You are on page 1of 30

Article

The Bone Marrow Protects and Optimizes


Immunological Memory during Dietary Restriction
Graphical Abstract Authors
Nicholas Collins, Seong-Ji Han,
Michel Enamorado, ...,
Dorian B. McGavern,
Pamela L. Schwartzberg,
Yasmine Belkaid

Correspondence
ybelkaid@niaid.nih.gov

In Brief
Calorie restriction triggers memory T cell
homing to the bone marrow to promote
survival and enhanced protective
function.

Highlights
d Dietary restriction promotes memory T cell accumulation
in BM

d BM trophic factors and adipocytes promote memory T cell


accumulation in BM

d Memory T cells display enhanced protective function during


dietary restriction

Collins et al., 2019, Cell 178, 1088–1101


August 22, 2019 ª 2019 Elsevier Inc.
https://doi.org/10.1016/j.cell.2019.07.049
Article

The Bone Marrow Protects and Optimizes


Immunological Memory during Dietary Restriction
Nicholas Collins,1 Seong-Ji Han,1 Michel Enamorado,1 Verena M. Link,1 Bonnie Huang,2 E. Ashley Moseman,3
Rigel J. Kishton,4 John P. Shannon,5 Dhaval Dixit,6 Susan R. Schwab,6 Heather D. Hickman,5 Nicholas P. Restifo,4
Dorian B. McGavern,3 Pamela L. Schwartzberg,2 and Yasmine Belkaid1,7,*
1Metaorganism Immunity Section, Laboratory of Immune System Biology, National Institute of Allergy and Infectious Diseases, National

Institutes of Health, Bethesda, MD 20892, USA


2Cell Signaling Section, Laboratory of Immune System Biology, National Institutes of Health, Bethesda, MD 20892, USA
3Viral Immunology and Intravital Imaging Section, National Institute of Neurological Disorders and Stroke, National Institutes of Health,

Bethesda, MD 20892, USA


4Surgery Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD 20814, USA
5Viral Immunity and Pathogenesis Unit, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD

20892, USA
6Skirball Institute of Biomolecular Medicine, New York University School of Medicine, New York, NY 10016, USA
7Lead Contact

*Correspondence: ybelkaid@niaid.nih.gov
https://doi.org/10.1016/j.cell.2019.07.049

SUMMARY et al., 2016; Sallusto et al., 1999) that are required for body-
wide immunosurveillance, whereas tissue resident memory cells
Mammals evolved in the face of fluctuating food avail- (TRM) are essential for initiating and amplifying local responses
ability. How the immune system adapts to transient (Jameson and Masopust, 2018; Mueller and Mackay, 2016). At
nutritional stress remains poorly understood. Here, steady state, memory T cell homeostasis is under the control
we show that memory T cells collapsed in secondary of various cytokines, transcription factors, and metabolic fuels
lymphoid organs in the context of dietary restriction (Buck et al., 2016; Cui et al., 2015; Kaech and Cui, 2012; Pan
et al., 2017; Surh and Sprent, 2008). However, these long-lived
(DR) but dramatically accumulated within the bone
cells are faced with numerous challenges throughout the life of
marrow (BM), where they adopted a state associated
the host, including their persistence and maintenance of protec-
with energy conservation. This response was coordi- tive function during stress and reduced nutritional availability.
nated by glucocorticoids and associated with a Indeed, food accessibility was and can remain highly contingent
profound remodeling of the BM compartment, which on encounters with distinct environments and climatic condi-
included an increase in T cell homing factors, tions. Thus, mechanisms may have evolved to ensure that the
erythropoiesis, and adipogenesis. Adipocytes, as host can adapt and thrive in situations where calories and nutri-
well as CXCR4-CXCL12 and S1P-S1P1R interactions, ents are limited. Of interest, caloric restriction or dietary restric-
contributed to enhanced T cell accumulation in BM tion (DR) has been shown to promote various aspects of host
during DR. Memory T cell homing to BM during DR fitness, including the improvement of metabolic profiles, preven-
was associated with enhanced protection against in- tion of cellular aging, and reduced incidence of cancer (Nikolich-
Zugich and Messaoudi, 2005; Redman et al., 2018; Robertson
fections and tumors. Together, this work uncovers a
and Mitchell, 2013; Speakman and Mitchell, 2011). However,
fundamental host strategy to sustain and optimize the consequence of DR on the memory T cell compartment re-
immunological memory during nutritional challenges mains to be addressed.
that involved a temporal and spatial reorganization of Due to the importance of memory T cells for host survival,
the memory pool within ‘‘safe haven’’ compartments. defined strategies or compensatory mechanisms may be in place
to sustain these cells in the context of nutritional challenges. Of
INTRODUCTION relevance, we and others have found that white adipose tissue
(WAT) is a reservoir for memory T cells (Han et al., 2017; Masopust
Host survival depends on the ability to adapt to challenges in a et al., 2001). While WAT is reduced during DR, the bone marrow
way that sustains and protects fundamental physiological pro- (BM) paradoxically shows increased adipogenesis in this context
cesses. Immunological memory is a cardinal feature of the adap- (Cawthorn et al., 2014; Devlin et al., 2010). These observations
tive immune system, which confers a survival advantage by raised the possibility that an alliance between defined tissue com-
allowing the host to rapidly and effectively control subsequent partments may serve the purpose of preserving immunological
challenges. Such responses rely on the ability of memory memory in the face of nutritional challenges.
T cells to persist long term, which can be divided into circulating Here, we show that DR induces a whole-body response, re-
and resident subsets. Circulating cells include central, effector, sulting in the collapse of circulating memory T cell populations
and peripheral memory T cells (TCM, TEM, and TPM) (Gerlach in secondary lymphoid organs (SLOs) and blood but enhanced

1088 Cell 178, 1088–1101, August 22, 2019 ª 2019 Elsevier Inc.
A

B C

F G
E

H I

Figure 1. Memory T Cells Accumulate in BM during DR


(A) Number of CD8+ CD44+ T cells in spleen (spl), cervical lymph node (cLN), blood, and gonadal adipose tissue (GAT) over time during 50% DR.
(B) Number of CD8+ CD44+ T cells in femur BM of mice on DR over time.
(C) Confocal microscopy of CD4+ (magenta) and CD8+ (yellow) T cells in BM after 3 weeks of DR.
(D) Number of CD8+ CD44+ T cells in BM from tibia, skull, thoracic vertebrate, humerus, and ilium of mice on DR for 3–6 weeks.
(E) Mice were infected and 4 weeks later put on DR for 4–6 weeks.
(F and G) (F) Number of memory CD8+ T cells in GAT, spl, and BM specific for the YopE antigen of Yersinia pseudotuberculolsis or (G) the nucleoprotein antigen of
influenza A virus after 4–6 weeks of DR.
(legend continued on next page)

Cell 178, 1088–1101, August 22, 2019 1089


accumulation in BM. Such a response was associated with pro- that when given at 50% restriction would contain normalized
found remodeling of the BM compartment, with increases in ad- levels of vitamins and minerals (V&M), essential amino acids
ipocytes and T cell trophic factors. The ability of memory T cells (EAA), or total protein. Normalizing V&M, EAA, or total protein still
to accumulate in BM not only protected the memory pool from induced memory T cell redistribution to BM, while reducing cal-
inhospitable conditions during DR, but also optimized their func- ories alone was sufficient to drive the response (Figure 1H).
tion in the face of secondary challenges. Altogether, this work Further, accumulation of memory T cells in BM during DR was
uncovers a fundamental host strategy to adapt to physiological reversible, with the steady-state number rapidly restored in BM
nutritional challenges, which are associated with a temporal and spleen upon refeeding (Figure 1I). Collectively, these results
and spatial reorganization of the memory pool within ‘‘safe indicate that memory T cells collapse in SLO and blood but
haven’’ tissue compartments. rapidly and reversibly accumulate in the BM compartment in
response to a reduction in calories.
RESULTS
Circulating Memory T Cells Accumulate in Bone Marrow
Memory T Cells Accumulate in the Bone Marrow during but Maintain the Ability to Recirculate during Dietary
Dietary Restriction Restriction
To assess the fate of memory T cells in the context of a transient Several memory T cell populations exist and are characterized
reduction in nutrition, mice were placed on DR, which involved by distinct migratory patterns and functional potential. DR
receiving 50% of their daily food intake. This resulted in approx- induced a decrease in the memory T cell subsets found in SLO
imately 10%–15% weight loss (Figure S1A) and a reduction in fat and WAT (Figure 2A). In contrast, the number of TCM and TEM
mass (Figure S1B) after 1 week, followed by a plateau (Fig- was significantly increased in BM during DR, whereas T cells ex-
ures S1A and S1B). DR caused a decrease in SLO cellularity (Fig- pressing a TRM or TPM phenotype were sustained (Figures 2B
ure S1C), resulting in a decrease in number of antigen-experi- and S2A). The expression of the canonical memory markers
enced CD8+ and CD4+ T cells (Figures 1A, S1D, and S1E), as CD127 (interleukin [IL]-7Ra), CD122 (IL-2Rb/IL-15Rb), CD25
well as regulatory T cells (Treg), natural killer (NK) cells, and (IL-2Ra), and CD62L (L-selectin), and transcription factors
mature B cells (Figures S1D and S1F–S1H). A similar decrease T-BET and EOMES, was similar between TCM and TEM from
was observed in blood and WAT (Figures 1A and S1E–S1H). mice fed ad libitum or on DR (Figures S2B–S2G). Thus, DR pref-
Thus, DR is associated with a rapid and profound collapse of an- erentially promoted the accumulation of TCM and TEM in the BM.
tigen-experienced T cells in the periphery, raising the possibility CD8+ T cells (isolated ex vivo or pre-activated in vitro) intrave-
that memory T cells may redistribute to a distinct niche under nously transferred into recipient mice on DR were found at
these conditions. reduced numbers in the spleen but were increased in BM
In contrast to other compartments examined, antigen-experi- compared to mice fed ad libitum (Figures 2C and S2H), support-
enced CD8+ and CD4+ T cells were significantly increased in BM ing the idea that BM accumulation was the result of cellular redis-
during DR (Figures 1B, 1C, and S1E). Lymphocyte redistribution tribution. In line with this, memory T cells in BM during DR did not
occurred by 1 week following the initiation of DR and was stable show increased levels of homeostatic proliferation (Figure S2I).
for at least 6 weeks (Figure 1B). Accumulation of antigen-experi- In fact, the rate of proliferation during DR was significantly lower
enced CD8+ T cells occurred across the entire BM compartment, as compared to mice fed ad libitum (Figure S2I). On the other
with an increase of T cells observed in BM from the femur, tibia, hand, memory T cells vigorously proliferated when mice on DR
skull, vertebrate, humerus, and ilium during DR (Figures 1B–1D). were re-fed ad libitum, which was particularly pronounced in
Such an increase was selective to memory T cells, as the number the TCM subset (Figure S2I). As such, TCM may be critical to
of Treg, NK cells, mature B cells, and plasma cells in BM was pre- restore the memory T cell compartment following the restoration
served during DR but not increased (Figures S1F–S1I). of calories.
To determine whether memory T cells induced following The major function of circulating memory T cells is immunosur-
infection redistributed during DR, we tracked antigen-specific veillance. To assess whether circulating memory T cells perma-
responses following an oral infection with Yersinia pseudo- nently accumulated in BM during DR or if they retained the ca-
tuberculosis DyopM (Yptb DyopM) (Han et al., 2017) or an intra- pacity to migrate, we performed parabiotic surgery on mice
nasal infection with influenza A virus (A/PR/8/34). Following path- that had been on DR for 1 week prior to surgery and for
ogen clearance and the establishment of memory (4 weeks), mice the remainder of the experiment or control mice that were fed
were placed on DR (Figure 1E), showing that these pathogen-spe- ad libitum throughout (Figure 2D). By 3 weeks post surgery,
cific memory CD8+ T cells were reduced in spleen and WAT but circulating memory T cell populations in the spleen and BM
found at higher numbers in BM during DR (Figures 1F and 1G). were in migratory equilibrium, while TRM cells in BM maintained
We next assessed whether a reduction in calories alone was their tissue-residency status (Figure 2E). This was the case
responsible for memory T cell redistribution during DR or if both in parabiotic pairs fed ad libitum and those on DR, indi-
defined nutrients played a role. To this end, we designed diets cating that circulating memory T cells retained the ability to

(H) Number of CD8+ CD44+ T cells in BM and spl of mice fed the indicated diet ad libitum or at 50% restriction for 3 weeks.
(I) Number of CD8+ CD44+ T cells in BM and spl of mice on DR for 3 weeks or on DR for 3 weeks then refed ad libitum for a further 1 or 3 week(s).
Each symbol represents an individual mouse. Data show the mean representative of four (C) or pooled from two to four experiments (A, B, D, and F–I) with two to
five mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; two-tailed unpaired Student’s t test. See also Figure S1.

