Abstract
Objective
Viral infection of cells leads to metabolic changes, but how viral
infection changes whole-body and tissue metabolism in vivo has not been
comprehensively studied. In particular, it is unknown how metabolism
might be differentially affected by an acute infection that the immune
system can successfully clear compared to a chronic persistent
infection.
Methods
Here we used metabolomics and isotope tracing to identify metabolic
changes in mice infected with acute or chronic forms of lymphocytic
choriomeningitis virus (LCMV) for three or eight days.
Results
Both types of infection alter metabolite levels in blood and tissues,
including itaconate and thymidine. However, we observed more dramatic
metabolite changes in the blood and tissues of mice with persisting
LCMV infection compared to those infected with the acute viral strain.
Isotope tracing revealed that the contribution of both glucose and
glutamine to the tricarboxylic acid (TCA) cycle increase in the spleen,
liver, and kidneys of mice infected with chronic LCMV, while acute LCMV
only increases the contribution of glutamine to the TCA cycle in the
spleen. We found that whole-body turnover of both glutamine and
thymidine increase during acute and chronic infection, whereas
whole-body glucose turnover surprisingly does not change. Activated T
cells in vitro produce thymidine and virus-specific T cells ex vivo
have increased thymidine levels, nominating T lymphocytes as the source
of thymidine in LCMV infection.
Conclusions
In sum, we provide comprehensive measurements of whole-body and tissue
metabolism in acute and chronic viral infection, and identify altered
thymidine metabolism as a marker of viral infection.
Keywords: Immunometabolism, Metabolomics, Isotope tracing, Whole-body
metabolism, Tissue metabolism
Highlights
* •
We performed metabolomics and isotope tracing in LCMV
virus-infected mice.
* •
Metabolites including thymidine and itaconate change in blood and
tissues.
* •
Chronic persisting infection changes metabolites more than acute
cleared infection.
* •
Turnover of glutamine and thymidine, but not glucose, increases in
infection.
* •
Activated T cells produce thymidine.
1. Introduction
Viral infection and the resultant host immune response cause major
physiologic changes. The virus replicates and spreads from cell to
cell, requiring RNA production (and DNA if a DNA virus) and protein
synthesis of the viral capsid and nonstructural proteins. Immune cells,
which account for around 2% of the mass of the human body in the
uninflamed state [[40]1], proliferate extensively-for example T cells
specific for the infecting virus can increase 1000-fold- and produce
effector proteins [[41]2,[42]3]. These phenomena suggest that the viral
infection and the host immune response may be supported by metabolic
changes, to produce energy and provide building blocks to enable
protein production and cell proliferation. However, how metabolism
changes during viral infection in vivo is incompletely understood.
One well-studied model of viral infection is the mouse RNA virus
lymphocytic choriomeningitis virus (LCMV). This virus infects most
tissues of the mouse, including the spleen, liver, and kidney [[43]4].
It induces a strong CD8+ T cell response, and the immune response
itself is responsible for the pathology and tissue damage during the
infection [[44]4]. Different strains of LCMV, distinguished by only a
few nucleotide changes, can result in dramatically different disease
outcomes. The host T cell response successfully clears the Armstrong
strain of LCMV, while the Clone 13 strain establishes a chronic
infection which the immune system cannot clear, resulting in immune
dysfunction and T cell exhaustion [[45]5]. Although an infection is not
considered truly chronic until day 15 or 30, mice infected with these
different strains diverge in viral load and in CD8+ T cell phenotypes
as early as day 8 [46][3], and as shorthand in this paper we will refer
to Armstrong as acute infection and Clone 13 as chronic infection.
Previous studies have demonstrated that LCMV infection of mice can
change serum and tissue metabolite levels, but there have been few
measurements of how metabolic flux may change during viral infection.
Both kynurenine [[47]6,[48]7] and itaconate [[49]8], two well-studied
immunometabolites [[50][9], [51][10], [52][11], [53][12], [54][13],
[55][14]], increase in the serum of LCMV infected mice. Tissue
metabolite levels have been less investigated, though one study showed
that itaconate and pyrimidine metabolites are elevated in the infected
liver [[56]8]. One study suggested that liver urea cycle flux may be
increased in LCMV infection [[57]15]. Another study from the same group
showed that mouse activity and whole-body oxygen consumption decrease
at the peak of infection (days 7–8) in chronic LCMV infection [[58]16].
These past studies are informative yet leave several key questions
unanswered. How do metabolite levels change during infection in tissues
other than liver? How does the use of specific nutrients change during
viral infection, both at the whole-body level and in specific tissues?
Physiologic changes like fasting-feeding [[59]17,[60]18], feeding a
ketogenic diet [[61]19], and cold exposure [[62]20] cause dramatic
changes to the use of nutrients like glucose, lactate, and glutamine,
but it is unknown what may happen during viral infection and immune
response. Finally, how does the metabolic response differ between acute
and chronic viral infection? Since these two types of infection vary
dramatically in viral kinetics, as well as the character of T cells
responding, we might expect resulting metabolic differences.
