Skip directly to content

Frontiers of Immunology

tagline 1999

Thymic Rebound, Naïve CD4+ T-Cell Repopulation Demonstrated in Individuals on HAART

Vive le thymus!

Even the most cutting edge immunology postdoc would have had his or her hands full preparing lesson plans from this month's "Frontiers of Immunology" piece, exhaustively stitched together from several late-breaking and provocative research papers by TAG's Mark Harrington: What really causes the T-cell depletion observed in HIV infection? Are T-cells being destroyed and replaced at the voracious clip we've been led to believe? What is the thymus' role in all of this? And might there be extrathymic sites for maturation and emigration of the new naïve T-cells?

While it's still a bit early for a definitive resolution to these complex biological inquiries, innovative studies performed with HIV-infected individuals on potent antiretroviral therapy are actively pushing forward the boundaries of our immunological knowledge. In this year alone, three ground-breaking papers have been published which weigh in on these sometimes highly contentious issues. And Mark Harrington skillfully identifies and elucidates the common thread which weaves them into a coherent whole. Out of necessity, most of the equations and more arcane methodology have been airbrushed out of this report in the interest of reader friendliness -- leaving, we trust, the underlying scientific arguments largely unadulterated. The more intrepid (or masochistic) reader may, however, wish to track down the original papers for a completely comprehensive immunological experience. With those TCR geeks and kinetic wonks in mind, abbreviated citations have been provided within the text itself -- and more complete ones as footnotes within the tables. Full text of the three papers (Hellerstein/McCune et al., Douek/Koup et al., and Mohri/Ho et al.) is also available on the web (but in some cases may require a user name and password).

Among the hoariest dogmas of human immunology is the notion that the thymus, finishing school for the mammalian T lymphocyte, is dormant in adulthood. Studies of conditions such as cancer and bone marrow transplant, among individuals whose immune system was destroyed by chemotherapy, demonstrated an age-dependent gradient of T cell repopulation. The younger the patient, the faster the immune reconstitution. However, until recently it had always been assumed, rather than demonstrated that the thymic atrophy associated with aging resulted in a complete lack of production of new naïve T lymphocytes.

Recently, several HIV immunologists have begun to investigate the role of the thymus in HIV infected children and adults, before and after highly active antiretroviral therapy (HAART). Their discoveries have overturned established immunological doctrine, suggesting that the thymus remains active throughout adult life and that HAART appears to permit significant, although not complete, CD4 immune reconstitution.

A raging controversy in HIV pathogenesis has long been -- what causes the CD4 depletion which leads to AIDS-impaired production of new CD4 T cells, viral or immune killing of existing CD4 T cells, or accelerated turnover or a shorter half-life of CD4 T cells in infected individuals?

Since 1995, a series of studies have increasingly challenged the paradigm of high CD4+ cell turnover and destruction = CD4+ cell depletion. First, the estimates of the proportion of CD4+ T cells destroyed and produced every day were calculated on the basis of the initial rapid rise observed after initiation of HAART. This initial rise is, however, mostly the result of redistribution of CD4+ cells from other lymphoid compartments to the blood. Second CD4+ T cell production in HIV-infected subjects is not significantly different from that in healthy HIV-negative subjects. Third, the Hellerstein and Sachsenberg studies have provided ample evidence that, at least on the basis of blood analysis, CD4+ T cell turnover is two- to three-fold higher in HIV-infected compared with HIV-negative subjects (Pantaleo 1999).

Mark Hellerstein, Mike McCune and colleagues in San Francisco set out to directly measure CD4 T cell production, lifespan and turnover. They developed a non-radioactive labeling technique to measure the rate of T cell proliferation and the half-life of the T cells in uninfected individuals as well as in HIV-infected people before and after starting HAART.

They gave study volunteers glucose (sugar) infusions labeled with deuterium, a heavy form of hydrogen which is not radioactive, but which is incorporated into the DNA of newly-dividing cells. Presumably, then, all T cells formed after the infusions of the labeled glucose would have deuterium (heavy hydrogen) in their DNA, while older cells would be unlabeled. Each time the cell divided, the deuterium content would be diluted by fifty percent. They measured the amount of heavy hydrogen in cells removed from the volunteers at 5-7 days and 10-14 days post infusion, allowing them to estimate the T cell production rate, replacement rate, and half-life or survival time (see table below).

These results suggested that the replacement (proliferation) rate was 300% higher in HIV+ than HIV- subjects. Neither viral load nor CD4 nadir correlated with the replacement rate or rate of production of new T cells. After 12 weeks of HAART, the half-life of CD4 and CD8 cells appeared to be shorter and the replacement rate higher. Therefore, overall T cell production appeared to be "significantly higher in the subjects on HAART" than in either other group. Thus, at least in the blood, the half-life of both CD4 and CD8 lymphocytes appears dramatically shorter in HIV infected people, either untreated or treated, as compared with HIV-negative controls. However, the half-life of these lymphocytes on HAART appears to be even shorter than among individuals not on HAART -- at least when they have been on HAART for only up to twelve weeks.