1090 Cell 178, 1088–1101, August 22, 2019


A

B C

D E

F G

Figure 2. Circulating Memory T Cells Accumulate in BM during DR


(A) Number of CD8+ TCM (CD44+ CD62L+ CD69 ), TEM (CD44+ CD62L CD69 ), and TRM (CD44+ CD62L CD69+) in the indicated tissue of mice on DR for
3–4 weeks.
(B) Plot showing CD62L and CD69 expression by CD8+ CD44+ cells in BM. Number of the indicated subset in BM from mice on DR for 3–4 weeks.
(C) Purified CD8+ CD44+ T cells were transferred into hosts fed ad libitum or on DR. Number of transferred cells in BM and spl after 1 week is shown.
(D) Congenically distinct mice were put on DR for 1 week and joined to form parabiotic pairs then maintained on DR for 3 weeks. Control pairs were fed ad libitum
throughout.
(E) Frequency of host-derived memory T cells in spl and BM from (D).
(F) In vitro activated OT-I mTomato+ T cells (pseudocolored green) were transferred into hosts fed ad libitum or on DR for 3 weeks and skull BM was imaged
1 week later. Images show representative frames. Bottom images show track displacement length from 0–25 or more mm.
(legend continued on next page)

Cell 178, 1088–1101, August 22, 2019 1091


migrate in the context of DR and that the increase in number of to the number of WT cells in the spleen of mice fed ad libitum
T cells within the BM was associated with transient, rather than (Figure 3G). In contrast, the number of Nr3c1 / T cells was
permanent, accumulation. not reduced in the spleen of mice on DR compared to the num-
We next assessed the motility of individual T cells in BM by ber in the spleen of mice fed ad libitum (Figure 3G). On the other
intravital imaging. To this end, we transferred in vitro activated hand, WT and Nr3c1 / T cells were both able to accumulate in
mTomato+ OT-I cells and performed two-photon imaging on the BM to a similar degree during DR (Figure 3G). Together,
skull BM. More T cells were observed within the BM of mice on these results support the idea that a direct interaction between
DR compared to mice fed ad libitum (Figure 2F). The behavior GCs and the GC receptor on T cells promoted their peripheral
of T cells was impacted by DR, evidenced by significantly collapse during DR, while GCs induce T cell accumulation in
reduced speed and track displacement length compared to the BM indirectly.
T cells in BM of mice fed ad libitum (Figures 2F and 2G;
Video S1). Collectively, these results support the idea that circu- Bone Marrow Remodeling during Dietary Restriction
lating memory T cells maintain their ability to migrate in the Promotes Enhanced Recruitment and Retention of
context of DR but display enhanced dwell time within the BM Memory T Cells
compartment. Our results supported the idea that the BM milieu was impacted
in a way that promoted the accumulation and survival of memory
Glucocorticoids Promote the Accumulation of T cells during DR. Whole-tissue RNA sequencing revealed that
Circulating Memory T cells in Bone Marrow during the BM was profoundly remodeled during DR, with 3,981 genes
Dietary Restriction significantly upregulated and 4,173 genes downregulated
The accumulation of memory T cells in BM suggested that this compared to mice fed ad libitum (Figure 4A). The most prominent
niche provided a survival advantage in the context of DR. DR is changes involved the upregulation of genes associated with adi-
associated with responses aimed at restoring energy balance, pogenesis, whereas those involved in B cell development were
a process dominantly coordinated by glucocorticoids (GCs) downregulated (Figure 4A). Of note, cytokines involved in mem-
(Cain and Cidlowski, 2017). However, heightened levels of GC ory T cell homeostasis (IL-7 and IL-15) were not differentially
can promote T cell death (Fujita et al., 2002; Wing et al., 1988). expressed (Figure S4A).
We found that while GCs were elevated in the blood of mice on In regard to T cell homing, several chemokines were differen-
DR compared to those fed ad libitum, the concentration of GC tially expressed in the BM during DR (Figure 4A). This included
in BM was significantly lower both at baseline and during DR Cxcl12, previously shown to promote T cell migration to the
(Figure 3A). In agreement, the frequency of dead (DAPI+) memory BM (Mazo et al., 2005), which was abundantly expressed at
T cells was increased in the spleen of mice early post DR but was baseline and modestly increased during DR (Figures 4A and
lower in BM both at steady state and during DR (Figure 3B). S4B). Of interest, CXCR4, the receptor for CXCL12, was
Circulating memory T cells in the BM also expressed higher increased on the surface of TCM and TEM during DR (Figures
levels of the anti-apoptotic factor BCL-2 during DR (Figures 3C 4B and S4C). To assess a role for this axis in promoting T cell
and S3A), suggesting that this compartment can promote mem- migration to BM during DR, pre-activated WT CD8+ T cells or
ory T cell survival in the context of nutritional stress. those lacking Cxcr4 were adoptively transferred into mice fed
To directly address a role for GC in orchestrating memory ad libitum or on DR. This showed that CD8+ T cells lacking
T cell redistribution during DR, we placed mice lacking adrenal Cxcr4 had a reduced ability to home to the BM compared to
glands (adrenalectomy or ADX) on DR (Fujita et al., 2002; Wing WT cells (Figure 4C). Furthermore, memory T cells were reduced
et al., 1988). Under this setting, memory T cells did not collapse in number in BM during DR following treatment with an inhibitor
in SLO, nor did they accumulate in BM (Figures 3D and S3B). of CXCR4 (AMD3100) (Figure S4D). Thus, enhanced T cell accu-
Furthermore, administration of the synthetic GC dexamethasone mulation within the BM (via recruitment and/or retention) during
(Dex) reproduced the phenotype seen during DR, with a reduc- DR was controlled by a CXCR4-CXCL12 axis.
tion of TCM in the spleen and an increase in BM (Figure 3E). While A striking component of the BM remodeling during DR was an
Dex treatment resulted in increased expression of BCL-2 by both increase in red blood cells (RBCs), which we observed macro-
subsets in BM (Figures 3F and S3C), its impact in terms of redis- scopically and by confocal imaging (Figure 4D). This was
tribution was more pronounced on TCM than TEM (Figures 3E and confirmed by flow cytometry, which revealed an increase in
S3D), suggesting that these cells may have different levels of both precursor (TER-119+ CD71+) and mature (TER-119+
sensitivity to GCs. CD71 ) RBCs (Figure 4E). In agreement with previous work (Fly-
We next transferred CD8+ T cells from wild-type (WT) mice or gare et al., 2011; Zhang et al., 2013), mice fed ad libitum that
those lacking the Nr3c1 gene that encodes the GC receptor were treated with Dex showed enhanced erythropoiesis (Fig-
(Mittelstadt et al., 2012) into mice fed ad libitum or on DR. ure S4E). Mature RBCs are the major source of sphingosine-1-
Approximately half the number of WT cells could be recovered phosphate (S1P) in the blood (Hänel et al., 2007; Pappu et al.,
1 week post transfer in the spleen of mice on DR when compared 2007), a molecule that plays a critical role in controlling T cell

(G) Speed and track displacement length of OT-I cells in skull BM of mice on DR.
Each symbol represents an individual mouse except for (F), in which each symbol represents an individual cell. Data show the mean pooled from two (A, C, and E)
or three experiments (B and G) or representative of three experiments (F). **p < 0.01, ***p < 0.001, ****p < 0.0001; ns, not significant. Two-tailed unpaired Student’s
t test. ND, not detected. See also Figure S2 and Video S1.

1092 Cell 178, 1088–1101, August 22, 2019


A B C Figure 3. GCs Drive Memory T Cells into BM
during DR
(A) Concentration of corticosterone in serum and the
extracellular environment of BM in mice on DR for
3 weeks.
(B) Frequency of DAPI+ cells in spl and BM of mice
on DR for 1 week.
(C) BCL-2 expression by BM TCM in mice fed ad
libitum (blue) or on DR (red) for 3 weeks.
(D) Mice received sham surgery or an adrenalec-
D E tomy (ADX) 1 week prior to being put on DR for
1 week. Graphs show number of TCM in spl and BM.
(E and F) (E) Number of TCM in spl and BM and (F)
BCL-2 expression by these cells in mice fed ad
libitum that were administered veh (blue) or dex (red)
daily for 2 weeks.
(G) WT or Nr3c1fl/fl x Lck-Cre mice (Cre ; WT, Cre+;
Nr3c1 / ) CD8+ CD44+ T cells were transferred into
mice fed ad libitum or on DR for 3 weeks. Cells were
enumerated 1 week post transfer.
G Graphs show transferred cells recovered in spl and
F BM, expressed as the fold-change in DR mice over
the average number found in mice fed ad libitum.
Each symbol represents an individual mouse. Data
show the mean, pooled from 3 (A) or 2 (B–G) ex-
periments. **p < 0.01, ***p < 0.001, ****p < 0.0001,
ns; not significant. Two-tailed unpaired Student’s t
test. See also Figure S3.

adipocyte gene Fabp4 (Figures 4A and


4H). To determine if BM adipocytes
contributed to memory T cell survival or
migration (Cyster and Schwab, 2012). RBCs can also store and accumulation, we generated mice in which adipocytes could
release chemokines and cytokines (Karsten et al., 2018a; Kars- be inducibly deleted (Adipoq-CreERT2 3 Rosa26-DTA). These
ten et al., 2018b), suggesting that increased numbers of RBCs mice were placed on DR for 2 weeks before tamoxifen treatment
may contribute to T cell accumulation in the BM during DR. then maintained on DR for an additional 3 weeks. A reduction in
To determine a role for S1P-S1P1R (S1P receptor 1) in retaining BM adipocytes was confirmed by reduced expression of Fabp4
memory T cells in BM during DR, mice were treated with FTY720, in whole-BM samples (Figure S4G). In mice with reduced adipo-
which prevents migration toward S1P (Matloubian et al., 2004). cytes, TCM were unable to efficiently persist in BM during DR,
This treatment resulted in reduced T cell numbers in BM of both suggesting a direct role for BM adipocytes in promoting memory
control and DR mice (Figures 4F and S4F). As FTY720 promotes T cell retention or survival in this compartment (Figure 4I). Previ-
accumulation of T cells in lymph node (LN) (Mandala et al., ous work suggested that memory T cell survival depends on
2002), we could not exclude that reduced accumulation in BM long-chain fatty acids (LC-FAs) to fuel mitochondrial fatty-acid
was due to impaired egress from LN. To circumvent this, mice oxidation (FAO) (O’Sullivan et al., 2014; Pan et al., 2017; Pearce
were treated with both FTY720 and an anti-CD62L antibody, the et al., 2009). However, memory T cells in the BM of mice on DR
latter restricting access to LN (Harp et al., 2010). Anti-CD62L treat- were comparable in their ability to take up and store lipids
ment alone resulted in an increase in T cell numbers in BM (Fig- compared to cells in the BM of mice fed ad libitum (Figures
ure 4F), presumably due to an inhibition of LN entry. In mice on S4H and S4I).
DR (but not in mice fed ad libitum) treated with both FTY720 Collectively, these data show that DR is associated with a
and the anti-CD62L antibody, the number of TCM in BM was signif- complex set of alterations within the BM, which is associated
icantly reduced compared to vehicle treated DR mice (Figure 4F). with an increase in chemokines, RBC, and adipocytes, all of
Further, transfer of pre-activated CD8+ T cells lacking the S1P1R which have the potential to promote circulating memory T cell
showed reduced accumulation in BM of mice on DR (Figure 4G). accumulation within the BM niche.
Together, these results supported the idea that CXCR4-CXCL12
and S1P-S1P1R contributed to increased recruitment and/or Memory T Cells Are in a State of Energy Conservation
retention of circulating memory T cells in BM during DR. during Dietary Restriction
DR is associated with increased adipogenesis in the BM (Caw- We next assessed whether memory T cells had undergone
thorn et al., 2014; Devlin et al., 2010). This was confirmed by an intrinsic adaptations to maximize long-term survival during DR.
increase in cells expressing PERILIPIN-1, a marker of lipid drop- Principal-component analysis (PCA) of mRNA expression by
lets, in BM of mice on DR, as well as increased expression of the TCM from BM and spleen revealed that the majority of the

Cell 178, 1088–1101, August 22, 2019 1093


A

B C

D E F

G H I

Figure 4. BM Remodeling during DR Promotes Memory T Cell Recruitment and Retention