Here, we used infusion of stable-isotope-labeled metabolites to measure
whole body and tissue-specific nutrient use during early (up to day 8)
acute Armstrong and chronic Clone 13 LCMV infection, as well as
metabolomics to measure serum and tissue metabolites. Both acute and
chronic infection alter blood and tissue metabolite levels, but chronic
infection changes more metabolites to a greater extent. We observed
increased whole-body turnover of both glutamine and thymidine, a
pyrimidine metabolite used for nucleotide salvage production, in acute
and chronic infection. Chronic infection caused more dramatic changes
in nutrient use in highly-infected tissues: glucose and glutamine
contribution to the TCA cycle both increase in spleen, liver, and
kidney in chronic LCMV infection, while only spleen glutamine use
increases in acute infection. Finally, we propose that thymidine
production in LCMV infection may be mediated by T cells, since
thymocytes, activated T cells in culture, T-acute lymphoblastic
leukemia, and virus-specific T cells ex vivo all display high thymidine
levels. Overall, this work provides an atlas of whole-body and tissue
metabolite and metabolic flux changes in acute and chronic LCMV
infection, and nominates thymidine levels and turnover as a biomarker
of viral infection.
2. Results
2.1. Chronic LCMV infection changes serum metabolite levels more dramatically
than acute infection
We set out to measure how acute and chronic lymphocytic
choriomeningitis virus (LCMV) infections in mice change whole-body and
tissue metabolism. Conceptually, performing metabolomics in LCMV can
nominate pathways with altered metabolic flux, and these fluxes can
then be measured using stable-isotope metabolite infusion. We first
examined changes in serum metabolite levels in acute and chronic LCMV
infection. Since the blood delivers metabolites to all tissues and
removes metabolites that tissues release, measuring blood can give an
overview of whole-body metabolic changes.
We measured blood metabolite levels in mice on the third day of LCMV
infection (the peak of viremia in acute infection [[63]3]), mice on the
eighth day of infection (the peak of T cell numbers in chronic and
acute infection), and in uninfected mice. Both viral strains alter many
serum metabolites ([64]Figure 1A–B, Supplementary Tables 1 and 2);
pyrimidine metabolites increase ([65]Figure 1C–D) consistent with a
previous report [[66]8], while purine metabolites tend to decrease
([67]Fig. S1A–B). Two notable changed metabolites were itaconate and
thymidine; these are increased by both types of viral infections
([68]Figure 1E–F), but more greatly by chronic Clone 13 infection.
Itaconate is an immunometabolite produced by activated macrophages
[[69]9,[70]10] which affects macrophage function by a number of
proposed mechanisms including inducing type I interferon [[71]21] and
reducing IL-6 production [[72]22]. Deoxypyrimidines involved in DNA
synthesis including thymidine, deoxycytidine, and deoxyuridine are
increased in both types of infection, while most other nucleoside
levels are unchanged ([73]Fig. S1C–D).
Figure 1.
[74]Figure 1
[75]Open in a new tab
Chronic LCMV infection changes serum metabolite levels more
dramatically than acute infection. (A–B) Serum metabolite changes in
day 8 acute Armstrong strain LCMV infection (A) or chronic Clone 13
strain (B) compared to uninfected; n = 4 uninfected for each, n = 6 for
Armstrong infected, n = 5 for Clone 13. Two-tailed t-tests in (A–B)
performed on log[2] transformed metabolite intensities. (C–D) Metabolic
pathways enriched among significantly increased metabolites in serum of
(C) Armstrong infected mice or (D) Clone 13 infected mice compared to
uninfected. (E–F) Serum metabolite levels of itaconate (E) and
thymidine (F) in LCMV infected mice. Includes multiple experiments, all
infected sera normalized to uninfected sera from the same experiment; n
= 27 uninfected, n = 6 acute day 3, n = 6 chronic day 3, n = 18 acute
day 8, n = 11 chronic day 8; two-tailed t-tests in performed on log[2]
fold changes from uninfected. (G–H) All (G) and selected (H) serum
metabolite changes in chronic versus acute day 8 infection, ratios of
fold changes compared to uninfected shown. (I) In which strain of LCMV
did serum metabolites exhibit a greater absolute-value fold-change
compared to uninfected. For (G–I), n = 18 acute day 8, n = 8 chronic
day 8, two-tailed t-tests done comparing log[2] fold changes from
uninfected.
We next directly compared blood metabolite changes observed on day 8 of
acute infection to day 8 of chronic infection, to test whether blood
metabolite levels could distinguish these two types of infection. Since
we did not conduct infections with both strains in the same experiment,
we normalized each metabolite to its level in uninfected mice in the
same experiment, then compared metabolite fold changes across three
independent experiments of acute infection and two of chronic
infection. We found that a number of metabolites increase more in
chronic infection, including kynurenine [[76]6], itaconate, and
thymidine, while both uridine and inosine decrease more in chronic than
acute infection ([77]Figure 1G–H). In general, chronic LCMV infection
causes greater metabolite increases and decreases (i.e. greater
absolute fold change compared to uninfected, [78]Figure 1I). Of all
metabolites significantly different between acute and chronic
infection, 89% are changed to a greater extent in chronic infection.
2.2. Acute LCMV alters tissue metabolites
We next measured which metabolites are altered in tissues on day 8 of
acute LCMV infection. The spleen, liver, and kidney are all sites of
LCMV infection [[79]4], while the spleen also serves as a main site for
mounting the adaptive immune response to LCMV, so we expected that
metabolite levels might be changed by infection in these tissues. Using
principal component analysis, infected spleens, livers, and kidneys
could be separated from their uninfected counterparts, although
unsurprisingly, tissue identity causes more dramatic metabolite
differences than infection status ([80]Figure 2A). Similarly,
hierarchical clustering of these tissue metabolomics data showed that
each tissue clusters separately from other tissues, yet infected and
uninfected samples of each tissue are intermixed ([81]Figure 2B), and
do not show major heatmap differences. However, a small set of
metabolites in each tissue change dramatically with acute LCMV
infection ([82]Figure 2C–F). These changes include increased itaconate
in spleen, liver, kidney, and intestine and increased thymidine in the
liver. Overall, upregulated tissue metabolites are enriched for
pyrimidine metabolites and phenylalanine metabolites ([83]Fig. S2A–C),
while purine metabolites tend to be downregulated ([84]Fig. S2D–F).
