Separate and Unequal

If their numbers are so small, how are these critical immune cells replenished? Vast quantities of battle-ready neutrophils can be churned out from blood stem cell precursors located in the bone marrow in a production line-like operation. Lymphocytes, however, cannot be maintained simply by continual or increased production of duplicates from the cells, as is done for neutrophils. Yes, blood stem cells can produce the immature predecessor of lymphocytes, but these cells lack the critical receptor to recognize antigen. The mature lymphocytes, which each express a unique receptor, are effectively irreplaceable by traditional stem cells, because each one is bar-coded to occupy its own singular place in the defense system.

Yet a mechanism for the replenishment of useful lymphocytes would seem to exist, since a cardinal feature of lymphocyte-mediated responses is that they 'remember' prior encounters by responding faster and better on subsequent challenges than during the initial confrontation. One of the major questions, therefore, facing those of us in the field has been: When a useful and unique lymphocyte is called to battle, how is its replacement insured? We now think we may be closing in on the answer.

Naive T cells are continually patrolling the lymph nodes and related lymphoid tissues, spending roughly a day moving randomly through a lymph node and then returning to the blood briefly before entering another lymph node for a day of patrolling. Antigens from microbes that enter the body are picked up by dendritic cells and transported to the closest lymph node, where they are presented to patrolling T cells. When the rare, patrolling T cell finally encounters a dendritic cell bearing antigen and the marks of an encounter with a microbe, its dynamic movements within the lymph node take a different course. In the last few years these interactions have been recorded in real-time movies.

At the outset of the immune response, the T cell abandons its random motion and begins to engage in short interactions with the antigen-loaded dendritic cell. These contacts last three minutes or less and occur repeatedly for the first 10 hours. Then the rapid activity and motion ceases as the T cell engages in prolonged contact with the dendritic cell, lasting for the next 20 hours. This usually ends with the T cell dividing into two daughter cells.

During the initial activation process of a naive T cell, it receives specific instructions from the dendritic cell about the nature of the threat: in which tissue the infection is located, how abundant it is, and whether it is a virus, bacterium, fungus, or parasite. This information will ultimately dictate the differentiation fate, which is a set of discrete and stable changes in gene expression in the cells that actually do the fighting and eliminating, namely the effector cells. In addition to that specific information from the dendritic cell, there is always another instruction that is generic to the immune response: Duplicate, duplicate, duplicate! A naive cell must divide based on its sheer rarity; more is better when it comes to the elimination of microbes that are themselves rapidly replicating.

Ten years ago immunologists were concerned with finding the precise order of these two events. Does a naive cell first transform itself into an effector cell and then divide to produce many duplicates of similarly fated cells? Or does duplication come before the adoption of a new fate, and rather it is the daughter or the granddaughter cells that are actually doing the converting? Despite some initial resistance, the latter scenario is a prevalent working model in the field now. This model, moreover, opens up the intriguing possibility that the descendants of a single T cell can adopt different fates, so that one naive cell could spawn diversity among her daughters, rather than spewing out homogeneous carbon copies.

At the same time that we were studying cell division as a formative step in converting cell fate, and thinking about T cells giving rise to diverse daughter cells, spectacular advances were being made elsewhere. A series of papers were published describing the organization and segregation of T cell molecules at the site of interaction between the T cell and its antigen-presenting cell. These studies revealed a rapid and elaborate kaleidoscope of arrangements, by both T cell-surface receptors and subsurface molecules. These components converge and reorganize around the intercellular junction between the T cell and its stimulatory neighbor. The term given to the point of contact is the immunologic synapse, because of the similarity to the synapses of the nervous system.

The immunologic synapse was first observed in mature effector T cells, where this multimolecular and cytoskeletal polarization makes perfect sense. By directing the secretion of toxic or inflammatory cargo in the direction of antigen, it allows the killer T cell to deliver the kiss of death to an infected target cell with precision, achieving maximal efficiency with minimal collateral damage. What was not clear was whether naive T cells became polarized in this manner when they first encounter stimulatory dendritic cells in an immune response, and if so, why. We began to entertain the idea that this polarization and segregation of cellular components could have a critical impact on the naive T cell that was preparing for its first division. The idea was a relatively simple one; testing it would be much harder.

Our hypothetical scenario was a fusion of three ideas: First, the T cell may want to diversify its daughter cells; second, T cells have a capacity to become lopsided in orientation to a stimulus; and third, T cells seem to remain in contact with a stimulatory neighboring cell until they divide. We were intrigued that a newly activated naive T cell crosses the point of no return (the G1/S boundary) of the cell cycle at approximately 10 hours, the same time that the movies reveal the transition to prolonged contact. It seemed as if the T cell was making prolonged contact from the time it knew it would give birth until the time it would complete that first division.

We predicted that the signaling molecules and the nucleation of cytoskeletal microtubules would segregate to the side of the T cell that was in contact with the dendritic cell, radiating from the immunologic synapse. Since the T cell has by this point committed to the act of cell division, it would soon have to contend with other chores. Notable among these is duplication of the microtubule organizing center and its migration 180 degrees away from the original nucleation of microtubules. This is required to pull apart the two sets of chromosomes into each of the incipient daughter cells.