(A) Scatterplot showing gene expression of whole-BM samples from mice fed ad libitum or on DR for 3 weeks. TPR, transcripts per million.
(B) CXCR4 expression by TCM in BM and spl of mice fed ad libitum (blue) or on DR (red).
(C) CD8+ T cells from Cxcr4fl/fl 3 UBC-CreERT2 (Cre ; WT, Cre+; Cxcr4 / ) mice treated with tamoxifen were in vitro activated and transferred into mice fed
ad libitum or on DR for 3 weeks. Analysis was performed 1 week post transfer. Graphs expressed in the same way as 3G.
(D) Images of BM from mice on DR for 3 weeks, showing TEr-119 (green) and DAPI (blue).
(E) Number of RBC (CD71+ precursors and CD71 mature cells) in BM of mice on DR for 3 weeks.
(F) Number of CD8+ TCM in BM after 1 week of DR treated with FTY720 (FTY), an anti-CD62L antibody, or a combination of both. Treatment commenced at same
time as initiation of DR.
(G) CD8+ T cells from S1pr1fl/fl x UBC-CreERT2 (Cre ; WT, Cre+; S1pr1 / ) mice that had been treated with tamoxifen were activated in vitro and transferred into
mice fed ad libitum or on DR for 3 weeks. Graph expressed in same way as 3G.
(H) Image showing BM of mice on DR for 3 weeks, staining for adipocytes by PERILIPIN-1 (cyan), CD8+ T cells (red), and DAPI (blue). Graph shows Fabp4 gene
expression in whole BM as determined by qPCR.
(I) Adipoq-CreERT2 3 Rosa26-DTA mice were put on DR for 2 weeks, treated with tamoxifen, and then maintained on DR for 3 weeks. Graph shows the number of
CD8+ TCM in BM.
Each symbol represents an individual mouse except for (A), in which each symbol represents an individual gene. Data show the mean representative of at least
two (D, F, and H) or pooled from two (C, E, and I) or three (A, B, and G) experiments with three to five mice per group. *p < 0.05, **p < 0.01, ***p < 0.001,
****p < 0.0001; two-tailed unpaired Student’s t test. ns, not significant. See also Figure S4.

variance could be explained by the nutritional status of the host (GSEA) revealed an enrichment for pathways associated with
(PC1, 24.29%), while effects of the tissue also had an effect on heat-shock protein chaperone binding and regulation of protein
gene expression (PC2, 12.87%) (Figure S5A). Pathway analysis folding in TCM from both BM and spleen of mice on DR (Figures
by gene ontogeny (GO) terms and gene set enrichment analysis 5A and 5B). In agreement with unaltered fatty-acid metabolism in

1094 Cell 178, 1088–1101, August 22, 2019


A

C D

Figure 5. Memory T Cells Are in a State of Energy Conservation during DR


(A and B) CD8+ TCM were sorted from spl and BM of mice fed ad libitum or on DR for 3 weeks for RNA sequencing. (A) GO terms and (B) GSEA from genes
upregulated in TCM from BM (blue) or spl (green) of mice on DR.
(C and D) (C) OCR and (D) SRC in TCM from SLO and BM of mice on DR for 3 weeks.
(E) Expression of phosphorylated mTOR2448, S6240/244, and AKT473 in TCM from spl and BM of mice on DR for 3 weeks.
Each symbol represents an individual mouse except for (C), which is pooled from 10–15 mice. Data are expressed as mean, or mean ± SD, pooled from
three biological replicates with five mice per group (A and B), representative of 3 (C), or pooled from three to four experiments (D and E). *p < 0.05, ***p < 0.001,
****p < 0.0001; ns, not significant. Two-tailed unpaired Student’s t test. See also Figure S5.

the context of DR (Figures S4H and S4I), we did not observe an oxygen consumption rate (OCR) (Figure 5C) and spare respi-
enrichment in pathways associated with lipid metabolism (Fig- ratory capacity (SRC) (Figure 5D), suggesting that memory
ures 5A and 5B). Of interest, genes associated with amino-acid T cells were in a state of reduced metabolic activity or quies-
deprivation and the cellular response to rapamycin, a negative cence during DR. Examination of mTOR in TCM from BM of
regulator of the mechanistic target of rapamycin (mTOR), fell mice on DR showed reduced phosphorylation at serine 2448
within the protein folding module (Figures 5A and 5B), supporting (Figure 5E), indicating a reduction in mTOR activity. Further,
the idea that TCM may have reduced levels of mTOR signaling TCM from the BM of mice on DR showed reduced phosphory-
during DR. lation of ribosomal protein S6 (pS6) at serine 240/244 and AKT
mTOR is a protein kinase that functions in larger multipro- at serine 473, downstream targets of mTORC1 and mTORC2,
tein complexes known as mTOR complex 1 (mTORC1) and respectively (Figure 5E). Of note, mTOR downregulation was
mTOR complex 2 (mTORC2) (Saxton and Sabatini, 2017). more pronounced in TCM from BM compared to those in the
Reduced mTOR signaling promotes survival during nutrient spleen in the context of DR (Figure 5E), supporting the idea
deprivation by decreasing anabolic processes and promoting that circulating memory T cells may integrate local cues
those that are catabolic (Aramburu et al., 2014). In agreement, when in BM to adopt a program compatible with energy con-
TCM in both the BM and SLO of mice on DR showed reduced servation or quiescence.

Cell 178, 1088–1101, August 22, 2019 1095


Memory T Cells Mediate Enhanced Protection against FTY720 had a similar bacterial burden to mice treated with
Secondary Challenges during Dietary Restriction vehicle (Figure 6D). In contrast, mice on DR that were treated
The low metabolic activity of memory T cells during DR and their with FTY720 no longer displayed enhanced protection (Fig-
BM tropism begged the question of whether mice on DR could ure 6D), suggesting that enhanced protection to secondary
rapidly respond to secondary challenges. To address this, we first infections was lost during DR if memory T cells were even
employed a model of acute oral infection with Yptb DyopM in transiently restricted from gaining access to the BM.
which rapid clearance of the microbe is associated with the induc- To determine if enhanced memory T cell function during DR
tion of a robust population of memory CD8+ T cells (Han et al., extended to other settings, we employed a model in which virally
2017). To investigate circulating memory T cells, naive or previ- induced T cells provide control of melanoma (Overwijk et al.,
ously infected mice were first challenged intravenously (i.v.) with 2003). Mice received naive transgenic CD8+ pmel-1 cells that
the virulent WT strain of Yptb. At 2 days post primary infection, are specific for an epitope derived from the melanoma-associ-
no difference was observed in bacterial burden between naive ated antigen gp100 before being infected with a recombinant
mice fed ad libitum or on DR (Figure 6A, columns A versus B). vaccinia virus that expresses human gp100 (VV-hgp100) (Fig-
As expected, previously infected mice fed ad libitum showed ure 6E). 1 month later, a time at which gp-100-specific memory
enhanced protection compared to naive mice 2 days post sec- CD8+ T cells had formed, DR was initiated for 3 weeks, and
ondary challenge (Han et al., 2017) (Figure 6A, columns A versus mice were inoculated with B16 melanoma cells. All mice fed ad
C). Antibody depletion in mice fed ad libitum revealed that memory libitum without pmel-1 cells reached the experimental endpoint
CD8+ T cells played a role in mediating protection (Figure 6A, col- (15 mm 3 15 mm) 26 days post tumor inoculation (Figure 6E,
umns C versus E), whereas mice depleted of CD4+ T cells trended black line), and consistent with previous reports (Kalaany and
toward having defective protection, although this did not reach Sabatini, 2009), DR alone provided a survival benefit (Figure 6E,
statistical significance (Figure 6A, columns C versus G). Next, blue). As previously shown (Xiao et al., 2011), infection with
mice received a primary infection and developed memory while VV-hgp100 alone, without pmel-1 cells, did not provide addi-
fed ad libitum before being placed on DR for 3–4 weeks and tional benefit compared to controls (Figure 6E, magenta and
then were challenged i.v. with WT Yptb. Remarkably, these mice green), whereas mice fed ad libitum that received both pmel-1
on DR during the secondary challenge had an enhanced ability cells and VV-hgp100 had a survival benefit (Figure 6E, red),
to control the infection, with approximately 500-fold fewer col- although the endpoint was reached by day 35 (Figure 6E, red).
ony-forming units (CFUs) of Yptb per gram of spleen compared Critically, mice on DR that received both pmel-1 cells and VV-
to previously infected mice fed ad libitum throughout (Figure 6A, hgp100 had the greatest survival rate (Figure 6E, orange), with
columns C versus D). Enhanced protection was also observed approximately 15% of mice from this group being completely tu-
in mice on DR during an oral secondary challenge with WT Yptb mor free after more than 2 months (Figure 6E). These results indi-
(Figure 6B). Thus, DR initiated post the establishment of memory cate that DR may enhance the protective function of memory
resulted in significantly enhanced protective responses. On the T cells in the context of both anti-bacterial and tumor immunity.
other hand, mice on DR throughout both the primary and second-
ary infections displayed a similar ability to control a secondary DISCUSSION
infection as mice fed ad libitum (Figure S6A). Thus, the timing of
DR was critical for the enhancement of secondary responses. Host fitness depends on the ability to survive and adapt to
Increased protection was associated with an enhanced changing environments. In these contexts, the long-term persis-
breadth of effector responses. Indeed, depletion of CD8+ (Fig- tence of memory T cells is of the utmost importance. How the
ure 6A, columns E versus F) or CD4+ (Figure 6A, columns G host responds in a way that prioritizes and protects defined
versus H) T cells alone did not revert the enhanced protection branches of the immune system during nutritional stress remains
afforded by DR. We found that enhanced protection was only poorly understood. Previous work supported the idea that the
reverted by the simultaneous depletion of both T cell subsets anorexic response following acute primary infections could pro-
(Figure 6A, columns I versus J). This indicates that both subsets mote host survival in defined contexts (Hart, 1988; Wang et al.,
contribute to enhanced secondary responses during DR and are 2016) and that a reduction in vitamin A results in decreased
able to effectively compensate for each other in this context. adaptive immunity but a compensatory increase in innate re-
Further, Yptb-specific memory CD8+ T cells isolated from the sponses aimed at protecting barrier sites (Spencer et al.,
spleen had a greater potential to produce interferon (IFN)g during 2014). Consistent with a recent report (Contreras et al., 2018),
DR (Figure 6C). Altogether, this indicates that memory T cells we found that a physiological decrease in calories was associ-
have enhanced functional capacity and ability to control second- ated with a complex set of alterations that resulted in the stra-
ary infections during DR. tegic protection of memory T cells within the BM (Figure 6F).
We next assessed whether appropriate homing during DR Our results supported the idea that this response was domi-
contributed to the optimization of immunological memory. We nantly coordinated by the steroid hormones GCs, which play a
previously showed that short-term treatment with FTY720 critical role in the regulation of energy balance (Cain and Cidlow-
reduced the accumulation of T cells in BM (Figures 4F and ski, 2017). Here, we show that one of the consequences of
S4F). Mice that had developed memory to Yptb were placed GC-mediated accumulation of T cells in BM was to protect these
on DR for 2 weeks then treated over the course of 1 week with cells from the detrimental effects of GCs themselves. While
FTY720 followed by a period of 2 weeks before challenge (Fig- circulating memory T cells were particularly enriched within the
ure 6D). Mice fed ad libitum throughout and treated with BM, other key subsets, such as Treg cells, NK cells, and plasma

1096 Cell 178, 1088–1101, August 22, 2019


****
ns
ns
A ns ** B C Spleen
****
ns **** Ctrl ****

WT Yptb CFU/g mLN


****

IFN + of YopE69-77+(%)
**** * * * 9
WT Yptb CFU/g spl

9
10 DR 10 ns Ctrl 100 **
108 Oral Oral DR
108 80
Yptb ΔyopM 1° Yptb WT 2° ***
107 107
106 60
50% DR 106
105 40
CFU 105
104 4wk 3wk 3d
10 4 20
103
102 103 0
CD8 - - - - + + - - + + Naive Memory

trl
R
D
CD4 - - - - - - + + + +

C
A B C D E F G H I J
Naive Memory *** ns
D Oral i.v. 109
ns ** Veh.