Figure 2.
[85]Figure 2
[86]Open in a new tab
Acute LCMV infection alters tissue metabolites. (A) Principal component
analysis of metabolite levels in spleen, liver, kidney, and intestine
of mice on day 8 of acute Armstrong strain LCMV infection or uninfected
mice. Metabolites are log transformed, row-mean normalized, and
standard deviation set to 1. (B) Heatmap with hierarchical clustering
of metabolite levels, same data as (A). Metabolites are log transformed
and row-mean normalized. (C–F) Tissue metabolites in (C) spleen, (D)
liver, (E) kidney, (F) small intestine of mice on day 8 of acute LCMV
infection compared to uninfected. Selected metabolites significantly
changed from uninfected tissue are labeled. In all panels, n = 4 mice
uninfected, n = 6 infected. Two-tailed t-tests in (C–F) performed on
log[2] transformed metabolite intensities.
2.3. Chronic LCMV infection changes tissue metabolites more than acute
infection
We then measured tissue metabolites on day 8 of chronic LCMV infection.
A number of metabolites change in the spleen and liver, including
increased itaconate and kynurenine in the spleen ([87]Figure 3A), and
increased itaconate and thymidine in the liver ([88]Figure 3B).
However, unlike in the blood, tissue thymidine and itaconate change
similarly in acute and chronic LCMV infection ([89]Figure 3C–F).
Pathway enrichment analysis showed that many pyrimidine metabolites and
glycerophospholipid metabolites are both up- and down-regulated in
chronic-LCMV-infected tissues ([90]Fig. S3A–D).
Figure 3.
[91]Figure 3
[92]Open in a new tab
Chronic LCMV infection changes tissue metabolites more than acute
infection. Tissue metabolites in (A) spleen or (B) liver of mice on day
8 of chronic infection compared to uninfected; n = 4 uninfected, n = 5
infected. Two-tailed t-tests in (A–B) performed on log[2] transformed
metabolite intensities. (C–F) Levels of (cC spleen or (D) liver
thymidine, and (E) spleen and (F) liver itaconate in uninfected mice or
mice on day 8 of LCMV infection. n = 8 uninfected, n = 6 acute LCMV, n
= 5 chronic LCMV; infected values normalized to uninfected mice from
the same experiment. (G–H) Spleen (G) and liver (H) metabolite changes
from uninfected on day 8 of chronic or acute infection. (I–J) Selected
spleen (I) and liver (J) metabolite changes on day 8 of acute or
chronic infection, relative to uninfected serum. (K) In which strain of
LCMV did spleen or liver metabolites exhibit a greater absolute-value
fold-change compared to uninfected tissues. For (G–K), n = 15 acute day
8, n = 8 chronic day 8. All two-tailed t-tests performed on log[2] fold
changes from uninfected.
We asked whether 8 days of chronic Clone 13 infection results in any
distinct metabolic changes from acute Armstrong infection. Though
itaconate and thymidine are not changed between the infection states,
certain other tissue metabolites are significantly different. Examples
include splenic sulfoglycolithocholate (a bile salt) and liver
deoxyinosine, both of which increase more in chronic than acute
infection ([93]Figure 3G–J). As in the blood, tissue metabolites
display more extreme changes in chronic than acute infection
([94]Figure 3K). Around 76% of altered spleen and liver metabolites
have a greater absolute-value fold-change in chronic infection.
2.4. LCMV infection increases glutamine and thymidine whole-body turnover
We next measured whole-body turnover of specific metabolites in acute
and chronic LCMV infection. Glucose, glutamine, and lactate are three
of the nutrients with the highest whole-body turnover and are some of
the dominant substrates for ATP generation in tissues
[[95]17,[96]19,[97]23,[98]24]. We hypothesized that viral infection and
immune response might change tissue energy and nutrient requirements,
and thus might change turnover of these nutrients. Indeed, we had
observed that blood glucose level decreases on day 8 of chronic viral
infection ([99]Figure 1B, 0.65x of levels in uninfected mice), and a
previous report suggested that carbohydrate oxidation is reduced in
chronic LCMV infection [[100]16]. Therefore we asked whether turnover
of these nutrients might be altered in chronic or acute viral
infection.
To measure nutrient turnover, we infused carbon-13- or
nitrogen-15-labeled nutrients one at a time intravenously in infected
or uninfected mice for 2.5 h, and sampled blood from the submandibular
vein ([101]Figure 4A). Measuring the blood enrichment of the
fully-labeled form of the nutrient infused allows calculation of that
nutrient’s turnover (also known as rate of appearance or F[circ]) using
the equation
[MATH: Ra=(RL)−R :MATH]
[1]
where R is rate of labeled nutrient infusion in nanomoles per minute
per gram body weight, and L is blood enrichment of fully-labeled (i.e.
with same number of labeled atoms as original infusate) nutrient
[[102]17,[103]25,[104]26]. Surprisingly, we found that glucose turnover
does not change on day 3 or day 8 of acute or chronic infection
([105]Figure 4B). This observation underlines the importance of
measuring nutrient turnover directly, rather than using nutrient level
as a proxy for metabolic flux. Glutamine turnover increases
significantly on day 3 of chronic infection and day 8 of acute
infection ([106]Figure 4C, 1.4x and 1.3x of uninfected turnover
respectively). Lactate turnover does not change in infected compared to
uninfected mice ([107]Figure 4D). Therefore, turnover of major
carbohydrate nutrients does not change on day 3 or 8 of LCMV infection
compared to uninfected mice, but glutamine turnover increases.