Our thinking was that the original cluster of microtubule ends becomes anchored at the synapse side of the cell to direct the secretion of hormones towards the dendritic cell. The presence of this first microtubule-organizing center might dictate the plane of cell division, as the second pole of the spindle is positioned 180 degrees away from the first. In this way the naive T cell would always have one daughter that was close to the point of contact with the dendritic cell and one daughter that was remote from the point of contact. Given that the contact point might also be highly enriched for certain molecules, we suddenly have a mechanism that not only tells the naive T cell where it should cleave, but also one that would reliably create inequality, or lopsidedness, in the inheritance of various molecules to the two daughters. Many of those predictions are essentially what we found in our initial experiments to test this model.

My colleagues and I initiated the first division of an immune response in a model system. We performed microscopic analysis of well-studied T cells that recognize known bacteria and parasites. In parent T cells that were in prolonged contact with dendritic cells, we observed the segregation of cell-surface receptors and reorganization of the microtubule cytoskeleton - the hallmarks of the immunologic synapse. When the T cells were in the midst of division, we found that a number of molecules segregated to one or other end of the spindle of microtubules. As the naive cells were actually cleaving in half, it was clear that the segregation of components at the poles of the spindle resulted in unequal inheritance for the two daughters.

Because we could identify the daughter cell that was close to the contact with the dendritic cell (the proximal cell) and the daughter cell from the far side of the contact (the distal cell), we began to look at indicators of their cell fate. The two cells looked and behaved as if they were each already heading towards different cell lineages. The proximal daughter cell appeared as though she would beget effector cells, leading the fight to eliminate the threat. The daughter from the far side appeared as if she would spawn memory cells, essentially filling the void (and more) left by the naive cell.

This bears similarity to a common strategy for maintaining tissue function in long-lived beings with short-lived cells: the stem cell or regeneration paradigm. For example, in our skin, stem cells at the bottom-most layer divide and give rise to a daughter that replaces the mother and a daughter that grows upward and begets terminally differentiated granddaughters.

The concept of memory cells as the replacement patrollers of the lymph nodes once the mother cell is gone has been entertained throughout the last 10 years. For the most part, however, the prevailing view was that daughter cells were uniformly destined to become effector cells and that memory cells arose because some of the warriors were lucky enough to survive the battle. The problem with that notion is that migration from lymph node to tissue is usually a one-way trip. So, even though some effector cells might be long-lived beyond the battle, they would not fulfill the role of naive cells patrolling the lymph nodes. Moreover, these effector-like cells that survived the conflict wouldn't have the capacity to replace themselves when they finally did die of old age.

A stem cell-like model of memory is, therefore, a potential resolution to what is the quintessential paradox of the lymphocyte-based system of defense. That paradox is that the activation of a lymphocyte means its receptor is priceless but apparently lost as the descendants of this cell fight to the death. Recognizing that some descendants are immediately earmarked as replacements allows the lymphocyte to be simultaneously used and replenished.

In skewing the inheritance of the daughters, the T cell is exploiting an ancient strategy for multicelled beings to reliably diversify the fates of sibling cells. This process is known as asymmetric cell division and is used by a variety of organisms to form the pattern of the body plan. It is also widely used to allow stem cells to give birth to one daughter that preserves the lineage and another daughter that is destined for function and then demise. For stem cells, generally, the proximal daughter, the one close to the protective niche, is the replica of the mother, while the distal daughter undergoes terminal differentiation. Curiously, the situation may be reversed in the case of immunity: The proximal daughter that is in contact with the dendritic cell seems to undergo terminal differentiation, while the distal daughter inherits the capacity to patrol the lymph nodes like the naive mother cell. Perhaps this difference is owing to the dendritic cell's role as an alarm of new danger rather than as a protective niche.

Where do we go from here? We anticipate being able to see the moving components inside the cell as the first divisions of T cells occur in living beings. To achieve this we are trying to produce higher-resolution movies that will allow us to identify the specific interactions that trigger the T cell to move its key components around. We also aim to chronicle the behavior of the proximal and distal daughters after they have separated - to understand where they go next, which cells they communicate with, and what fates they beget. We are also speculating about other functions for the interaction with the dendritic cell: Perhaps it allows some T cells the opportunity for asymmetric divisions that would yield daughter cells widely suited to the kind of microbe being fought.

Looking beyond the immune system, we predict that asymmetric division plays an as-yet undiscovered role in a variety of biologic processes. The final head-to-tail pattern of a worm gets set up simply by the place where sperm enters egg during fertilization, because that points the direction of a first asymmetric cell division. Similarly, T cells seem to use the place where they temporarily lay down to organize asymmetric division. This raises the possibility that other cells, which transiently communicate with cellular and noncellular surfaces before they divide, might do the likewise - as embryos are growing, organs are forming, cancers are spreading. In a world full of beings that are filled with many different types of cells, we simply do not yet know the extent of the role of asymmetric division in making daughter cells different. It might rear its head in some unexpected places.

Steven Reiner is a professor at the Abramson Family Cancer Research Institute at the University of Pennsylvania.