The shorter half-life of circulating CD4+ T cells in the HAART group may reflect complex underlying population dynamics. It is possible that activated "memory" CD4+ T cells are dying at an accelerated rate while "naïve" cells are being produced in greater numbers, for example. Measurement of the dynamics of T-cell subsets will be required to establish the full consequences of HAART.

They stated that their results were consistent with Ki67 staining in advanced and early disease (Sachsenberg 1998; Zhang 1998; Fleury 1998) and with analyses from patients with unstable cellular DNA after radiation for cancer (Michie 1992; McLean 1995), but were inconsistent with reported rates of CD4 accumulation rates after HAART (Ho 1995).

The study had several limitations. It only measured T cells in the blood, not the lymph nodes. Moreover, it measured T cell changes after just 12 weeks of HAART, before the second phase of naïve cell reproduction has occurred. It did not measure T cell subsets such as naïve, memory, activated or resting. It measured cells drawn at only two time-points, just 5-7 and 10-14 days post-infusion; "cells that are slowly produced and/or released into the circulation (for example, recent thymic emigrants [emphasis added]) may not be measured even after 15 days of sampling."

They stated that "Our results are incompatible with a highly accelerated destruction of circulating CD4+ T cells that overcomes a higher than normal total production rate ('open drain/open tap') model," and concluded that their technique needs to be applied with "longer labeling times, variable times post-infusion for sampling, the inclusion of samples derived from tissue spaces of interest [and] the use of high-speed cell sorting [to] facilitate the kinetic analysis of even rare subpopulations of cells" (Hellerstein 1999).

T-Cell Kinetics: The Manhattan Version

In January, I asked David Ho about his reaction to Mark Hellerstein and Mike McCune's paper describing the non-radioactive labeling of T cells in HIV infection, which came up with very different T cell replacement kinetics than those described since 1995 by Ho and Perelson.

Ho explained that there is a common misunderstanding of what is meant by T cell production. If there is a stable T cell pool (T), which loses T cells by cell death (D), it replenishes its cells by two means -- proliferation (P) from existing cells (mainly memory cells) and production of new (mainly naïve) T cells from a source (S): P, S, T, D. S= Source (thymus?); P= Proliferation; D= Death. If T is at steady-state, S+P=D. If S is zero because of thymic damage, P=D. S is mainly thymocytes (see Ho's Science paper using bromodeoxyuridine (BrdU) labeling of T cells in rhesus macaques, Mohri 1998). Both Koup's and Ho's work demonstrates the existence of recent thymic emigrants (RTEs) throughout life, on an age-dependent gradient.

Neither the T cell labeling experiments of Hellerstein et al., nor the BrdU paper by Mohri et al., distinguishes between thymocytes and proliferating memory T cells. There are some other key differences between these labeling studies which differ in their estimate of T cell kinetics (see table below). Because the Hellerstein paper measured labeled T cells at only two timepoints it was unable to observe any differential (e.g., phasic) kinetics. In addition, lymphocyte division is a relatively rare event, so two days of labeling may have been insufficient.

Ho proposes to conduct a study using the same labeling system used by Hellerstein et al., but with five data points during labeling and ten data points after labeling to measure the fine kinetics of lymphocyte birth, proliferation and death. People would serve as their own control, by having their T cell kinetics measured for six weeks before starting on HAART and then for ten weeks on HAART, to see if therapy changed the kinetics. Ho and Perelson are working with Hellerstein and McCune to learn the technique of the latter while refining their math. It's a good example of collaboration in action.

Recent Thymic Emigrants: A Surrogate Marker for Thymic Output

Since 1997, studies by Brigitte Autran of the Hôpital Pitié-Salpètrière in Paris and others have demonstrated two phases of CD4 reconstitution in people starting HAART. After starting treatment, as HIV is removed and the lymph nodes quiet down, existing memory (CD45RO+) CD4 cells are redistributed from lymph nodes to the circulation. Then the number of naïve CD45RA+ CD4 cells begins a gradual yet persistent increase (Autran 1997, 1998; Haase 1998).

In 1998, Mike McCune and others showed that the thymus increases in mass after people start HAART, and that the degree of this increase is correlated with the number of new naïve cells created (McCune 1998). Could it be possible to measure more directly how recently the new naïve CD4 cells, which occur after initiation of HAART, have come from the thymus? Several researchers -- including Richard Koup of the University of Texas Southwestern Medical Center in Dallas, McCune and colleagues in San Francisco and, David Ho and colleagues in New York City -- are exploring the use of more specific markers of recent thymic emigration (RTE) to quantify thymic activity over the course of adult human life and before and after HAART.