Figure 4.
[108]Figure 4
[109]Open in a new tab
LCMV infection increases glutamine and thymidine whole-body turnover.
(A) Schematic of whole-body metabolite turnover measurement using
isotope-labeled metabolite infusion. (B) Whole-body turnover of glucose
determined by [U–^13C] glucose infusion. (5 experiments pooled, overall
n = 19 uninfected, n = 5 acute LCMV day 3, n = 3 chronic LCMV day 3, n
= 11 acute LCMV day 8, n = 11 chronic LCMV day 8.) (C) Whole-body
turnover of glutamine determined by [U–^13C] glutamine infusion. (3
experiments pooled, overall n = 14 uninfected, n = 3 acute LCMV day 3,
n = 4 chronic LCMV day 3, n = 6 acute LCMV day 8, n = 6 chronic LCMV
day 8.) (D) Whole-body turnover of lactate determined by [U–^13C]
lactate infusion. (4 experiments pooled, overall n = 10 uninfected, n =
6 acute LCMV day 3, n = 3 chronic LCMV day 3, n = 6 acute LCMV day 8, n
= 3 chronic LCMV day 8.) (E) Whole-body turnover of thymidine
determined by [^15N[2]] thymidine infusion. (2 experiments pooled,
overall n = 8 uninfected, n = 4 acute LCMV day 3, n = 4 chronic LCMV
day 3, n = 4 acute LCMV day 8, n = 9 chronic LCMV day 8). All t-tests
are Student’s two-tailed t-tests between uninfected vs other groups,
acute d3 vs chronic d3, or acute d8 vs chronic d8; all tests not
significant if not shown.
Next, we measure turnover of thymidine, levels of which increase in the
blood and liver in acute and chronic infection ([110]Figure 1,
[111]Figure 3). Thymidine is not a major substrate for energy
production; rather, it is a breakdown product of deoxynucleotide
synthesis, which can be recycled back into deoxynucleotide production
using thymidine kinase or broken down and excreted [[112]27,[113]28].
We found that indeed thymidine turnover doubles on day 8 of both acute
and chronic infection ([114]Figure 4E, 1.9x and 2.2x increase relative
to uninfected turnover respectively). Therefore, both glutamine and
thymidine turnover increase in LCMV infection.
2.5. Glucose contribution to tissue TCA cycle increases in chronic LCMV
infection
Carbon-13 nutrient infusion can also reveal which blood nutrients serve
as a source for tissue metabolites. For example, the tricarboxylic acid
(TCA) cycle, which gives rise to the majority of ATP in all tissues
[[115]23], can be fueled by glucose, glutamine, lactate, fatty acids,
and other nutrients. Though whole-body turnover of glucose and
glutamine is largely unchanged by LCMV infection, we asked whether
viral infection might change nutrient contribution to the TCA cycle of
specific tissues.
By infusing carbon-13 glucose and measuring the carbon-13 labeling of
tissue malate ([116]Figure 5A), we found that glucose contribution to
TCA cycle metabolite increases on day 8 of chronic but not acute LCMV
infection in spleen, liver and kidney ([117]Figure 5B–D). The
contribution of glucose to liver TCA cycle also increases on day 3 of
chronic but not acute infection ([118]Figure 5C). Some portion of this
contribution from glucose is via circulating lactate converted from
glucose in other tissues [[119]17], but overall infection does not
increase the contribution from carbon-13 lactate itself ([120]Figure
S4-no change in spleen or kidney, lactate contribution decreases in
liver on day 8 of acute infection). Contribution of glucose to serine
and glycine in the spleen increases, similar to the increase in TCA
cycle contribution of glucose, on day 8 of chronic infection ([121]Fig.
S5). However, contribution of glucose to quadriceps muscle remains
unchanged on day 8 of chronic LCMV infection ([122]Figure 5E). Muscle
represents the most massive tissue type in the body, accounting for
about 40% of mouse body weight [[123]29]. Therefore it appears that
highly-infected tissues like spleen, liver and kidney increase their
glucose use, but since muscle (and perhaps other tissues) do not
increase glucose use, whole body glucose turnover is unchanged
([124]Figure 4).
Figure 5.
[125]Figure 5
[126]Open in a new tab
Glucose contribution to tissue TCA cycle increases in chronic LCMV
infection. (A) Schematic of measurement of nutrient contribution to TCA
metabolites using labeled nutrient infusion. (B) Average carbon
labeling of malate TCA intermediate in spleen from [U–^13C] glucose
infusion in LCMV-infected and uninfected mice. (n = 14 uninfected, n =
5 acute LCMV day 3, n = 3 chronic LCMV day 3, n = 6 acute LCMV day 8, n
= 8 chronic LCMV day 8.) (C) Average labeling of malate in liver from
[U–^13C] glucose infusion. (n = 14 uninfected, n = 5 acute LCMV day 3,
n = 3 chronic LCMV day 3, n = 6 acute LCMV day 8, n = 8 chronic LCMV
day 8.) (D) Average labeling of malate in kidney from [U–^13C] glucose
infusion. (n = 9 uninfected, n = 5 acute LCMV day 3, n = 3 chronic LCMV
day 3, n = 6 acute LCMV day 8, n = 3 chronic LCMV day 8.) (E) Average
labeling of malate in quadriceps from [U–^13C] glucose infusion. (n = 7
uninfected, n = 3 chronic LCMV day 3, n = 8 chronic LCMV day 8.) (F)
Average labeling of malate in spleen from [U–^13C] glutamine infusion.