WT Yptb CFU/g spl


i.v. Oral i.v.
Yptb ΔyopM 1° Yptb WT 2° 108 FTY
Yptb WT Yptb ΔyopM 1° Yptb WT 2°
107
50% DR 50% DR 50% DR 106
CFU 105
CFU CFU 4wk 2wk 2wk 2d
3wk 2d 4wk 3wk 2d 104
FTYx4
every 2nd day 103
102
Ctrl DR
100 Ctrl (n=33)
E

****
Pmel-1 + n
DR (n=34)
80 n s
VV-hgp100 s.c. B16 Ctrl + VV-hgp100 (n=9)

*
s
Survival (%)

****
60 DR + VV-hgp100 (n=10)
50% DR Ctrl + VV-hgp100 + pmel-1 (n=22)

****
40 DR + VV-hgp100 + pmel-1 (n=21)
4wk 3wk Measure
20
0
F 0 10 20 30 40 50 60 70 80
Ad Dietary Days post tumour inoculation
Libitum Restriction
Adipocyte
Blood/ RBC
SLO GC Memory T cell
RBC
Pathogen
Cxcr4 Cxcr4 Cxcr4
Bcl-2
mTor
Cxcl12 Cxcl12

Bone
Marrow

Figure 6. Memory T Cells Mediate Enhanced Protection against Secondary Challenges during DR
(A) Mice that were naive or previously infected with the Yptb DyopM orally were put on DR for 3 weeks and challenged i.v. with the more virulent WT strain of Yptb.
Bacterial burden was assessed in spl 2 days after challenge. CD4+ or CD8+ T cells were depleted prior to challenge in the indicated groups.
(B) Mice were infected with Yptb DyopM orally then 4 weeks later put on DR for 3 weeks. During DR, mice were challenged orally with WT Yptb. Bacterial burden
was assessed in mesenteric lymph nodes 3 days after challenge.
(C) IFNg production following PMA/ionomycin stimulation by Yptb-specific splenic memory CD8+ T cells at day 2 after i.v. challenge with WT Yptb.
(D) Mice previously infected with Yptb DyopM orally were put on DR for approximately 5 weeks. Two weeks into DR, mice were treated with FTY720 four times
every other day followed by a 2-week period before i.v. challenge with WT Yptb. Bacterial burden was assessed in spl 2 days after challenge.
(E) Mice received naive pmel cells followed by an i.v. infection with vaccinia-hgp100. 1 month later, mice were put on 50% DR for 3 weeks and inoculated with B16
cells subcutaneously (s.c.). Mice were euthanized when tumors reached 15 mm 3 15 mm.
(F) Circulating GCs are increased during DR, inducing memory T cell recruitment and retention in BM, a site with low levels of GCs. The BM is drastically
remodeled during DR to contain increased adipocytes, RBCs, and CXCL12, which have the potential to maintain memory T cells in this niche. Once in BM during
DR, memory T cells express optimal levels of BCL-2 and mTOR to promote survival. Upon secondary challenges, memory T cells have an enhanced ability to
mediate protection during DR.
Each symbol represents an individual mouse. Data show the mean pooled from at least two (A, B, and E), 4 (C), or 3 (D) experiments. *p < 0.05, **p < 0.01,
***p < 0.001, ****p < 0.0001; ns, not significant. Two-tailed unpaired Student’s t test except for (E), which was performed by a log-rank (Mantel-Cox) test. See also
Figure S6.

Cell 178, 1088–1101, August 22, 2019 1097


cells, were preserved at this site during DR, supporting the idea (Scheller et al., 2015, 2016). Reports indicate that memory
that our findings may broadly apply to many keystone popula- T cells critically require FAO to fuel their metabolism and long-
tions. In addition to its fundamental role in hematopoiesis, the term survival (O’Sullivan et al., 2014; Pearce et al., 2009; van
BM is a site that circulating memory T cells traffic through at der Windt et al., 2012), although whether this is the case in vivo
steady state (Becker et al., 2005; Chaix et al., 2014; Di Rosa is unclear (Raud et al., 2018a, 2018b). The import of extracellular
and Gebhardt, 2016; Klonowski et al., 2004; Mazo et al., 2005; glycerol has been shown to be critical for this process in circu-
Pabst et al., 1986; Parretta et al., 2005). Under resting condi- lating memory T cells (Buck et al., 2016; Cui et al., 2015; O’Sulli-
tions, CXCR4-CXCL12 interactions contribute to the homing of van et al., 2014; van der Windt et al., 2012), while the survival of
memory T cells to BM (Mazo et al., 2005). Of relevance, GCs skin TRM requires uptake of exogenous lipids (Pan et al., 2017).
can upregulate CXCR4 expression by T cells (Besedovsky We found that memory T cells in BM during DR did not have
et al., 2014a, 2014b; Ghosh et al., 2009; Shimba et al., 2018), an enhanced ability to uptake, store, or process fatty acids.
while reduced mTORC2 signaling also results in increased Indeed, unlike white adipocytes that undergo high rates of lipol-
CXCR4 expression and sequestration of naive T cells in BM ysis, BM adipocytes undergo only minimal amounts of lipolysis
(Arojo et al., 2018). As such, our results support the idea that (Scheller et al., 2019). As such, a key question that remains to
the CXCR4-CXCL12 axis dominantly contributes to T cell reten- be addressed is how BM adipocytes contribute to T cell survival
tion and/or accumulation within the BM during DR. and/or persistence in the BM niche during DR.
Protection of memory T cells within the BM was not associated Our work supports the idea that memory T cells from hosts on
with permanent sequestration but rather increased dwell time in DR had entered a state of energy conservation or quiescence
a privileged site, a response that may result from the dramatic in- that was associated with reduced mTOR signaling. When nutri-
crease in erythropoiesis observed during DR. RBCs are the ma- ents are abundant, mTORC1 is activated to stimulate anabolic
jor source of S1P in blood (Hänel et al., 2007; Pappu et al., 2007), processes that supports cell growth, whereas when these fac-
a molecule critical in orchestrating T cell migration (Cyster and tors are reduced, mTORC1 signaling is suppressed, stimulating
Schwab, 2012). As such, an increase of RBC in BM may impact catabolic processes (Saxton and Sabatini, 2017). This places
the local concentration of S1P within the BM, delaying the ability mTOR as a key rheostat that balances the metabolic profile of
of T cells to re-acquire S1P1R and re-enter the circulation cells to align with the level of nutrients or stress within the host
(Maeda et al., 2010). Furthermore, S1P may play an important (Aramburu et al., 2014). Restricting the downregulation of
role in the survival of T cells within the BM compartment during mTOR under conditions of nutrient deprivation in vitro is detri-
DR (Mendoza et al., 2017). mental for cell survival and function (Choo et al., 2010). Of rele-
The role of GCs in coordinating these responses also supports vance to our observation, reduced mTOR activity enhances
the idea that our observation may be relevant to numerous forms memory T cell development, maintenance, and function
of stress, promoting cell survival in the privileged environment of (Araki et al., 2009). Consistent with this, mTOR inhibitors show
the BM. Conversely, this adaptive response may have detri- promise in the clinic by enhancing vaccine responses and anti-
mental consequences. Notably, in the context of glioblastoma, viral immunity (Mannick et al., 2018). As such, this phenomenon
T cells were severely depleted from the periphery and tumor may provide a mechanistic link for our observation of enhanced
site but found at high numbers in BM, and as in our settings, protective responses in the context of DR.
S1P-S1P1R interactions contributed to this response (Chongsa- Together, our work raises intriguing questions regarding the
thidkiet et al., 2018). Thus, what may have evolved as a means to optimal state for immunological memory preservation and func-
adapt to physiological alterations in food intake or stress may tion. Access to a constant level of calories and nutrients is only a
also be co-opted in the setting of cancer, depriving the tumor recent occurrence in high-income countries, and all animals
microenvironment of the adaptive immune repertoire. have evolved in settings of variable food availability. As such,
We and others have shown that WAT is a hub for memory fluctuations in nutrition or intermittent fasting may represent
T cells (Han et al., 2017; Masopust et al., 2001). However, this the ideal state for supporting the function of memory T cells.
compartment rapidly declines during DR, whereas the BM para- Furthermore, understanding the nature of the factors within the
doxically becomes enriched in adipocytes (Cawthorn et al., BM milieu that potentially optimizes immunological memory
2014; Devlin et al., 2010). Under these settings, the BM acts as could be utilized to improve and protect fundamental arms of
an endocrine organ that is critical for compensating for the loss the immune system in the context of disease.
of WAT in the periphery (Cawthorn et al., 2014). Further in align-
ment with WAT being a privileged site for memory T cells, abla- STAR+METHODS
tion of adipocytes during DR reduced the ability of memory
T cells to accumulate in BM. Of note, increased adipogenesis Detailed methods are provided in the online version of this paper
in BM occurs in several settings, such as irradiation and chemo- and include the following:
therapy (Zhou et al., 2017), suggesting that this response could
represent a general response following stress aimed at preser- d KEY RESOURCES TABLE
ving physiological processes, including immunological memory. d LEAD CONTACT AND MATERIALS AVAILABILITY
An intriguing question that remains to be addressed is the na- d EXPERIMENTAL MODEL AND SUBJECT DETAILS
ture of the adipocytes and adipocyte-derived fuels that develop B Mice
in BM during DR. BM adipocytes express a complex gene B Cell lines
signature associated with both white and brown adipocytes B Bacteria and virus strains

1098 Cell 178, 1088–1101, August 22, 2019


d METHOD DETAILS Arojo, O.A., Ouyang, X., Liu, D., Meng, T., Kaech, S.M., Pereira, J.P., and Su,
B Dietary restriction B. (2018). Active mTORC2 Signaling in Naive T Cells Suppresses Bone Marrow
B Surgery Homing by Inhibiting CXCR4 Expression. J. Immunol. 201, 908–915.

B Tissue processing and flow cytometry Becker, T.C., Coley, S.M., Wherry, E.J., and Ahmed, R. (2005). Bone marrow is
B Confocal microscopy a preferred site for homeostatic proliferation of memory CD8 T cells.
+ J. Immunol. 174, 1269–1273.
B In vitro activation of CD8 T cells and adoptive transfer
Besedovsky, L., Born, J., and Lange, T. (2014a). Endogenous glucocorticoid
B In vivo treatments
receptor signaling drives rhythmic changes in human T-cell subset numbers
B Ex vivo assays
and the expression of the chemokine receptor CXCR4. FASEB J. 28, 67–75.
B RNA extraction, cDNA synthesis and Quantitative PCR
Besedovsky, L., Linz, B., Dimitrov, S., Groch, S., Born, J., and Lange, T.
B RNA Sequencing
(2014b). Cortisol increases CXCR4 expression but does not affect CD62L
B ELISA and CCR7 levels on specific T cell subsets in humans. Am. J. Physiol. Endocri-
B 2-photon imaging and analysis nol. Metab. 306, E1322–E1329.
B Oxygen consumption flux evaluation Buck, M.D., O’Sullivan, D., Klein Geltink, R.I., Curtis, J.D., Chang, C.H., Sanin,
d QUANTIFICATION AND STATISTICAL ANALYSIS D.E., Qiu, J., Kretz, O., Braas, D., van der Windt, G.J., et al. (2016). Mitochon-
d DATA AND CODE AVAILABILITY drial Dynamics Controls T Cell Fate through Metabolic Programming. Cell
166, 63–76.
SUPPLEMENTAL INFORMATION Cain, D.W., and Cidlowski, J.A. (2017). Immune regulation by glucocorticoids.
Nat. Rev. Immunol. 17, 233–247.
Supplemental Information can be found online at https://doi.org/10.1016/j. Cawthorn, W.P., Scheller, E.L., Learman, B.S., Parlee, S.D., Simon, B.R., Mori,
cell.2019.07.049. H., Ning, X., Bree, A.J., Schell, B., Broome, D.T., et al. (2014). Bone marrow ad-
ipose tissue is an endocrine organ that contributes to increased circulating adi-
ACKNOWLEDGMENTS ponectin during caloric restriction. Cell Metab. 20, 368–375.
Chaix, J., Nish, S.A., Lin, W.H., Rothman, N.J., Ding, L., Wherry, E.J., and
Y.B. is supported by the Division of Intramural Research of NIAID (NIAID; ZIA-
Reiner, S.L. (2014). Cutting edge: CXCR4 is critical for CD8+ memory T cell ho-
AI001132, ZIA-AI001133), NIH. N.C. and Y.B. were supported by the Office of
meostatic self-renewal but not rechallenge self-renewal. J. Immunol. 193,
Dietary Supplements Research Scholar program, NIH. Work in the Schwab lab
1013–1016.
was supported by NIH grants R01AI085166 and R01AI123308, and a Vilcek
Scholar Award to D.D. We thank the Belkaid lab for their suggestions, support, Chongsathidkiet, P., Jackson, C., Koyama, S., Loebel, F., Cui, X., Farber, S.H.,
and critical reading of the manuscript and J. Kehr and N. Bouladoux for man- Woroniecka, K., Elsamadicy, A.A., Dechant, C.A., Kemeny, H.R., et al. (2018).
aging the program of the laboratory. We thank K. Beacht, E. Lewis, and J. Le- Sequestration of T cells in bone marrow in the setting of glioblastoma and other
Grand for technical assistance; the NIAID animal facility staff; T. Hawley (NIAID intracranial tumors. Nat. Med. 24, 1459–1468.
flow cytometry facility); O. Schwartz (NIAID biological imaging facility); and B. Choo, A.Y., Kim, S.G., Vander Heiden, M.G., Mahoney, S.J., Vu, H., Yoon,
Tran and J. Sheti (NCI sequencing core facility). S.O., Cantley, L.C., and Blenis, J. (2010). Glucose addiction of TSC null cells
is caused by failed mTORC1-dependent balancing of metabolic demand
AUTHOR CONTRIBUTIONS with supply. Mol. Cell 38, 487–499.

