Lymph nodes as barriers to T‐cell rejuvenation in aging mice and nonhuman primates

Abstract In youth, thymic involution curtails production of new naïve T cells, placing the onus of T‐cell maintenance upon secondary lymphoid organs (SLO). This peripheral maintenance preserves the size of the T‐cell pool for much of the lifespan, but wanes in the last third of life, leading to a dearth of naïve T cells in blood and SLO, and contributing to suboptimal immune defense. Both keratinocyte growth factor (KGF) and sex steroid ablation (SSA) have been shown to transiently increase the size and cellularity of the old thymus. It is less clear whether this increase can improve protection of old animals from infectious challenge. Here, we directly measured the extent to which thymic rejuvenation benefits the peripheral T‐cell compartment of old mice and nonhuman primates. Following treatment of old animals with either KGF or SSA, we observed robust rejuvenation of thymic size and cellularity, and, in a reporter mouse model, an increase in recent thymic emigrants (RTE) in the blood. However, few RTE were found in the spleen and even fewer in the lymph nodes, and SSA‐treated mice showed no improvement in immune defense against West Nile virus. In parallel, we found increased disorganization and fibrosis in old LN of both mice and nonhuman primates. These results suggest that SLO defects with aging can negate the effects of successful thymic rejuvenation in immune defense.

mice and nonhuman primates. These results suggest that SLO defects with aging can negate the effects of successful thymic rejuvenation in immune defense.

| INTRODUCTION
The thymus undergoes age-related involution, that includes progressive loss of thymic epithelial and hematopoietic lineage cellularity, an increase in adiposity, and reduced T-cell output (Hale, Boursalian, Turk, & Fink, 2006;Nikolich-Žugich, 2014). In the periphery, fewer naïve T cells are available (Appay & Sauce, 2014), and the old T-cell compartment is less able to respond to infections and cancer. This is believed to contribute to increased vulnerability of older adults to emerging and reemerging infections (Nikolich-Žugich, 2014). More recent evidence suggests that secondary lymphoid organ (SLO) organization and structure also undergo changes with increased age (Aw et al., 2016;Becklund et al., 2016;Davies, Thompson, Pulko, Padilla Torres, & Nikolich-Žugich, 2017;Thompson, Smithey, Surh, & Nikolich-Žugich, 2017), and the impact of these changes upon naïve T-cell survival (Link et al., 2007) and function is beginning to be understood.
A "holy grail" of T-cell aging research is to achieve functional rejuvenation of T-cell function (Nikolich-Žugich, 2018). Early experiments with surgical castration have shown that transient thymic rejuvenation is possible, as measured by increased thymic volume and cellularity (Fitzpatrick, Kendall, Wheeler, Adcock, & Greenstein, 1985). Similar results have since been obtained using pharmacological sex steroid blockade as well as injection of growth factors Min et al., 2007;Velardi et al., 2014). While some of these studies have shown some improvement in peripheral immune function in treated mice (Heng et al., 2012;Min et al., 2007), the ultimate tests of functional immunity in the face of microbial challenge were not performed. Therefore, the question remains how well thymic rejuvenation improves the peripheral T-cell pool with aging, and whether it confers improved protection against infection.
To address this question, we examined the effects of (a) keratinocyte growth factor (KGF) administration in mice and nonhuman primates, or (b) sex steroid ablation (SSA) in mice using an antagonist of the luteinizing hormone-releasing hormone receptor, degarelix (Firmagon). Despite robust thymic rejuvenation in response to both interventions, we found no evidence of improved peripheral T-cell maintenance. KGF-treated old mice were not more effective at mounting CD8 T-cell responses to, or clearance of, Listeria monocytogenes. Similarly, degarelix did not improve CD8 T-cell responses to, or survival of old mice following challenge with, West Nile virus (WNV). While rejuvenated thymi produced substantial numbers of recent thymic emigrants (RTE), these RTE did not significantly contribute to T-cell populations in the SLO of old mice compared to adults. We further found that old lymph nodes exhibited considerable fibrosis and degeneration of structure. These data indicate that restoration of thymic function by itself may not be sufficient to improve the immune response in elderly and suggest that interventions to simultaneously alleviate defects in aging SLO may need to be considered when designing strategies to improve immune response in older organisms.

| Age-related decline in naïve T cells
Early in life, thymic involution progressively limits the production of new naïve T cells (Nikolich-Žugich, 2014). Thereafter, the naïve T-cell pool is successfully maintained peripherally until the last tertile or quartile of life (den Braber et al., 2012), at which point this process also eventually deteriorates. This leads to a loss of naïve T cells that is concordant with an increase in susceptibility to infection with advanced age (Heng et al., 2012;Nikolich-Žugich, 2014). Figure 1a shows the cross-sectional kinetics of naïve CD8 and CD4 T cells (CD62L HI , CD44 LO ) decline in our mouse colony, measured as a fraction of total CD8 and CD4 cells, respectively. We have previously shown a similar loss in the absolute number of naïve CD8 T cells with age in the mouse spleen (Smithey, Li, Venturi, Davenport, & Nikolich-Žugich, 2012). The decline in representation and/or numbers of naïve CD8 T cells in the blood had also been described by several groups in nonhuman primates (Janković, Messaoudi, & Nikolich-Žugich, 2003;Okoye et al., 2015;Pitcher et al., 2002) and humans (Fagnoni et al., 2000;Olsson et al., 2001;Wertheimer et al., 2014).