(G) Average labeling of malate in liver from [U–^13C] glutamine
infusion. (H) Average labeling of malate in kidney from [U–^13C]
glutamine infusion. For f-h, n = 12 uninfected, n = 3 acute LCMV day 3,
n = 4 chronic LCMV day 3, n = 3 acute LCMV day 8, n = 6 chronic LCMV
day 8. (I) Average labeling of malate in quadriceps muscle from
[U–^13C] glutamine infusion; n = 2 uninfected, n = 3 chronic LCMV day
8. All t-tests are Student’s two-tailed t-tests between uninfected vs
other groups, acute d3 vs chronic d3, or acute d8 vs chronic d8; all
tests not significant if not shown.
Glutamine contribution to the TCA cycle, measured by carbon-13
glutamine infusion, increases in the spleen in all infection conditions
([127]Figure 5F), but with a significantly greater increase on day 8 of
chronic infection compared to day 8 of acute infection. Glutamine
contribution also increases on day 3 of chronic infection in the liver,
and days 3 and 8 of chronic infection in the kidney ([128]Figure 5G–I).
In sum, the contribution of glucose and glutamine to the TCA cycle
rises in infected tissues (spleen, liver, and kidney) in chronic
infection, while in acute infection the only contribution change
observed was increased glutamine use in the spleen. Therefore,
increased carbon-13 glucose contribution to the TCA cycle in infected
tissues distinguishes these two LCMV strains.
2.6. Thymidine synthesis increases in the spleen in LCMV infection
Since we observed that thymidine whole-body turnover increases in LCMV
infection, indicating more whole-body thymidine production and more
consumption/excretion ([129]Figure 6A) [[130]25,[131]26], we wondered
which tissue and cell type might be responsible for the increased
production of thymidine. We found that thymidine synthesis rate
increases in the spleen in LCMV infection, particularly in day 8
chronic infection ([132]Figure 6B), by measuring labeling of thymidine
from carbon-13 glucose infusion. (Note that because nucleotide
synthesis is relatively slow, not reaching steady state until around 12
hours of infusion [[133]30], a 2.5 hour infusion of carbon-13glucose
captures thymidine synthesis rate, unlike TCA cycle metabolites which
have reached steady-state labeling at this timepoint [[134]26].)
Synthesis of thymidine from glucose was not detected in liver, kidney,
or intestine, so its production may be specific to the spleen, a major
site of the adaptive immune response in LCMV.
Figure 6.
[135]Figure 6
[136]Open in a new tab
T lymphocytes may be a major source of thymidine. (A) Schematic of
thymidine metabolism pathway. (B) Labeling of spleen thymidine from
[U–^13C] glucose infusion. (5 experiments pooled, overall n = 13
uninfected, n = 5 acute LCMV day 3, n = 3 chronic LCMV day 3, n = 6
acute LCMV day 8, n = 7 chronic LCMV day 8.) T-tests are Student’s
two-tailed t-tests between uninfected vs other groups, acute d3 vs
chronic d3, or acute d8 vs chronic d8; all tests not significant if not
shown. (C) Fractional composition of CD45+ spleen cells in n = 6
uninfected, n = 6 day 8 acute, and n = 5 day 8 chronic LCMV infected
mice measured by flow cytometry. (D) Activation status (CD44+ fraction)
of transferred LCMV-specific P14 CD8 T cells in the spleen in n = 5
uninfected, n = 6 day 8 acute, and n = 5 day 8 chronic LCMV infected
mice measured by flow cytometry. (E) Thymidine levels in tissues of
healthy mice, n = 3 mice. (F) Serum thymidine in mice with NOTCH-1
induced T-cell acute lymphocytic leukemia compared to healthy controls,
n = 5 for each group. (G) Production of thymidine by mouse T cells upon
activation, n = 4 biological replicates per group. (H) Thymidine level
in CD8+ T cells, either CD8+ cells from uninfected mice or P14
virus-specific CD8+ T cells isolated from LCMV-infected mice on day 8,
n = 11 uninfected and chronic d8, n = 12 acute d8. All t-tests are
Student’s two-tailed t-tests, tests in (F–G) are performed on log[2]
transformed metabolite intensities.
2.7. T lymphocytes may be a major source of thymidine
We next asked which cells in the spleen might be responsible for
producing thymidine during LCMV infection. B and T lymphocytes compose
most of the spleen by cell number, both in uninfected and LCMV-infected
mice ([137]Figure 6C, not including red blood cells which don’t contain
nuclei or most metabolic pathways). However, the proportion of
activated and proliferating lymphocytes increases during LCMV infection
([138]Figure 6D), with associated changes in cellular metabolism, such
as an increase in the rate of DNA replication [[139][31], [140][32],
[141][33], [142][34]].
We hypothesize that activated lymphocytes are the likely cellular
source of increased thymidine during LCMV infection. In healthy mice,
the tissue exhibiting by far the highest level of thymidine is the
thymus ([143]Figure 6E), a tissue primarily composed of T cell
precursors. Moreover, mice with T-acute lymphocytic leukemia, in which
∼99% of the splenic nucleated cells are T cell blasts [[144]23],
display an even greater increase in serum thymidine than LCMV-infected
mice do, suggesting that these dividing T cell blasts are major
producers of thymidine ([145]Figure 6F). Activated T cells in culture
release thymidine into media ([146]Figure 6G). Finally, virus-specific
P14 CD8+ T cells transferred into mice before infection and then
isolated on day 8 of acute or chronic infection display increased
thymidine concentrations relative to CD8+ T cells from uninfected mice
([147]Figure 6H, median of 8- or 5-fold increase in thymidine relative
to T cells from uninfected mice). Collectively, these data suggest that
activated T cells in the spleens of LCMV-infected mice are dominant
producers of thymidine, contributing to the high thymidine turnover
observed in LCMV infection.