Contreras, N.A., Fontana, L., Tosti, V., and Nikolich-Zugich, J. (2018). Calorie
N.C. and Y.B. designed the study and wrote the manuscript. N.C. performed restriction induces reversible lymphopenia and lymphoid organ atrophy due to
experiments and analyzed the data. S.-J.H., M.E., V.M.L., B.H., E.A.M., cell redistribution. Geroscience 40, 279–291.
J.P.S., R.J.K., and D.D. participated in performing experiments, provided intel-
Cui, G., Staron, M.M., Gray, S.M., Ho, P.C., Amezquita, R.A., Wu, J., and
lectual expertise, and helped to interpret experimental results. S.R.S., N.P.R.,
Kaech, S.M. (2015). IL-7-Induced Glycerol Transport and TAG Synthesis Pro-
H.D.H., D.B.M., and P.L.S. provided reagents and intellectual expertise and
motes Memory CD8+ T Cell Longevity. Cell 161, 750–761.
helped to interpret experimental results.
Cyster, J.G., and Schwab, S.R. (2012). Sphingosine-1-phosphate and lympho-
DECLARATION OF INTERESTS cyte egress from lymphoid organs. Annu. Rev. Immunol. 30, 69–94.
Devlin, M.J., Cloutier, A.M., Thomas, N.A., Panus, D.A., Lotinun, S., Pinz, I.,
The authors declare no competing interests. Baron, R., Rosen, C.J., and Bouxsein, M.L. (2010). Caloric restriction leads
to high marrow adiposity and low bone mass in growing mice. J. Bone Miner.
Received: December 18, 2018 Res. 25, 2078–2088.
Revised: May 28, 2019 Di Rosa, F., and Gebhardt, T. (2016). Bone Marrow T Cells and the Integrated
Accepted: July 29, 2019 Functions of Recirculating and Tissue-Resident Memory T Cells. Front. Immu-
Published: August 22, 2019 nol. 7, 51.
Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut,
REFERENCES
P., Chaisson, M., and Gingeras, T.R. (2013). STAR: ultrafast universal RNA-seq
aligner. Bioinformatics 29, 15–21.
Acosta-Rodrı́guez, V.A., de Groot, M.H.M., Rijo-Ferreira, F., Green, C.B., and
Takahashi, J.S. (2017). Mice under Caloric Restriction Self-Impose a Temporal Flygare, J., Rayon Estrada, V., Shin, C., Gupta, S., and Lodish, H.F. (2011). HI-
Restriction of Food Intake as Revealed by an Automated Feeder System. Cell F1alpha synergizes with glucocorticoids to promote BFU-E progenitor self-
Metab. 26, 267–277. renewal. Blood 117, 3435–3444.
Araki, K., Turner, A.P., Shaffer, V.O., Gangappa, S., Keller, S.A., Bachmann, Fujita, Y., Murakami, M., Ogawa, Y., Masuzaki, H., Tanaka, M., Ozaki, S., Na-
M.F., Larsen, C.P., and Ahmed, R. (2009). mTOR regulates memory CD8 kao, K., and Mimori, T. (2002). Leptin inhibits stress-induced apoptosis of T
T-cell differentiation. Nature 460, 108–112. lymphocytes. Clin. Exp. Immunol. 128, 21–26.
Aramburu, J., Ortells, M.C., Tejedor, S., Buxadé, M., and López-Rodrı́guez, C. Gerlach, C., Moseman, E.A., Loughhead, S.M., Alvarez, D., Zwijnenburg, A.J.,
(2014). Transcriptional regulation of the stress response by mTOR. Sci. Signal. Waanders, L., Garg, R., de la Torre, J.C., and von Andrian, U.H. (2016). The
7, re2. Chemokine Receptor CX3CR1 Defines Three Antigen-Experienced CD8 T

Cell 178, 1088–1101, August 22, 2019 1099


Cell Subsets with Distinct Roles in Immune Surveillance and Homeostasis. Im- Mendoza, A., Fang, V., Chen, C., Serasinghe, M., Verma, A., Muller, J., Chalu-
munity 45, 1270–1284. vadi, V.S., Dustin, M.L., Hla, T., Elemento, O., et al. (2017). Lymphatic endothe-
Ghosh, M.C., Baatar, D., Collins, G., Carter, A., Indig, F., Biragyn, A., and Taub, lial S1P promotes mitochondrial function and survival in naive T cells. Nature
D.D. (2009). Dexamethasone augments CXCR4-mediated signaling in resting 546, 158–161.
human T cells via the activation of the Src kinase Lck. Blood 113, 575–584. Mittelstadt, P.R., Monteiro, J.P., and Ashwell, J.D. (2012). Thymocyte respon-
siveness to endogenous glucocorticoids is required for immunological fitness.
Han, S.J., Glatman Zaretsky, A., Andrade-Oliveira, V., Collins, N., Dzutsev, A.,
J. Clin. Invest. 122, 2384–2394.
Shaik, J., Morais da Fonseca, D., Harrison, O.J., Tamoutounour, S., Byrd, A.L.,
et al. (2017). White Adipose Tissue Is a Reservoir for Memory T Cells and Pro- Mueller, S.N., and Mackay, L.K. (2016). Tissue-resident memory T cells: local
motes Protective Memory Responses to Infection. Immunity 47, 1154–1168. specialists in immune defence. Nat. Rev. Immunol. 16, 79–89.

Hänel, P., Andréani, P., and Gräler, M.H. (2007). Erythrocytes store and release Nikolich-Zugich, J., and Messaoudi, I. (2005). Mice and flies and monkeys too:
sphingosine 1-phosphate in blood. FASEB J. 21, 1202–1209. caloric restriction rejuvenates the aging immune system of non-human pri-
mates. Exp. Gerontol. 40, 884–893.
Harp, J.R., Gilchrist, M.A., and Onami, T.M. (2010). Memory T cells are en-
riched in lymph nodes of selectin-ligand-deficient mice. J. Immunol. 185, O’Sullivan, D., van der Windt, G.J., Huang, S.C., Curtis, J.D., Chang, C.H.,
5751–5761. Buck, M.D., Qiu, J., Smith, A.M., Lam, W.Y., DiPlato, L.M., et al. (2014). Mem-
ory CD8(+) T cells use cell-intrinsic lipolysis to support the metabolic program-
Hart, B.L. (1988). Biological basis of the behavior of sick animals. Neurosci. ming necessary for development. Immunity 41, 75–88.
Biobehav. Rev. 12, 123–137.
Overwijk, W.W., Theoret, M.R., Finkelstein, S.E., Surman, D.R., de Jong, L.A.,
Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X., Vyth-Dreese, F.A., Dellemijn, T.A., Antony, P.A., Spiess, P.J., Palmer, D.C.,
Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage- et al. (2003). Tumor regression and autoimmunity after reversal of a functionally
determining transcription factors prime cis-regulatory elements required for tolerant state of self-reactive CD8+ T cells. J. Exp. Med. 198, 569–580.
macrophage and B cell identities. Mol. Cell 38, 576–589.
Pabst, R., Miyasaka, M., and Dudler, L. (1986). Numbers and phenotype of
Jameson, S.C., and Masopust, D. (2018). Understanding Subset Diversity in T lymphocytes emigrating from sheep bone marrow after in situ labelling with
Cell Memory. Immunity 48, 214–226. fluorescein isothiocyanate. Immunology 59, 217–222.
Kaech, S.M., and Cui, W. (2012). Transcriptional control of effector and mem- Pan, Y., Tian, T., Park, C.O., Lofftus, S.Y., Mei, S., Liu, X., Luo, C., O’Malley,
ory CD8+ T cell differentiation. Nat. Rev. Immunol. 12, 749–761. J.T., Gehad, A., Teague, J.E., et al. (2017). Survival of tissue-resident memory
Kalaany, N.Y., and Sabatini, D.M. (2009). Tumours with PI3K activation are T cells requires exogenous lipid uptake and metabolism. Nature 543, 252–256.
resistant to dietary restriction. Nature 458, 725–731. Pappu, R., Schwab, S.R., Cornelissen, I., Pereira, J.P., Regard, J.B., Xu, Y.,
Karsten, E., Breen, E., and Herbert, B.R. (2018a). Red blood cells are dynamic Camerer, E., Zheng, Y.W., Huang, Y., Cyster, J.G., and Coughlin, S.R.
reservoirs of cytokines. Sci. Rep. 8, 3101. (2007). Promotion of lymphocyte egress into blood and lymph by distinct sour-
ces of sphingosine-1-phosphate. Science 316, 295–298.
Karsten, E., Hill, C.J., and Herbert, B.R. (2018b). Red blood cells: The primary
Parretta, E., Cassese, G., Barba, P., Santoni, A., Guardiola, J., and Di Rosa, F.
reservoir of macrophage migration inhibitory factor in whole blood. Cytokine
(2005). CD8 cell division maintaining cytotoxic memory occurs predominantly
102, 34–40.
in the bone marrow. J. Immunol. 174, 7654–7664.
Klonowski, K.D., Williams, K.J., Marzo, A.L., Blair, D.A., Lingenheld, E.G., and
Pearce, E.L., Walsh, M.C., Cejas, P.J., Harms, G.M., Shen, H., Wang, L.S.,
Lefrançois, L. (2004). Dynamics of blood-borne CD8 memory T cell migration
Jones, R.G., and Choi, Y. (2009). Enhancing CD8 T-cell memory by modulating
in vivo. Immunity 20, 551–562.
fatty acid metabolism. Nature 460, 103–107.
Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold
Raud, B., McGuire, P.J., Jones, R.G., Sparwasser, T., and Berod, L. (2018a).
change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550.
Fatty acid metabolism in CD8+ T cell memory: Challenging current concepts.
Maeda, Y., Seki, N., Sato, N., Sugahara, K., and Chiba, K. (2010). Sphingosine Immunol. Rev. 283, 213–231.
1-phosphate receptor type 1 regulates egress of mature T cells from mouse Raud, B., Roy, D.G., Divakaruni, A.S., Tarasenko, T.N., Franke, R., Ma, E.H.,
bone marrow. Int. Immunol. 22, 515–525. Samborska, B., Hsieh, W.Y., Wong, A.H., Stuve, P., et al. (2018b). Etomoxir
Mandala, S., Hajdu, R., Bergstrom, J., Quackenbush, E., Xie, J., Milligan, J., Actions on Regulatory and Memory T Cells Are Independent of Cpt1a-Medi-
Thornton, R., Shei, G.J., Card, D., Keohane, C., et al. (2002). Alteration of ated Fatty Acid Oxidation. Cell Metab. 28, 504–515.
lymphocyte trafficking by sphingosine-1-phosphate receptor agonists. Sci- Redman, L.M., Smith, S.R., Burton, J.H., Martin, C.K., Il’yasova, D., and Rav-
ence 296, 346–349. ussin, E. (2018). Metabolic Slowing and Reduced Oxidative Damage with Sus-
Manglani, M., and McGavern, D.B. (2018). Intravital Imaging of Neuroimmune tained Caloric Restriction Support the Rate of Living and Oxidative Damage
Interactions Through a Thinned Skull. Curr. Protoc. Immunol. 120, Theories of Aging. Cell Metab 27, 805–815.
24.2.1–24.2.12. Robertson, L.T., and Mitchell, J.R. (2013). Benefits of short-term dietary re-
Mannick, J.B., Morris, M., Hockey, H.P., Roma, G., Beibel, M., Kulmatycki, K., striction in mammals. Exp. Gerontol. 48, 1043–1048.
Watkins, M., Shavlakadze, T., Zhou, W., Quinn, D., et al. (2018). TORC1 inhibi- Sallusto, F., Lenig, D., Förster, R., Lipp, M., and Lanzavecchia, A. (1999). Two
tion enhances immune function and reduces infections in the elderly. Sci. subsets of memory T lymphocytes with distinct homing potentials and effector
Transl. Med. 10. functions. Nature 401, 708–712.
Masopust, D., Vezys, V., Marzo, A.L., and Lefrançois, L. (2001). Preferential Saxton, R.A., and Sabatini, D.M. (2017). mTOR Signaling in Growth, Meta-
localization of effector memory cells in nonlymphoid tissue. Science 291, bolism, and Disease. Cell 169, 361–371.
2413–2417. Scheller, E.L., Doucette, C.R., Learman, B.S., Cawthorn, W.P., Khandaker, S.,
Matloubian, M., Lo, C.G., Cinamon, G., Lesneski, M.J., Xu, Y., Brinkmann, V., Schell, B., Wu, B., Ding, S.Y., Bredella, M.A., Fazeli, P.K., et al. (2015). Region-
Allende, M.L., Proia, R.L., and Cyster, J.G. (2004). Lymphocyte egress from specific variation in the properties of skeletal adipocytes reveals regulated and
thymus and peripheral lymphoid organs is dependent on S1P receptor 1. Na- constitutive marrow adipose tissues. Nat. Commun. 6, 7808.
ture 427, 355–360. Scheller, E.L., Cawthorn, W.P., Burr, A.A., Horowitz, M.C., and MacDougald,
Mazo, I.B., Honczarenko, M., Leung, H., Cavanagh, L.L., Bonasio, R., We- O.A. (2016). Marrow Adipose Tissue: Trimming the Fat. Trends Endocrinol.
ninger, W., Engelke, K., Xia, L., McEver, R.P., Koni, P.A., et al. (2005). Bone Metab. 27, 392–403.
marrow is a major reservoir and site of recruitment for central memory CD8+ Scheller, E.L., Khandaker, S., Learman, B.S., Cawthorn, W.P., Anderson, L.M.,
T cells. Immunity 22, 259–270. Pham, H.A., Robles, H., Wang, Z., Li, Z., Parlee, S.D., et al. (2019). Bone