CD4 T-cell numbers in blood
We next examined whether increased thymic size and cellularity contributed to the peripheral naïve T-cell pool in the blood and found that the increase in thymus cellularity failed to increase the percentage of naïve CD8 (Figure 2a numbers of CD3 T cells in blood tend to decrease with age (e.g., # CD3/ml blood 1.99 ± 0.43 × 10 6 vs. 1.03 ± 0.27 × 10 6 ; p = 0.0159), these results are even more pronounced in absolute terms. Similarly, old rhesus macaques (RM) treated with KGF showed no statistical improvement in the percentage of naïve CD8 or CD4 T cells (Fig-ure 2e,f) (defined, as previously described (Okoye et al., 2015), as CD28 INT CD95 LO ) in blood. We could not directly measure thymic cellularity in RM treated with KGF as that would have been a terminal study, and our attempts to measure thymic size by MRI were not satisfactory. However, we did observe comparable clinical signs of KGF activity in old and adult RM, as manifested by transient redness/flushing of the face and lips and increased salivation, which suggested that the administered KGF had the expected impact on epithelia. Moreover, the same dose (in mg/kg) produced a manifest increase in murine thymus cellularity. We conclude that thymic rejuvenation via SSA or KGF administration did not correlate to an increased frequency of naïve peripheral T cells in SLO.

| SSA treatment of old mice did not improve survival after WNV challenge
While in our hands the increased thymic size and cellularity did not translate into an overall increase in naïve T-cell frequency in the blood, it was possible that the newly produced "young" cells  (Hale et al., 2006). Such replacement of "old" naïve T cells could lead to improved immunity and was invoked by Haynes, Swain, and colleagues to explain improved T-cell function following depletion of peripheral T cells in old mice (Haynes, Eaton, Burns, Randall, & Swain, 2005). To test whether the T-cell pool was functionally improved by SSA, we challenged old degarelix-treated and control mice with WNV 42 days after SSA. WNV is an age-sensitive virus that induces significantly higher mortality rates in older humans (Petersen & Marfin, 2002) and mice (Brien, Uhrlaub, Hirsch, Wiley, & Nikolich-Žugich, 2009) and is ideal to test possible improvements in protective immunity of older organisms. We found that the survival of old mice treated with degarelix was no better, and tended to be worse, compared to untreated controls ( Figure 3a). However, as the defense against WNV is mediated by multiple arms of immunity ( Figure S1).
Prior work with KGF treatment of 14-month-old mice found improved anti-KLH antibody responses and took it as evidence that CD4 function must have been improved (Min et al., 2007). To assess that possibility, we examined anti-WNV antibody responses. Adult mice exhibited significantly higher overall anti-WNV IgG than old mice, and degarelix again provided no advantage over old controls in that regard ( Figure 3c). Together, these results indicate that thymic rejuvenation using degarelix or KGF was insufficient to improve immunity against intracellular infections in old mice.

| Increased thymic output does not increase naïve T cells in secondary lymphoid organs
To examine the mechanistic basis of the failure of the rejuvenated thymus to improve functional immunity, we took advantage of For RM experiments, n = 3 adults, n = 7 old, and n = 7 treated old from three independent experiments. For degarelix experiments, n = 7 to 13 mice per group pooled from two independent experiments. Bar graphs means + SEM are shown (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Absolute numbers provided in the text Rag2pGFP transgenic mice, in which the RTE are labeled with GFP, and where this label persists in newly exported cells until diluted by cell division and/or until the cell is removed from the population (Boursalian, Golob, Soper, Cooper, & Fink, 2004;Hale et al., 2006).
The Rag2pGFP reporter marks RTE for approximately 3 weeks following thymic export (Boursalian et al., 2004) and is therefore a robust marker of RTE production and initial distribution. In this model, we found a 2.12-fold increase in the percentage of CD3-expressing RTE in the blood in old mice on day 42 following degarelix Changes in the stromal microenvironment in secondary lymphoid organs (SLO), particularly lymph nodes (LN), have recently been reported to influence naïve T-cell homeostasis (Becklund et al., 2016;Link et al., 2007). We found that old LN exhibit a profound decline in Tcell cellularity while the total T-cell numbers in the spleen were not significantly altered in old mice (