3. Discussion
In this study, we used metabolomics and labeled nutrient infusions to
identify metabolic changes occurring during days 3 and 8 of Armstrong
and Clone 13 LCMV infection. We found that while both forms of
infection change metabolite levels and metabolite use in tissues, more
dramatic changes occur during infection with the Clone 13 strain, which
continues to replicate at day 8 unlike the Armstrong strain. For
example, blood levels of itaconate, kynurenine, thymidine, and uridine
all change more in chronic Clone 13 than acute Armstrong strain LCMV
infection. Similarly, glucose and glutamine contribution to the TCA
cycle in the highly-infected tissues spleen, liver and kidney all
increase in chronic infection, while only glutamine use in the spleen
increases in acute infection. Whole-body nutrient turnovers were more
consistent between acute and chronic: thymidine turnover increases in
both cases, while glutamine turnover increases on day 3 of chronic and
day 8 of acute infection. Our data further suggest that activated T
cells may be the producers of the increased thymidine in both acute and
chronic LCMV infection, although more mechanistic investigation would
be valuable in future.
Overall, we observed that chronic LCMV infection changes metabolism
more than acute infection. There are two major differences between
these disease states: in chronic infection at day 8 virus persists in
blood and tissues, while it is low or absent at that time in acute
infection; as a consequence, the innate and adaptive immune response
persists in chronic infection, while in acute infection at day 8 the
immune response is beginning to resolve [[148]3,[149]35]. It is likely
that both the innate and adaptive immune response contribute to the
metabolic changes observed: for example, itaconate is a hallmark of
innate immune activation of macrophages [[150]9,[151]10] while our data
suggests that lymphocytes may be dominant producers of thymidine. In
future, it would be valuable to examine metabolite levels and metabolic
fluxes at even later stages of infection, where the outcomes of the two
types of infection have diverged yet further. It would also be valuable
to use cell-type isolation paired with isotope tracing [[152]36], or
imaging mass spectrometry [[153]37,[154]38], to identify which cell
types change their metabolism in acute and chronic infection.
This work nominates increased thymidine in the blood and thymidine
whole-body turnover as biomarkers of viral infection. Thymidine
turnover increases in both acute and chronic infection ([155]Figure 4),
though blood level ([156]Figure 1) and spleen synthesis rate
([157]Figure 6) increase more in chronic infection. Based on our
observations, we hypothesize that activated T lymphocytes may be
primary producers of thymidine. It may be that thymidine turnover
increases during any adaptive immune response, or any state when
lymphocyte-like cells are proliferating (for example T cell leukemia,
[158]Figure 6). In future, we hope to test this association further in
other types of infection or vaccination in both mice and humans.
However, thymidine metabolism differs between mice and humans, with
serum thymidine levels around 10-fold higher in mice [[159]39], so this
feature of infection might not be shared between the species. We also
do not yet know the fate of the thymidine produced during LCMV
infection: thymidine can be broken down by the liver and excreted or
can be recycled into further deoxynucleotide synthesis ([160]Figure
6A). If the latter fate is dominant, then modulating thymidine levels
during an infection or vaccination may modulate the immune response or
the formation of immune memory, and it would be valuable to test this
in future.
We were surprised that LCMV infection does not change whole-body
turnover of glucose. How can we reconcile this with the increases in
glucose contribution to the TCA cycle in specific tissues ([161]Figure
5), and the previously reported decrease in respiratory quotient and
thus carbohydrate oxidation during chronic LCMV infection [[162]16]?
First, spleen, liver, and kidney are relatively small contributors to
whole-body TCA flux in the mouse, summing to around 10% of the
whole-body total [[163]23], so the increased contribution of glucose
measured would not necessarily lead to changes in whole-body glucose
consumption. Second, in the study of Baazim and colleagues [[164]16],
the decrease in carbohydrate oxidation seemed to be driven by a
decreased feeding by chronic-infected mice [[165]40]. Since the
standard mouse diet is very high in carbohydrates, reduced food
consumption would be expected to lower whole-body carbohydrate
oxidation. However in our studies, all mice were fasted for 5 hours
before the start of 2.5-hour labeled nutrient infusions, which may have
masked the metabolic effect of infection-reduced feeding.
Overall, pairing isotope-labeled nutrient infusions with metabolomics
is a powerful approach to measure changes in whole-body and
tissue-specific metabolism. These approaches have most comprehensively
been used to measure metabolism in the context of diabetes/obesity
[[166]41,[167]42], burn wounds [[168]43], and in recent years, in
cancer [[169]44]. However, as a field we know less about the in vivo
metabolism of other physiologic and pathologic states including
infection and immune response, so this study aims to help fill this gap
in the field. In future, similar approaches should be used to measure
the metabolism of other physiologic states such as cytokine storm,
aging, and neurodegenerative disease. Most importantly, such studies
can nominate metabolic pathways specifically changed in different
disease states, and then such pathways can be modulated to test whether
this can improve disease pathology.
4. Methods
4.1. Mice
All animal studies were approved either by the Memorial Sloan Kettering
Institutional Animal Care and Use Committee (majority of experiments),
the Princeton Institutional Animal Care and Use Committee (healthy
tissue thymidine concentration and mouse T cell isolation for in vitro
culture), or the Rutgers Institutional Animal Care and Use Committee
(T-acute lymphocytic leukemia experiment). All experiments were
performed in C57Bl/6 mice from Charles River Laboratories. Mice for
nutrient infusions were purchased with jugular vein catheters implanted
by Charles River Laboratories.