1100 Cell 178, 1088–1101, August 22, 2019


marrow adipocytes resist lipolysis and remodeling in response to b-adrenergic van der Windt, G.J., Everts, B., Chang, C.H., Curtis, J.D., Freitas, T.C., Amiel,
stimulation. Bone 118, 32–41. E., Pearce, E.J., and Pearce, E.L. (2012). Mitochondrial respiratory capacity is
a critical regulator of CD8+ T cell memory development. Immunity 36, 68–78.
Shimba, A., Cui, G., Tani-Ichi, S., Ogawa, M., Abe, S., Okazaki, F., Kitano, S.,
Wang, A., Huen, S.C., Luan, H.H., Yu, S., Zhang, C., Gallezot, J.D., Booth, C.J.,
Miyachi, H., Yamada, H., Hara, T., et al. (2018). Glucocorticoids Drive Diurnal
and Medzhitov, R. (2016). Opposing Effects of Fasting Metabolism on Tissue
Oscillations in T Cell Distribution and Responses by Inducing Interleukin-7 Re-
Tolerance in Bacterial and Viral Inflammation. Cell 166, 1512–1525.
ceptor and CXCR4. Immunity 48, 286–298.
Wing, E.J., Magee, D.M., and Barczynski, L.K. (1988). Acute starvation in mice
Speakman, J.R., and Mitchell, S.E. (2011). Caloric restriction. Mol. Aspects reduces the number of T cells and suppresses the development of T-cell-
Med. 32, 159–221. mediated immunity. Immunology 63, 677–682.
Spencer, S.P., Wilhelm, C., Yang, Q., Hall, J.A., Bouladoux, N., Boyd, A., Nut- Xiao, H., Peng, Y., Hong, Y., Liu, Y., Guo, Z.S., Bartlett, D.L., Fu, N., and He, Y.
man, T.B., Urban, J.F., Jr., Wang, J., Ramalingam, T.R., et al. (2014). Adapta- (2011). Lentivector prime and vaccinia virus vector boost generate high-quality
tion of innate lymphoid cells to a micronutrient deficiency promotes type 2 bar- CD8 memory T cells and prevent autochthonous mouse melanoma.
rier immunity. Science 343, 432–437. J. Immunol. 187, 1788–1796.
Zhang, L., Prak, L., Rayon-Estrada, V., Thiru, P., Flygare, J., Lim, B., and Lod-
Surh, C.D., and Sprent, J. (2008). Homeostasis of naive and memory T cells.
ish, H.F. (2013). ZFP36L2 is required for self-renewal of early burst-forming unit
Immunity 29, 848–862.
erythroid progenitors. Nature 499, 92–96.
Tripathi, S., Pohl, M.O., Zhou, Y., Rodriguez-Frandsen, A., Wang, G., Stein, Zhou, B.O., Yu, H., Yue, R., Zhao, Z., Rios, J.J., Naveiras, O., and Morri-
D.A., Moulton, H.M., DeJesus, P., Che, J., Mulder, L.C., et al. (2015). Meta- son, S.J. (2017). Bone marrow adipocytes promote the regeneration of
and Orthogonal Integration of Influenza ‘‘OMICs’’ Data Defines a Role for stem cells and haematopoiesis by secreting SCF. Nat. Cell Biol. 19,
UBR4 in Virus Budding. Cell Host Microbe 18, 723–735. 891–903.

Cell 178, 1088–1101, August 22, 2019 1101


STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER


Antibodies
Anti-mouse B220, PE-CF594 (RA3-6B2) BD Cat# 562290; RRID:AB_11151901
Anti-mouse BrDU, FITC BD 557891
Anti-mouse CD4, BV605 (RM4-5) Biolegend BioLegend Cat# 100548; RRID:AB_2563054
Anti-mouse CD8b, PE (H35-17.2) eBioscience Cat# 12-0083-82; RRID:AB_657767
Anti-mouse CD8b, BV650 (H35-17.2) BD Cat# 740552; RRID:AB_2740253
Anti-mouse CD16/32, purified (2.4G2) Bio-X-Cell Cat# CUS-HB-197; RRID:AB_2687830
Anti-mouse CD25, AF488 (PC61.5) eBioscience Cat# 53-0251-82; RRID:AB_763472
Anti-mouse CD31, AF647 (MEC13.3) Biolegend Cat# 102516; RRID:AB_2161029
Anti-mouse CD44, PE-Cy7 (IM7) eBioscience Cat# 25-0441-82; RRID:AB_469623
Anti-mouse CD44, AF700 (IM7) eBioscience Cat# 56-0441-82; RRID:AB_494011
Anti-mouse CD45, APC-eFluor 780 (30-F11) eBioscience Cat# 47-0451-82; RRID:AB_1548781
Anti-mouse CD45.1, FITC (A20) eBioscience Cat# 11-0453-85; RRID:AB_465059
Anti-mouse CD45.2, APC-eFluor 780 (104) eBioscience Cat# 47-0454-82; RRID:AB_1272175
Anti-mouse CD62L, FITC (MEL-14) eBioscience Cat# 11-0621-85; RRID:AB_465110
Anti-mouse CD69, PE (H1.2F3) eBioscience Cat# 12-0691-83; RRID:AB_465733
Anti-mouse CD71, APC (R17217) eBioscience Cat# 17-0711-82; RRID:AB_1834355
Anti-mouse CD90.2, BV785 (30-H12) Biolegend Cat# 105331; RRID:AB_2562900
Anti-mouse CD122, PerCP-eFluor 710 (TM-b1) eBioscience Cat# 46-1222-82; RRID:AB_11064442
Anti-mouse CD127, BV605 (A7R34) Biolegend Cat# 135041; RRID:AB_2572047
Anti-mouse CXCR4, PE (2B11) eBioscience Cat# 12-9991-81; RRID:AB_891393
Anti-mouse CX3CR1, PE (SA001F11) Biolegend Cat# 149005; RRID:AB_2564314
Anti-mouse EOMES, eFluor450 (Dan11mag) eBioscience Cat# 48-4875-82; RRID:AB_2574062
Anti-mouse FOXP3, PerCP (FJK-16S) eBioscience Cat# 45-5773-82; RRID:AB_914351
Anti-mouse NK1.1, PeCy7 eBioscience Cat# 25-5941-82; RRID:AB_469665
Anti-mouse TCRgd, PE-CF594 (eBioGL3) BD Cat# 563532; RRID:AB_2661844
Anti-mouse TER-119, eFluor450 (TER-119) eBioscience Cat# 48-5921-82; RRID:AB_1518808
Anti-mouse Va2, PE (B20.1) eBioscience Cat# 12-5812-82; RRID:AB_465949
Anti-mouse BCL-2, eFluor450 (10C4) eBioscience Cat# 48-6992-42; RRID:AB_2574099
Anti-mouse IFN-g, eFluor450 (XMG1.2) eBioscience Cat# 48-7311-82; RRID:AB_1834366
Anti-human/mouse T-BET, BV421 (eBio4B10) Biolegend Cat# 644816; RRID:AB_10959653
Anti-mouse TCRb, PerCP-Cy5.5 (H57-597) eBioscience Cat# 45-5961-82; RRID:AB_925763
Anti-mouse AKT473, PeCy7 (MRRBY) eBioscience Cat# 12-9715-42; RRID:AB_2637101
Anti-mouse mTOR2448, PE (SDRNR) eBioscience Cat# 25-9718-42; RRID:AB_2573550
Anti-mouse pS6240/244, AF488 (D68F8) Cell signaling technologies Cat# 5018; RRID:AB_10695861
Anti-mouse PERILIPIN1, purified Abcam Cat# ab3526; RRID:AB_2167274
Normal Goat Serum Jackson ImmunoResearch Cat# 005-000-121; RRID:AB_2336990
Laboratories
Rat Gamma Globulin Jackson ImmunoResearch Cat# 012-000-002; RRID:AB_2337135
Laboratories
Goat anti-rabbit AF647 Life Technologies Cat# A27040; RRID:AB_2536101
Purified anti-mouse CD3 BD Cat# 553057; RRID:AB_394590
Purified anti-mouse CD28 BD Cat# 553294; RRID:AB_394763
(Continued on next page)

e1 Cell 178, 1088–1101.e1–e7, August 22, 2019


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
Purified anti-mouse CD62L (Mel-14) Bio-X-Cell Cat# BE0021; RRID:AB_1107665
Purified anti-mouse CD8 (2.43) Bio-X-Cell Cat# BE0061; RRID:AB_1125541
Purified anti-mouse CD4 (GK1.5) Bio-X-Cell Cat# BE0003-1; RRID:AB_1107636
Bacterial and Virus Strains
Yersinia pseudotuberculosis (32777) Laboratory of Y.B. N/A
Influenza A Virus, A/Puerto Rico/8/1934 (PR8), Laboratory of H.D.H, originally N/A
strain H1N1 from Mount Sinai School
of Medicine
Recombinant vaccinia-virus expressing Laboratory of N.P.R N/A
human gp100
Chemicals, Peptides, and Recombinant Proteins
2-Mercaptoethanol Sigma-Aldrich M3148-25ML
Brefeldin A (GolgiPlug) BD Biosciences 555029
BSA Sigma-Aldrich A3059-500G
DAPI Sigma-Aldrich D9542
DNase I Sigma-Aldrich DN25-5G
DMSO Sigma-Aldrich D2650
EDTA (0.5M) Corning 46-034-Cl
FBS Hyclone SH30070.03
L-Glutamine Corning 25-005-Cl
HEPES Corning 25-060-Cl
Ionomycin Sigma-Aldrich I0634-5MG
Bodipy FL C16 Life Technologies D-3821
BrdU Thermo Fisher or BD B23151 or 550891
Biosciences
Liberase TL Roche 5401020001
Sodium Pyruvate (100X) Corning 25-000-Cl
SIINFEKL peptide Genscript N/A
Triton X Sigma-Aldrich T9284
Liberase TL Roche 05401020001
MEM Non-essential Amino Acids (100X) Corning 25-025-Cl
Paraformaldehyde Electron Microscopy 15714-S
Sciences
Pennicillin-Streptomycin (100X) Corning 30-002-Cl
Phorbol 12-myristate 13-acetate (PMA) Sigma-Aldrich P8139-10MG
ProLong Gold Antifade Mountant Molecular Probes P36930
RPMI 1640 medium Corning 10-040-CV
Methanol JT Baker 9093-03
Bodipy FL-C16 Life Technologies D-3821
LipidTOX Life Technologies H34476
DNase I Sigma DN25-5G
Liberase TL Roche 5401020001
Trizol Thermo Fisher 15596018
OCT Fisher 23-730-571
Recombinant human IL-2 Biolegend 589106
Recombinant human IL-7 Biolegend 581906
Dexamethasone Sigma-Aldrich D1756-100MG
Tamoxifen Sigma-Aldrich T5648-5G
Corn oil Sigma-Aldrich C8267-500ML
(Continued on next page)