| Increased thymic output contributes differently in adult and old mice to the peripheral Tcell pool in Slo
Old mice exhibit defects in the stromal architecture of SLO (Aw et al., 2016;Becklund et al., 2016;Davies et al., 2017;Masters, Haynes, Su, & Palmer, 2016;Thompson et al., 2017) and that could play a role in the suboptimal ability of old LN to recruit and properly direct T-cell trafficking during an immune response (Richner et al., 2015). Less is known about whether such defects could also affect ingress of RTE into old LN. To test whether the kinetics of thymic and peripheral (SLO) reconstitution may be different between adult and old mice, we treated old and adult mice with degarelix and measured numbers of double-positive (DP) thymocytes (as a measure of thymic generative activity (Hale et al., 2006)) against the number of naïve CD3 T cells (CD3 + CD62L HI C- This raised the possibility that bioavailability/access to IL-7 may be the key mechanism behind the above inability of old LN to recruit and/or retain RTE. Indeed, in response to TGFβ and/or Th2 cytokines (particularly IL-13), FRC has the potential to produce excessive collagen, and to lead to the process of fibrosis, that results in a thick- found an improvement in thymocyte number following either KGF or SSA/degarelix treatment of old mice (Heng et al., 2012;Velardi et al., 2014) (Figure 1). However, while the thymi of old mice treated with degarelix or KGF were rejuvenated, we found no increase in naïve CD8 and CD4 T-cell frequencies in the blood in mice or monkeys ( Figure 2). We also found that thymic rejuvenation resulted in a surge of RTE in the blood, but that it did not significantly improve the number of naïve T cells or RTE in spleen or LN in old mice (Figures 4, 5). Even when RTE numbers were trending higher (but were not significantly improved), these numbers were too low (only 2%-10% of all total naïve T cells) to contribute to an increase in the naïve T-cell pool or overall LN cellularity. This further suggested that LN were not capable of either recruiting and/or retaining new RTE produced by the thymus, an issue discussed further below.
It has been shown that a fraction of RTE can home directly to the gut and produce intraepithelial lymphocytes (Staton et al., 2006).  . Fibrosis in mouse LNs was quantified using ImageJ software and presented as (f) capsule thickness and (g) percentage of area positive for collagen inside the capsule. Mean + SEM are shown (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001), n = 4-6 mouse/group) We did not examine this migration, as our aim was to deal with immunity and homeostasis in SLO, and therefore, aging of mucosal lymphoid tissues will have to be addressed at a later date.
KGF is secreted by mesenchymal stem cells and is known to promote the proliferation of epithelial cells including thymic epithelial cells (Chaudhry, Velardi, Dudakov, & Brink, 2016;Nikolich-Žugich, 2007). KGF is currently licensed for the reduction in oral mucositis after bone marrow transplantation (Seggewiss et al., 2007). KGF was shown to increase thymopoiesis and total T-cell numbers in the spleen of aged mice (18 months old) due to an expansion of both naïve and memory subsets (Chaudhry et al., 2016). However, T-cell proliferation in mixed lymphocyte reactions or following concanavalin A stimulation was not improved by KGF (Alpdogan et al., 2006). Min et al. treated 14-month-old mice with KGF and observed increased thymopoiesis and increased total CD4 and CD8 T cells in the blood 1 month after treatment, which correlated to improved secondary T-dependent antibody responses to keyhole limpet hemocyanin (KLH) given in adjuvant (Min et al., 2007). There are several differences between these experiments and the ones described here that may explain the discrepancy of the results. First, we performed our experiments on bona fide old animals (18-20 months at treatment), as opposed to the "almost" old 14-month-old mice (Min et al., 2007). Second, it is possible that secondary KLH + adjuvant stimulation is a more potent stimulus compared to live WNV in our experiments; third, the experiments may not be directly comparable due to different rejuvenation reagents. Finally, there was a difference in the immunization route (s.c. for WNV following degarelix, and i.v. for Listeria following KGF in our hands; vs. i.p. for KLH + IFA in the experiments of Min et al. [2007]). At a minimum, protective immunity against Listeria, that is mediated by CD8 T cells, was not improved by any parameter tested (CD8 tetramer + cell numbers or CD8+ Tcell polyfunctionality).
Sex steroid ablation has also been extensively studied for its ability to rejuvenate the aged thymus and the naïve T-cell compartment (Chaudhry et al., 2016;Heng et al., 2012). SSA via surgical castration (Heng et al., 2012) or degarelix (Shore, 2012) can restore both thymic cellularity and the ratios of naïve to memory peripheral T cells in 9-month-old mice. However, surgical castration is progressively less effective in restoring peripheral naïve:memory T-cell ratios in 18month-old and 24-month-old mice , and degarelix has not been studied in that regard so far. More importantly, surgical SSA in 9-month-old mice led to improved virus-specific CD8 T-cell percentages and numbers following influenza A viral infection (Heng et al., 2012). That effect was somewhat reduced at 18 months and abrogated in 24-month-old mice (Heng et al., 2012). In our hands, degarelix improved thymic cellularity and the numbers of RTE in the peripheral blood of old (>18, typically 20 months) mice, but did not improve either total naïve or RTE numbers in their SLO. Further, degarelix treatment did not improve survival, or T-or B-cell responses against WNV in old mice. Our functional data are reminiscent to data of Heng et al. (2012), in that they both highlight the existence of age limits to the ability of SSA to improve T-and B-cell responses in truly old mice.
One key difference between our and the above studies is that we conclusively tracked RTE export and migration by rejuvenated thymi into the blood, spleen, and LN. Our finding of the reduced presence of RTE in SLO of old mice raises important issues about the exact mechanism of the age-related defect in this case. Such a defect(s) could be intrinsic to newly produced T cells, or intrinsic to LN stroma, or both, and experiments are in progress to address this issue. Regardless, this defect results in impaired homing and/or retention in SLO of the aged animals.
While there is much to be learned about long-term maintenance of stromal and lymphoid compartments in SLO with aging, we know that RTE need signals from SLO to survive and mature properly (Houston, Nechanitzky, & Fink, 2008;Link et al., 2007). Our data are consistent with recent work documenting a decline in FRC in old LN (Becklund et al., 2016) and showing that fewer transferred naïve T cells can be recovered from the old compared to adult SLO post-WNV infection (Richner et al., 2015). Of interest, this study and a prior study of one of us (Becklund et al., 2016) have demonstrated that IL-7 does not decline with age at the mRNA (Becklund et al., 2016) and protein (this study) levels, even though the FRC, a critical component of LN stroma, that produce IL-7, is reduced with aging (Becklund et al., 2016;Davies et al., 2017). Together with our results demonstrating increased LN fibrosis with age, and our data, to be reported separately, on the increase in profibrotic cytokines in old LN, this suggests that IL-7 bioavailability may be suboptimal with aging, making LN fibrosis an interesting potential target for both mechanistic and therapeutic studies.
We conclude that restoration of thymic cellularity and of new Tcell production is not, by itself, sufficient to improve immune protection against intracellular pathogens. Rather, we show that deterioration of SLO in aged animals contribute to the decline of T-cell maintenance and function, and have the potential to negate the beneficial effects of thymic rejuvenation. These defects, therefore, must be taken into account when considering immune rejuvenation in the old age. Thousand Oaks, CA) at 5 mg kg −1 day −1 for 3 consecutive days i.p. (Min et al., 2007). Experiments were conducted under the approval of the Institutional Animal Care and Use Committee (IACUC) and the Institutional Biosafety Committee, in accordance with all applicable federal, state, and local regulations. All WNV experiments were completed within a USDA-inspected biosafety level 3 facility.

