally distinct types of T cells than previously appreciated. Recent data from another group who also used nanofluidic approaches to survey complex patterns of proteins secreted by individual cells also showed staggering but finite complexity of functional profiles of T cells (15). Thus, the job of understanding heterogeneity in the T-cell compartment may be orders of magnitude more complex than we currently believe.

As exciting as these results are they still leave many questions unanswered. First, Han et al. studied cells stimulated not with cognate antigen but with nonphysiologic stimuli designed to kick-start many T cells at once. It will be important to confirm their results in populations of antigen-specific of cells tickled by antigen, not bludgeoned with phorbol esters. Second, the frequency of cells displaying any one cytokine trajectory is very low, making it technically demanding to investigate the mechanisms that regulate that functional pattern. Last, it remains to be seen whether this time-integrated approach to surveying polyfunctionality is a better proxy for immunologic protection than the snapshot afforded by flow cytometry. However, the work by Han et al. suggests that understanding the immense heterogeneity inherent to a T-cell response to pathogen may require equivalently comprehensive experimental approaches such as those that they have developed.