Cell 178, 1088–1101.e1–e7, August 22, 2019 e2


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
AMD3100 Sigma-Aldrich 239820-5MG
FTY720 Sigma-Aldrich SML0700-25MG
Glutamine Sigma-Aldrich 1294808-100MG
Glucose Sigma-Aldrich G8270-100G
Oligomycin Sigma-Aldrich 75351-5MG
FCCP Sigma-Aldrich C2920-10MG
Rotenone Sigma-Aldrich R8875-5G
Antimycin A Sigma-Aldrich A8674-100MG
Fatty-acid free BSA Sigma-Aldrich A8806
2X YT medium Sigma-Aldrich Y1003
Regular mouse chow Lab Diet Advanced 5V75
Protocol
Control isocaloric diet Envigo TD.160179
Adequate vitamins and minerals diet Envigo TD.190106
Adequate essential amino acids diet Envigo TD.190107
Adequate protein diet Envigo TD.190108
Diet in which only calories at 50% Envigo TD.190105
Critical Commercial Assays
Anti-PE microbeads Miltenyi Biotec 130-048-801
Arcturus PicoPure RNA Isolation kit Thermo Fisher KIT0204
FITC BrdU Flow kit BD Biosciences 557891
BD Cytofix/Cytoperm BD Biosciences 554722
BD Perm/Wash BD Biosciences 554723
Foxp3 / Transcription Factor Staining Buffer Set eBioscience 00-5523-00
High Sensitivity D1000 ScreenTape Agilent 5067-5584
LIVE/DEAD Fixable Blue Dead Cell Staining Kit Life Technologies L23105
MACS Cell Separation Column LS Miltenyi Biotec 130-042-401
Omniscript RT kit QIAGEN 205111
Taq PCR Master Mix Life Technologies K0171
Mouse corticosterone ELISA Thermofisher EIACORT
Nextera XT DNA Library Prep Kit Illumina FC-131-1096
SMARTer Ultra Low Input RNA Kit Clonetech 634946
Deposited Data
Raw RNA-Seq data This manuscript GEO: GSE124063
Experimental Models: Cell Lines
B16 melanoma The laboratory of N.P.R. N/A
Experimental Models: Organisms/Strains
Mouse: C57BL/6 Taconic Farms Mouse strain: C57BL/6NTac
Mouse: B6.SJL-Ptprca/BoyAiTac (CD45.1) Taconic Farms – NIAID exchange Mouse strain: Tac 8478
/
Mouse: OT-I CD45.1 RAG Taconic Farms – NIAID exchange Mouse strain: Tac 300
Mouse: C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT-I) Jackson laboratories Mouse strain: Jax, 003831
Mouse: B6.129(Cg)-Gt(ROSA)26Sortm4 Jackson laboratories Mouse strain: Jax, 007676
(ACTB-tdTomato,-EGFP)Luo/J (mTomato)
Mouse: mTomato x OT-I Laboratory of D.B.M N/A
Mouse: C57BL/6-Tg(Adipoq-cre/ERT2)1Soff/J Jackson laboratories Mouse strain: Jax, 025124
Mouse: B6.129P2-Gt(ROSA)26Sortm1(DTA)Lky/J Jackson laboratories Mouse strain: Jax, 009669
Mouse: Adipoq-CreERT2 x Rosa26-DTA Laboratory of Y.B. N/A
Mouse: Cxcr4fl/fl x UBC-CREERT2 Laboratory of S.R.S N/A
(Continued on next page)

e3 Cell 178, 1088–1101.e1–e7, August 22, 2019


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER
fl/fl ERT2
Mouse: S1pr1 x UBC-CRE Laboratory of S.R.S N/A
Mouse: B6Cg.Thy1a/Cy Tg(TcraTcrb)8Rest/J Jackson laboratories Mouse strain: Jax, 005023
Oligonucleotides
Hprt Thermofisher 4331182 (Mm03024075_m1)
Fabp4 Thermofisher 4331182 (Mm00445878_m1)
Software and Algorithms
Flowjo software Treestar RRID: SCR_008520
Imaris software Bitplane RRID: SCR_007370
Prism software Graphpad RRID: SCR_002798
ImageJ https://imagej.net/ RRID: SCR_003070
STAR aligner Dobin et al., 2013 RRID: SCR_015899
HOMER http://homer.ucsd.edu/ RRID: SCR_010881
DESeq2 https://bioconductor.org/packages/ RRID: SCR_015687
release/bioc/html/DESeq2.html
Metascape http://metascape.org/gp/index.html#/ RRID: SCR_016620
main/step1
GSEA http://www.broadinstitute.org/gsea/ RRID: SCR_003199
FastQC Babraham Bioinformatics RRID: SCR_014583
R https://www.r-project.org/ N/A

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Yasmine
Belkaid (ybelkaid@niaid.nih.gov).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice
C57BL/6NTac mice were purchased from Taconic Farms. B6.SJL-Ptprca/BoyAiTac (CD45.1), OT-I CD45.1 RAG / (OT-I) were
obtained through the NIAID Taconic exchange program. C57BL/6-Tg(TcraTcrb)1100Mjb/J (OT-I) and B6.129(Cg)-Gt(ROSA)26Sort-
m4(ACTB-tdTomato,-EGFP)Luo/J (mTomato) mice were purchased from the Jackson Laboratories and crossed to generate
mTomato OT-I mice. C57BL/6-Tg(Adipoq-cre/ERT2)1Soff/J and B6.129P2-Gt(ROSA)26Sortm1(DTA)Lky/J were purchased from Jack-
son laboratories and crossed to generate Adipoq-CreERT2 x Rosa26-DTA mice. B6Cg.Thy1a/Cy Tg(TcraTcrb)8Rest/J (pmel) were
from the laboratory of N.P.R. All experiments involved female mice 6–23 weeks of age. Mice were bred and maintained under specific
pathogen-free conditions at an American Association for the Accreditation of Laboratory Animal Care (AAALAC)-accredited animal
facility at the NIAID and housed in accordance with the procedures outlined in the Guide for the Care and Use of Laboratory Animals.
Cxcr4fl/fl x UBC-CREERT2 and S1pr1fl/fl x UBC-CREERT2 mice were housed at the Skirball institute animal facility. Experiments were
performed under an animal study proposal (LISB-18E, 19E and 20E) approved by the NIAID Animal Care and Use Committee.

Cell lines
B16 melanoma cells (National Cancer Institute tumor repository) were cultured at 37 C, 5% CO2 in complete media DMEM (GIBCO)
with 10% FBS, 1% glutamine and 1% penicillin-streptomycin.

Bacteria and virus strains


Wild-type or mutant Yersinia pseudotuberculosis (32777) strains were grown from bacterial culture in 2X YT media at 25 degrees
overnight. Mice were infected with 1 3 107 colony forming units (CFU) of Yptb DyopM via oral gavage (mice were fasted for
12 h prior). For secondary infections, mice were injected intravenously with 2 3 102 CFU WT Yptb and bacterial burden was as-
sessed in the spleen 48 h later, or for an oral challenge, mice were gavaged with 5 3 109 CFU WT Yptb and bacterial burden was
assessed in the mesenteric lymph node after 72 h. Bacterial burden was determined by serial plating on MacConkey plates and
colonies were counted after incubation at room temperature (RT) for 48 h.

Cell 178, 1088–1101.e1–e7, August 22, 2019 e4


For influenza infections, A/Puerto Rico/8/1934 (PR8) (originally obtained from Mount Sinai School of Medicine, strain H1N1) was
grown in allantoic cavities of 10-day-old specific pathogen free (SPF) embryonated chicken eggs. Allantoic fluid was collected 48 h
post infection and purified by differential centrifugation. Viral titers were determined by 50% tissue culture infective dose (TCID50) on
MDCK cells using the Reed-Muench method. Mice were anesthetized using isoflurane and intra-nasally infected with PR8 at 5
TCID50 in 25 mL sterile saline solution supplemented with 0.1% BSA.
For experiments involving recombinant vaccinia virus expressing human-gp100 followed by a B16 challenge, mice initially received
3 3 105 naive enriched pmel cells then were vaccinated intravenously with 1 3 107 plaque-forming units of this virus. After one month,
mice were randomly assigned to undergo DR for the remainder of the experiment or to remain being fed ad libitum. Following 3 weeks
of DR, mice were inoculated subcutaneously with 7.5 3 105 B16 cells and tumor burden was monitored over time. Mice were eutha-
nized when tumors reached a size of 15 mm x 15 mm.

METHOD DETAILS

Dietary restriction
The daily intake of regular chow (LabDiet Advanced Protocol PicoLab Verified – 75 IF; 20% protein, 5% fat) by individually caged mice
was determined by giving a defined amount of food and weighing the remainder daily for 2 weeks. From this, we found that individual
mice consumed approximately 2.75 g of food per day, consistent with previous reports (Acosta-Rodrı́guez et al., 2017). This equates
to an intake of roughly 11.4 kcal per day. As such, we provided 1.375 g of food to mice daily (5.7 kcal per day) to ensure 50% DR. Mice
on DR consumed all the food provided. Diets in which specific factors were normalized were designed in consultation with a nutri-
tionist (Envigo). These included the CR diet (only calories at 50%), in which everything (macromolecules, V & M, EAA) except for
calories was doubled. Thus, when given at 50% restriction everything is normalized except calories. Normalized V & M diet, in which
V & M were doubled. Thus, vitamins and minerals are normalized when given at 50% restriction. Normalized EAA diet, in which EAA
were doubled. Thus, EAA are normalized when given at 50% restriction. Normalized total protein diet, in which total protein was
doubled. Thus, total protein levels are normalized when given at 50% restriction.

Surgery
For experiments involving parabiotic surgery, mice were put on DR 1 week prior to surgery and then maintained on it for a further
3 weeks while conjoined. For the surgery, mice were anaesthetized with ketamine and xylazine and longitudinal incisions were
made from the elbow to the knee joint of each mouse, which were then connected at the joints. The skin of the two animals was
then sutured together. Adrenalectomy surgery was performed by Jackson laboratories.

Tissue processing and flow cytometry


Mice were euthanized with CO2, perfused via the left ventricle of the heart with 10 mL PBS before tissues were harvested and placed
in cold PBS. Bone marrow was flushed using cold PBS and processed through a 70 mM filter. Spleen and lymph nodes were pro-
cessed through a 70 mM filter. Blood was collected in heparinized tubes. White adipose tissue was chopped in digestion media con-
taining DMEM with 1% low fatty acid BSA, 50 mM HEPES, Liberase TL (0.05 mg/mL) (Roche) and DNase I (0.25 mg/mL) (Sigma) and
incubated for 25 min at 37 C with agitation. The tissue was then filtered through a 70 mm filter. Cells were stained for flow cytometry
for 15–30 min in PBS on ice. Antibodies for flow cytometry were purchased from eBioscience, BD Pharmingen or Biolegend and were
conjugated to Pacific Blue, BV605, BV785, BV510, BV650, eFluor 450, APC, AlexaFluor 647, Pe-Texas Red, PE-CF594, FITC, PE,
PerCPCy5.5, PECy7, APCCy7, APCe780, PerCPe710, BV421 or Alexa Flour 700. DAPI or Live/Dead fixable stain (Life Technologies)
was used to identify dead cells in all experiments. Intracellular staining for transcription factors and cytokine producing cells was per-
formed using BD intracellular staining kit or the eBiosciences FOXp3 staining kit. To determine cytokine production, cells were stim-
ulated with phorbol myristate acetate (PMA) (50 ng/mL), ionomycin (1 mM), and BD GolgiPlug (BFA) (1 mL/mL) for 2.5 h at 37 C before
staining for flow cytometry. The following antibodies were used for staining of murine cells: CD4 (RM4-5), CD8b (H35-17.2), CD25
(PC61.5) T-BET (4B10), FEXP3 (FJK-16S), EOMES (Dan11mag), CXCR4 (2B11), TER-119 (TER-119), CD71 (R17217), BCL-2
(10C4), IFN-g (XMG1.2), CD45 (30-F11), CD45.1 (A20), CD45.2 (104), CD69 (H1.2F3), CD127 (A7R34), CD90.2 (30-H12), TCRb
(H57-597), CD62L (MEL-14), CD44 (IM7), NK1.1 (PK136), gdTCR (GL3), B220 (RA3-6B2), CD122 (TMb1). Phosphostaining was per-
formed using either the eBiosciences kit (for AKT473 and mTOR2448 - eBiosciences) or by methanol fixation (for pS6240/244 - Cell
signaling technologies). For the latter, cells were first stained with surface antibodies, washed then treated with 4% PFA at RT for
10 min in PBS. Cells were then incubated with methanol at 20 C for 1 h or at 4 C overnight, followed by phospho-staining for
1 h on ice in the dark. BrdU staining was performed according to the manufacturers protocol (FITC BrdU flow kit - BD Biosciences).
Flow cytometric data were acquired on an LSR II or LSR Fortessa (BD Biosciences) and analyzed using FlowJo software (Tree Star).

Confocal microscopy
To image whole-mounted bone marrow tissue, one side of the bone (femur or tibia) was cut open, fixed for 8-16 h in PLP-buffer at 4 C
and left in 30% sucrose for another 10-16 h. Fixed bones were embedded in optical cutting temperature compound (OCT) over dry
ice and stored at 80 C. For staining, bones were trimmed 60-100 mm on a cryostat and collected by melting the OCT. Exposed
bones were fixed again in 4% PFA for 30 min, blocked for 1 h at RT with 1% BSA, 0.25% Triton X-100, 1% Fc-Block. If rabbit

e5 Cell 178, 1088–1101.e1–e7, August 22, 2019


anti-PERILIPIN-1 antibody (Abcam) was used, 10% normal goat serum was added in the blocking buffer. Antibodies (TER-119, CD8,
CD4 are the same as used for flow cytometry) or nuclei stain (DAPI) were diluted in blocking buffer and tissues were stained for 1 h at
RT, washed 3x in PBS and then imaged directly, or if PERILIPIN-1 was used, bones were incubated for 1 h at RT with the secondary
antibody AF647 goat anti-rabbit IgG. Bones were imaged in a PBS filled 4-well chambered cover glass using an upright inverted
confocal microscope from Leica (Leica TCS SP8). Imaris software (Bitplane) was used for all confocal image processing.