At the Gallo lab meeting in August 1998, Koup previewed a study which appeared in Nature during December (Douek 1998). He counted circular fragments of DNA which are produced in T cells during thymic maturation, known as "alpha circles" or "TCR rearrangement excision circles" (TREC), to measure how recently a T cell emerged from the thymus. TCR rearrangement is the mechanism by which cells are programmed in the thymus to respond to foreign antigens but not to self tissue, preventing autoimmunity. TREC are formed by excision of DNA fragments during the formation of the T cell receptor (TCR) during T cell maturation inside the thymus.

When TCR rearrangement occurs, the TCR sequence is randomly reshuffled. This rearrangement determines which T cell will respond to CMV, which to HIV, and so on. When a naïve T cell meets its antigen, it proliferates in response, thus diluting the TREC over multiple rounds of cell division. Thus, counting TREC appears to be a surrogate marker for recent thymic emigration (RTE). TREC are found preferentially in naïve (CD45RA+45RO-) T cells.

First Douek, Koup and colleagues counted the number of TREC in the T cells of people from birth to 73 years old. While naïve (CD45RA+) T cells drop fourfold over the course of a lifetime (mostly between birth and adulthood), the number of TREC decreases continually over the lifespan, ultimately by over tenfold. Three groups of people had unusually low levels of TREC: those who'd had their thymus removed (thymectomized), those over the age of 78, and those with HIV but not on treatment. The low levels were particularly marked in the CD4 cells. Thus, in its effect on recent thymic emigration (RTE), untreated HIV infection mimics the outcome of total thymectomy. A likely explanation is that the decrease in TREC is thymic in origin, and results from HIV-induced inhibition of thymic function, inhibition of thymic precursors, or infection and death of thymocytes that express CD4 (Douek 1998).

The Dallas group followed ten previously antiretroviral-naïve HIV-infected individuals for up to nine months after initiating HAART and measured viral load, naïve CD4 cell count and naïve CD4 cell TREC number. The TREC count rebounded more rapidly than the overall naïve CD4 cell count. All the subjects (aged between 22 and 36) except the one aged 63 experienced increases in naïve CD4 cell TREC, even in cases where the naïve CD4 cell number itself did not rise. Two patients whose viral load rebounded had rapid declines in TREC. The TREC increase seen after HAART could only occur if 1) the thymus is active in adults or if 2) TCR maturation happens extrathymically (Douek 1998).

David Ho presented additional new data on recent thymic emigrants (RTEs) and T cell receptor rearrangement excision circles (TREC) at Keystone in January 1999. They applied a new technique to a larger dataset than Richard Koup used in his December 1998 Nature paper. The Diamond Center's Sharon Lewin developed a new assay to quantitate TREC, which also happens to be useful for counting any form of cellular DNA. When the DNA sequence being sought binds to a specific primer, a molecular beacon lights up. Thus, individual TREC can be counted.

To assess the relationship between thymic output and age, they carried out a population-based study of 542 people aged zero to 95, including 125 HIV-infected adults and 42 children with HIV. They also looked at TREC in 41 people whose viral load had been stably undetectable on HAART for up to 900 days. There was no relationship between TREC number and overall T lymphocyte count, and only a weak correlation between the number of TREC and CD4 cells or naïve CD45RA+ cells. TREC count is stable for the first fifteen years of life, dropping fast through age 25, declining slowly thereafter.

Koup had studied ten people with HIV, of whom eight (80%) had very low TREC counts. By contrast, Ho, Lewin and colleagues studied larger numbers and found a more complex picture. They looked at TREC count in both primary HIV infection (PHI) and in chronic infection.

At ADARC, among 16 seroconverters with PHI, half had normal TREC counts and half had TREC counts below 10% of the range of normal in uninfected persons. Among people with chronic HIV infection, about 60% had TREC counts within the normal range of variation and 40% had TREC counts below 10% of the normal range. The differences were more dramatic among HIV-infected children. They studied 42 KWAs, along with 124 uninfected age-matched controls. Here 50% had TREC counts below 10% of the normal range.

Why do some HIV-infected people have normal TREC levels, and others greatly reduced ones, in the absence of therapy? This implies that only some HIV-infected people have reductions in thymic output. "We don't know," said Ho. "We're going for the gold and looking at viral phenotype. We know thymic tissue is rich in CXCR4 [the co-receptor for syncytium-inducing, SI viruses], so we're looking to see whether people with low TREC counts have SI viruses. We know from studies of thymic implants, for example by Mike McCune in the SCID-hu mouse, that they do worse with SI viruses. We might know by later this week."