4.2. T-cell acute lymphocytic leukemia model
Generation of NOTCH1-induced mouse primary T-cell acute lymphocytic
leukemia and secondary transplantation into sub-lethally irradiated
recipients was performed as previously described [[170]45]: 1 × 10^6
leukemia cells were transplanted from primary recipients into
sub-lethally irradiated (4.5 Gy) 9-week-old C57BL/6 secondary
recipients, which were pre-catheterized by Charles River Laboratories
in the right jugular vein. Animals were monitored for signs of distress
and normal motor function daily. Blood was sampled 9–10 days post
leukemia transplantation.
4.3. Lymphocytic choriomeningitis virus infection
Armstrong and Clone 13 strains of LCMV were purchased from the European
Virus Archive. Mice were infected by intraperitoneal delivery of 2 ×
10^5 plaque-forming units (PFU) of LCMV Armstrong or intravenous
delivery of 4 × 10^6 PFU of LCMV Clone 13.
4.4. Nutrient infusions
Nutrient tracers were diluted as follows: [U–^13C] glucose (CLM-1396,
Cambridge Isotope Laboratories) was diluted to 300 mM in sterile
saline; [U–^13C] lactate (20 w/w% solution, CLM-1579, Cambridge Isotope
Laboratories) was diluted to 5% (4x dilution, 435 mM final
concentration) in water; [U–^13C] glutamine (CLM-1822, Cambridge
Isotope Laboratories) was diluted to 100 mM in sterile saline;
[^15N[2]] thymidine (NLM-3901, Cambridge Isotope Laboratories) was
diluted to 1 mM in sterile saline. Five hours prior to tracer infusion,
mice were fasted by switching to fresh cages without food. Mice were
weighed and infusions were begun at a rate of 0.1 μL per minute per
gram body weight. Each infusion was continued for 150 min. After 150
min, mouse blood was collected by submandibular vein sampling. After
animal sacrifice by cervical dislocation, tissues were rapidly
collected and freeze-clamped using a liquid nitrogen-cooled
Wollenberger clamp. Tissues were stored at −80 °C until processed.
4.5. Ex vivo T cell isolation for metabolomics
On the day prior to viral infection, 5 × 10^4 P14 transgenic
(LCMV-specific) CD8^+ T cells (CD45.1^+) were delivered intravenously
to catheterized recipient mice (CD45.2^+). On day 8 of infection,
spleens were harvested and rapidly processed into single-cell
suspensions on ice, subjected to brief (∼3 min) red blood cell lysis
(ACK Lysing Buffer; Gibco), followed by rapid (∼10 min) magnetic bead
purification of CD45.1^+ cells according to the manufacturer’s
instructions (MagniSort™ Mouse CD45.1 Positive Selection Kit;
ThermoFisher). To measure T cells from uninfected mice, spleens of
wild-type C57Bl/6 mice without P14 transfer were harvested and
processed in the same manner, followed by rapid (∼10 min) magnetic bead
purification of total CD8^+cells according to the manufacturer’s
instructions [Dynabeads™ Mouse CD8 (Lyt 2); Invitrogen]. Purified T
cells from infected and uninfected mice were counted, resuspended at
1.5 × 10^6 cells/100 ml in precooled acetonitrile-methanol–water
(40%/40%/20% v/v/v), and stored at −80 °C until processed.
4.6. T cell in vitro stimulation
Mouse spleens were harvested and pooled as single-cell suspensions by
passing through 70-μm cell strainers into RPMI 1640 media. After red
blood cell lysis (eBioscience, 00-4300-54), naïve CD8^+ T cells were
purified by magnetic bead separation using mouse naïve CD8a^+ T Cell
Isolation Kit (Miltenyi Biotec, 130-096-543) following vendor
instructions. Approximately 2–3 million purified naïve CD8+ T cells can
be obtained from each mouse. Cells were cultured in complete RPMI media
made with RPMI 1640 (11875119, ThermoFisher), supplemented with 10%
FBS, 100 U ml^−1 penicillin, 100 μg ml^−1 streptomycin and 55 μM
2-mercaptoethanol. Cells were maintained at 1 million cells per ml in 1
ml media in 12-well plates. T cells were stimulated for 24 h with
plate-bound anti-CD3 (10 μg ml^−1, Bio X Cell, BE0001-1) and anti-CD28
(5 μg ml^−1, Bio X Cell, BE0015-1) in complete RPMI media supplemented
with recombinant IL-2 (100 U ml^−1, Peprotech, 217-12). At 24 h after
stimulation, media was changed and then was sampled for mass
spectrometry at 32 h after stimulation.
4.7. Blood processing and extraction
Sampled blood was kept on ice for up to 60 min after sampling, then
centrifuged at 4 °C, 15000 RCF for 10 min. The serum fraction was
transferred to another tube and stored at −80 °C. For mass
spectrometry, 2–3 μL of serum were extracted in 50 volumes methanol on
dry ice, then centrifuged at 4 °C, 15000 RCF for 25 min, and then
transferred to mass spectrometry vials (Thermo Scientific 200046, caps
Thermo Scientific 501313) for measurement.
4.8. Tissue processing and extraction
Throughout processing until extraction, tissues were kept buried in dry
ice, or in metal Eppendorf racks surrounded by dry ice. Tissues were
ground into powder using the Retsch CryoMill. Tissue powder was weighed
(5–20 mg in dry-ice-precooled Eppendorf tubes), and tissues were
extracted by vortexing in 40x volumes precooled
acetonitrile-methanol–water (40%/40%/20% v/v/v), then left on water ice
over dry ice for 10 min. Then the solution was centrifuged at 4 °C,
14000 RCF for 25 min, moved to a new tube, then centrifuged again at 4
°C, 14000 RCF for 25 min, to remove any particulates. Then extract was
moved to mass spectrometry vials for measurement.