In vitro activation of CD8+ T cells and adoptive transfer


Spleen and lymph nodes from wild type and knock out mice were processed through a 70 mm filter, stained with an anti-CD8b
antibody conjugated to PE then positively enriched using anti-PE microbeads (Miltenyi) and a magnetic column (Miltenyi). Cells
were cultured in a 24-well plate coated with 5 mg anti-CD3 and anti-CD28 antibodies in complete media with 10 ng/mL IL-2. Cells
were split on days 2 and 3, with 10 ng/mL IL-7 added for days 4 and 5 of culture. OT-I cells were isolated from spleen of OT-I
mice (CD45.1 or mTomato+) background and activated with SIINFEKL peptide. 2–5 million cells were transferred intravenously via
the retroorbital vein.

In vivo treatments
Dexamethasone (Sigma) was administered to mice fed ad libitum at a dose of 2 mg/kg i.p. every day for 2 weeks. For inducible gene
deletion, 75 mg/kg tamoxifen (Sigma) in 90% corn oil: 10% ethanol for 5 consecutive days i.p. Experiments were performed 3 weeks
after the final treatment. FTY720 (Sigma) was administered i.p. four times every other day at a dose of 1 mg/kg, with or without 0.5 mg
of an anti-CD62L antibody (Mel-14, BioXcell). For AMD3100 (Sigma) experiments, mice were put on DR for 3 weeks then injected with
10 mg/kg subcutaneously for 4 consecutive days. 1 mg/kg BrdU (BD) was administered i.p. every day for 7 days prior to the exper-
iment. Anti-CD4 (GK1.5, BioXcell) and anti-CD8 (2.34, BioXcell) was administered i.p. at a dose of 0.2 mg for three consecutive days
prior to infection.

Ex vivo assays
Bodipy-C16 uptake assays were performed as previously described (Pan et al., 2017). LipidTOX (Thermofisher) staining was
performed on BM cells after surface staining according the manufactures instructions.

RNA extraction, cDNA synthesis and Quantitative PCR


Whole bone marrow from femur or tibia was flushed in Trizol (Thermofisher). Total RNA extraction was performed with RNeasy mini kit
(QIAGEN). DNase-treated total RNA (100 ng) was reverse transcribed into cDNA with Omniscript RT kit (QIAGEN) following
manufacturer’s instructions. qPCR was performed using the Taq PCR Master Mix (Life Technologies) on a Quantstudio 6 Flex
Real-Time PCR (Life Technologies).

RNA Sequencing
CD8+ TCM were sorted from the bone marrow and spleen of mice fed ad libitum or on DR. RNA was extracted using an Arcturus
PicoPure RNA Isolation Kit (Applied Biosystems), as per manufacturer’s instructions. cDNA was synthesized with the Ultralow V4
kit (Clonetech) and sequencing libraries were subsequently prepared with the Nextera XT DNA prep kit (Illumina). RNA-seq reads
were mapped with STAR (Dobin et al., 2013) to the mm10 reference genome with default parameters. Read counts were assessed
with HOMER’s (Heinz et al., 2010) analyzeRepeats with the option rna and parameters -noadj -condenseGenes and -count exons
for three replicates per condition. Differential gene expression was assessed with DESeq2 (Love et al., 2014) using HOMER’s
getDiffExpression.pl using a false discovery rate (FDR) of 5% and a fold change of 1.5. Metascape (Tripathi et al., 2015) was used
for analyzing enriched GO terms, and GSEA was used for gene set enrichment analysis. PCA was performed in R with prcomp,
center = TRUE, scale. = TRUE. GEO accession number is GSE124063.

ELISA
Mouse corticosterone ELISA was performed according to the instructions of the manufacturer (Thermofisher). To obtain bone
marrow extracellular fluid, femurs were flushed in approximately 2 mL of PBS, centrifuged and the supernatant was taken.

2-photon imaging and analysis


Mice fed ad libitum or on DR for 3 weeks received 20 3 106 mTomato+ OT-I cells i.v. and were imaged 1 week later. To image, mice
were anesthetized with ketamine (85 mg/kg), xylazine (13 mg/kg), and acepromazine (2 mg/kg) in PBS and maintained at a core
temperature of 37 C. 20 mg of an anti-CD31 AF647 antibody (clone EMC13.3, Biolegend) was injected i.v. to visualize vessels
immediately before imaging. Two-photon imaging through a surgically thinned skull was adapted from methods previously
described (Manglani and McGavern, 2018). Briefly, a metal bracket was glued to the skull such that the imaging window contained
the frontal bone sutures. The skull was thinned to leave 25 mm of bone above the bone marrow. 3D time lapses were captured
using a Leica SP8 two-photon microscope equipped with an 8,000-Hz resonant scanner, a 25 3 color corrected water-dipping
objective (1.0 NA), a quad HyD external detector array, a Mai Tai HP DeepSee Laser (Spectra-Physics) tuned to 905nm and an
Insight DS laser (Spectra-Physics) tuned to 1050nm. The following dichroic mirrors were used for imaging studies: 495nm-LP,

Cell 178, 1088–1101.e1–e7, August 22, 2019 e6


562nm-LP, and 624-LP. Z-stacks were acquired at 8.5 sec intervals and consisted of 7 optical slices (3.5mm step size). A second
harmonic signal was used to visualize skull bone. Cell speed and track displacement length of mTomato OT-I cells were obtained
using Imaris software (Bitplane). ImageJ software was used to generate the supplemental movies.

Oxygen consumption flux evaluation


CD8+ CD44+ T cells were enriched via negative selection and sorted from secondary lymphoid organs or bone marrow of mice fed
ad libitum or on DR. Real-time oxygen-consumption rate (OCR) in T cells were determined with an XF-96e Extracellular Flux
Analyzer (Seahorse Bioscience); 3 3 105 cells/well were used for each condition. Cell were previously attached by using Cell
Tak (25 ug/mL). The assay was performed in XF DMEM pH = 7.4 supplemented with 2 mM glutamine, 25 mM glucose and
1mM pyruvate. Three consecutive measurements were performed under basal conditions and after the sequential addition of
the following ETC inhibitors: 1 mM oligomycin, 1 mM FCCP, 1 mM rotenone and 1 mM antimycin A. Basal respiration rate (BRR)
was defined as last rate measurement before the first injection. Maximal respiration rate (MRR) was defined as the maximal
rate measurement after FCCP injection. Spare respiration capacity (SRC) was defined as the difference between MRR and BRR.

QUANTIFICATION AND STATISTICAL ANALYSIS

Groups were compared with Prism V8 software (Graphpad) using a two-tailed unpaired Student’s t test or Log-rank (Mantel-Cox) test
where appropriate. Differences were considered to be statistically significant when p < 0.05. Data are expressed as mean, or mean ±
SD. Refer to figure legends for the specific statistical tests used in each experiment. Sample sizes were determined based on
previous experience with similar experiments.

DATA AND CODE AVAILABILITY

The datasets generated during this study can be found using the GEO accession number GEO: GSE124063.

e7 Cell 178, 1088–1101.e1–e7, August 22, 2019


Supplemental Figures

Figure S1. Memory T Cells Accumulate in BM during DR, Related to Figure 1


(A) Percentage weight change of mice on 50% DR for up to 6 weeks. (B) Weight (grams) of gonadal adipose tissue (GAT) in mice on DR. (C) Total hematopoietic
cellularity of spl, cervical LN (cLN) and inguinal LN (iLN) after 3-4 wk of DR. (D) Gating strategy of BM populations analyzed during DR. (E) Number of CD4+ CD44+

(legend continued on next page)


T cells in spl, cLN, blood, GAT, and BM over time during DR. (F) Number of regulatory T cells (G) NK1.1+ cells and (H) B220high B cells in GAT, spl and BM after 5-6
wk of DR. (I) Number of B220- CD138+ plasma cells in BM after 5-6 wk of DR. Each symbol represents an individual mouse, except A, which is the mean of 5 mice.
Data shows the mean, or mean ± SD, representative of at least 5 (A, D), or pooled from 2–3 (B, C, E–I) experiments with 3–5 mice per group. *p < 0.05, **p < 0.01,
***p < 0.001, ****p < 0.0001, ns; not significant. Two-tailed unpaired Student’s t test.
Figure S2. Phenotypic Analysis of Memory T Cells during DR, Related to Figure 2
(A) Number of CD44+ CX3CR1int. TPM in BM of mice fed ad libitum or on DR for 3 wk. (B–G) Expression of various surface markers and transcription factors in TCM
and TEM from BM and spl of mice fed ad libitum or on DR for 3–4 weeks. (H) OT-I cells were in vitro activated with peptide and transferred into hosts fed ad libitum
(legend continued on next page)
or on DR for 3 wk. Graphs show number of transferred OT-I cells recovered from BM and spl 1 wk after transfer. (I) BrdU uptake (over 1 wk) by memory T cells in
BM and spl of mice fed ad libitum, on DR for 3 wk, or on DR for 3 wk then re-fed ad libitum for 1 wk. Gated on CD8+ CD44+ cells. Each symbol represents an
individual mouse. Data shows the mean, pooled from 2–3 experiments (A, I) or representative of at least 2 (B–H) experiments. *p < 0.05, **p < 0.01, ***p < 0.001,
****p < 0.0001, ns; not significant. Two-tailed unpaired Student’s t test.
A B

C D

Figure S3. GC Drive Memory T Cells into BM during DR, Related to Figure 3
(A) BCL-2 expression by TEM in BM of mice fed ad libitum (blue) or on DR (red) for 3 wk. (B) Mice received sham surgery or an adrenalectomy (ADX) 1 wk prior to
being put on DR for 1 wk. Graphs show number of TEM in spl and BM. (C) BCL-2 expression in BM and (D) number of TEM in spl and BM of mice fed ad libitum that
were treated with veh (blue) or dex (red) daily for 2 wk. Each symbol represents an individual mouse. Data shows the mean, pooled from 2 experiments. *p < 0.05,
**p < 0.01, ***p < 0.001, ****p < 0.0001, ns; not significant. Two-tailed unpaired Student’s t test.
Figure S4. BM Remodeling during DR Promotes Memory T Cell Recruitment and Retention, Related to Figure 4
(A) Scatterplot showing the Il7 and Il15 genes in whole BM samples from mice fed ad libitum or on DR for 3 wk. (B) Cxcl12 gene expression in whole BM of mice fed
ad libitum or on DR for 3 wk, determined by qPCR. (C) CXCR4 surface expression by TEM in the BM and spl of mice fed ad libitum (blue) or on DR for 3 wk (red). (D)
Mice on DR for 3 wk were treated with AMD3100 subcutaneously for 4 consecutive days prior to analysis. Graph shows number of CD8+ TCM in BM. (E) Number of
RBC in BM of mice fed ad libitum treated with veh or dex daily for 2 wk. (F) Mice that had been on DR for 3 wk were treated with FTY720 4 times every other day. (G)
Fabp4 gene expression in whole BM samples from tamoxifen-treated Adipoq-CreERT2 x Rosa26-DTA mice fed ad libitum or on DR for 5–6 wk determined by
qPCR. (H) Uptake of Bodipy-C16 ex vivo by TCM from BM of mice fed ad libitum or on DR. (I) Intracellular lipid content of TCM from BM of mice fed ad libitum or on
DR. Each symbol represents an individual mouse, except for (A), in which each symbol represents an individual gene. Data shows the mean, representative of at
least 2 experiments (C, E) or pooled from 2–3 (A, B, D, F–I) experiments with 3–6 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns; not
significant. Two-tailed unpaired Student’s t test.
Figure S5. Memory T Cells Are in a State of Energy Conservation during DR, Related to Figure 5
(A) Principle component analysis of mRNA expression by TCM sorted from the spl and BM of mice fed ad libitum or on DR for 3 wk. Each symbol represents a
biological replicate pooled from 5 mice.
Figure S6. Similar Protection to a Secondary Infection When on DR throughout Primary and Secondary Infections, Related to Figure 6
(A) Mice on DR for 3 wk were infected with Yptb DyopM orally, then challenged 7 wk later i.v. with WT Yptb. Bacterial burden was assessed in the spl 2 days after
challenge. Each symbol represents an individual mouse. Data shows the mean, pooled from 2 experiments with 5 mice per group. ns; not significant. Two-tailed
unpaired Student’s t test.

You might also like