They looked at whether HAART raises TREC counts by taking 41 HIV infected people, 19 with API and 22 chronically infected, whose viral load had been stably suppressed for up to 900 days on HAART. They stratified by baseline TREC number (see table below). After HAART, TREC counts decreased slightly in the group with API, increased slowly in half those with chronic infection and increased most dramatically in the group with low baseline TREC.

But Do They Work? Are the New Naïve CD4 Cells Clinically Relevant?

Finally, the most important question to any person with HIV and to any treating physician: what is the use of all these high-tech assays and arcane immunologic arguments unless the new CD4 T cells produced after people go on HAART are clinically useful? If it is not good enough to observe the decline in rates of opportunistic infections and death which has been observed since the introduction of HAART in 1996, one can refer to several papers demonstrating the return of antigen-specific CD4 cells after initiation of HAART. Another paper co-authored by Mike McCune, this time with Krishna Komanduri, Mark Jacobson and others from the SFGH/UCSF powerhouse demonstrated the return of cytomegalovirus-specific CD4 T cells among twelve patients with quiescent CMV retinitis who responded to HAART and ganciclovir. Their CMV-specific T cell response was similar to 18 CMV-infected individuals with no evidence of CMV disease, and much higher than in six individuals with active CMV retinitis or encephalopathy. Thus, it might make sense for HAART/ganciclovir responders to consider going off their ganciclovir therapy, as several groups have already done, without observing a CMV recurrence (Torriani 1997; Tural 1997; Komanduri 1998).

However, as San Diego clinician F.J. Torriani revealed at the Sixth Retrovirus Conference in Chicago during February, 1999, among fifteen patients whose CD4 count returned to over 70 cells/mm3, and who stopped CMV maintenance treatment for up to 25 months, five (33%) failed HAART and developed reactivated CMV retinitis. Four of these five (80%) had low lymphoproliferative responses (LPR) to CMV antigens. These results suggest that 1) the degree of immune reconstitution depends on magnitude of CD4 cell increase (and duration of virologic response to therapy?) and on restoration (or lack thereof) of antigen-specific immunity (Torriani 1999).

In Chicago, Constance Benson reviewed the results of eight studies in which patients whose CD4 cells returned above 200 cells/mm3 stopped PCP prophylaxis. Their risk of PCP seemed to have fallen to the level of those whose CD4 count never reached below 200 in the first place (Benson 1999). A large Spanish study randomized 332 HAART responders to maintain or discontinue PCP prophylaxis; after six months of follow-up, there were zero cases of PCP in both arms (Lopez 1999).

Results such as this have led opportunistic infection maven Bill Powderly of Washington University in St. Louis, Missouri, the "Show-Me" state, to conclude that "although incomplete, considerable immune recovery occurs [post-HAART], sufficient, in most cases, to provide adequate protection against most AIDS-associated opportunistic infections" (Powderly 1998).

Whatever the outcome of the tap vs. drain debates, then, and whatever the ultimate answers about T cell kinetics, dynamics and homeostasis, new discoveries pioneered by HIV immunologists are likely to continue producing significant clinical benefits for people living with HIV and with other infectious, chronic and autoimmune diseases.

T-Cell Kinetics in HIV-Negative and HIV-Infected Individuals

Mean HIV copies/mlN/A94,000<500
Mean lymphocytes/µL 
Absolute production rate 
Replacement rate (k)
CD487 days24 days14 days
CD877 days22 days14 days

*All previously antiretroviral naïve, they received 12 weeks of ritonavir, saquinavir and at least one nucleoside analogue.

Source: Nature Med. 5, 83-89 (1999)


Differences Between Hellerstein and Ho T-Cell Labeling Experiments

Model systemHumansRhesus macaques
Model virusHIVSIV
HAART usageYesNo
Label usedH2BrdU
Duration of labeling2 days3 weeks
Number of datapoints29
Timing of measurement
(days post-labeling)
5-7, 10-140, 7, 14, 21
Bone-marrow/lymph biopsiesNoWeek 3
Datapoints during delabelingNoWeeks 4, 5, 6, 7, 10
Estimated lymphocyte half-life (days)
Infected, on HAART1414NDND

*Half-life lower when viral load was higher.
ND = not done; NG = not given.

Source: Nature Med. 5, 83-89 (1999) and Science 279, 1223-1227 (1998).


Thymic Immigration

Baseline TREC count/µLHAART Induced Change in TREC* per 106 PBMC per day
*TREC= T-cell rearrangement excision circles, a high-tech measure of thymic output of new T cells, in units per one million peripheral blood mononuclear cells per day; see accompanying report

Source: Lewin S, Ho D et al.