| Animals, KGF, and degarelix treatments
Colony-bred rhesus macaques (Macaca mulatta, RM, Indian origin) of both sexes (two females and one male per group) were maintained according to federal, state, and local guidelines at the Oregon National Primate Research Center under the approval of the Center's IACUC. Old (20-30 years of age) and adult (9-15 years of age) RM were given a low-dose (250 μg/kg KGF) (Seggewiss et al., 2007) or high-dose treatments of 1,000 μg/kg or 5,000 μg/kg KGF for 3 consecutive days subcutaneously.

| LN analysis for the presence of fibrosis
LN were immediately fixed in 4% paraformaldehyde/PBS (mouse) or 10% neutral-buffered formalin (RM) at 4°C overnight, then routinely processed for paraffin embedding, and cut to obtain 5-μm-thick sections. Before use, sections were deparaffinized, rehydrated, and incu-
One-way and two-way ANOVA were used to compare groups. Some of the data were not normally distributed as determined by Shapiro-Wilk normality test. In those cases, we ran the Kruskal-Wallis test to confirm significance and it was maintained. Adjusted p-values of <0.05 were considered significant. Nonlinear fit analysis was used to compare whether best-fit values were shared between data sets. Significance is noted as follows throughout: ns = not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Error bars denote SEM.

ACKNOWLEDG MENTS
The authors wish to thank Dr. Michael K. Axthelm, Alfred Legasse, and Shannon Planer from the ONPRC/VGTI for expert help with KGF treatment of RM; and the members of the Nikolich, Kuhns, Schenten, Wu, and Frelinger laboratories for constructive input and supported by the USPHS awards from the NIH/NIAID U01 AI82529 (L.J.P. and J.N.Ž., co-PI) and NIH/NIA P01 052359 (J.N.-Ž.,