4.9. T cell media processing and extraction
Twenty microliters of media was extracted in 80 μL methanol at −20°C.
Samples were centrifuged at 4 °C, 14000 RCF for 25 min. Then 30 μL
supernatant was added to 90 μL precooled acetonitrile-methanol–water
(40%/40%/20% v/v/v) and moved to mass spectrometry vials for
measurement.
4.10. Mass spectrometry
Water soluble metabolite measurements were obtained by running samples
on the Q Exactive Plus hybrid quadrupole-orbitrap mass spectrometer
(Thermo Scientific) coupled with hydrophilic interaction chromatography
(HILIC). An XBridge BEH Amide column (150 mm × 2.1 mm, 2.5 μM particle
size, Waters, Milford, MA) was used. The gradient was solvent A (95%:5%
H[2]O:acetonitrile with 20 mM ammonium acetate, 20 mM ammonium
hydroxide, pH 9.4) and solvent B (100% acetonitrile) 0 min, 90% B; 2
min, 90% B; 3 min, 75%; 7 min, 75% B; 8 min, 70% B, 9 min, 70% B; 10
min, 50% B; 12 min, 50% B; 13 min, 25% B; 14 min, 25% B; 16 min, 0% B,
20.5 min, 0% B; 21 min, 90% B; 25 min, 90% B. The flow rate was 150
μL/min with an injection volume of 10 μL and a column temperature of 25
°C. The MS scans were in negative ion mode with a resolution of 140,000
at m/z 200. The automatic gain control (AGC) target was 5 × 10^6 and
the scan range was m/z 75−1000. Xcalibur (Thermo Fisher) was used to
collect raw data.
4.11. Flow cytometry
Flow cytometry was performed on single-cell suspensions of splenocytes
after red blood cell lysis. Fluorescently labelled antibodies were
purchased from Biolegend: anti-CD3 PE (17A2), anti-CD4 BV711 (RM4-5),
anti-CD8 BV785 (53–6.7), anti-CD11b BV605 (M1/70), anti-CD19 BV510
(6D5), anti-CD44 FITC (IM7), and anti-Ly6G FITC (1A8); or eBioscience:
anti-PD1 PE-Cy7 (J43). Data was collected on a Cytek 5 Laser Aurora and
analyzed with FlowJo 10 software.
4.12. Mass spectrometry analysis
All LC-MS data, both raw abundances and abundances of isotope-labeled
forms, were analyzed by El-Maven v0.5.0.
4.13. Metabolomics analysis
Peaks were selected automatically by El-Maven and then all peaks were
manually inspected. Metabolite identities were based on matching
mass-to-charge ratio and retention time of standards previously run
with this method. Peaks matching standards for compounds not expected
in mammalian blood and tissues were discarded.
Global metabolomic analyses in each type of infection were performed
with data from a single experiment to avoid data normalization
([171]Figure 1, [172]Figure 2, [173]Figure 3A-B, [174]Figs. S1–S3), and
in this case t-tests were performed on log-transformed metabolite
intensities. For figure panels examining specific metabolites, or
comparing different viral strains in infections conducted at different
times, each metabolite intensity in viral infection was normalized to
its intensity in uninfected samples from the same experiment, and
fold-changes were pooled and compared across experiments ([175]Figure
1, [176]Figure 3, [177]Figure 6H) and t-tests were performed on
log[2]-fold-changes from uninfected samples. Numbers of mice per group
are listed in figure legends.
Heatmaps and principal component analysis plots were produced using
Metaboanalyst ([178]https://www.metaboanalyst.ca/) using the
“Statistical Analysis [one factor]” tool, and for this analysis
metabolomics data were log transformed and row-mean centered. When
comparing metabolite levels using t-tests, any metabolite intensities
less than 1000 were set to 1000 (approximate limit of detection for
this mass spectrometer). Pathway analysis was performed using
Metaboanalyst, and then selecting pathways where a) p-value of
enrichment was p < 0.05, b) at least two metabolites from the pathway
were changed in the dataset, c) discarded duplicate pathways where
multiple pathways were enriched based on the same two or more
metabolites.
4.14. Labeling analysis
For infusion experiments involving carbon-13 or nitrogen-15 labeling,
outputs were corrected for natural ^13C abundance using the accucor
package in R (Version 0.2.3).
Whole body turnover of an infused nutrient was calculated using
equation [179]1.
In the case of carbon-13 lactate infusion, all measured L were
multiplied by 1.45 before use in equation [180](1), to correct for the
reduced labeling observed when sampling venous blood compared to
arterial blood during carbon-13 lactate infusion [[181]17]. (This
reduced labeling is likely due to a combination of the high turnover of
lactate and the high production of lactate by the tissue bed drained by
the vein.)
To calculate the contribution of an infused nutrient to a tissue
metabolite, we calculated the average carbon-13 labeling of the infused
nutrient in the blood, as well as the average carbon-13 labeling of the
nutrient of interest in the tissue, using equation
[MATH:
Lavg=1n∗∑i=1ni∗proportionilabeled :MATH]
[2]
where n is the total number of carbons in the nutrient (i.e. 6 for
glucose, 3 for lactate, 5 for glutamine, 4 for malate). Then the
contribution of an infused nutrient x to a tissue nutrient y such as
malate is calculated by taking the quotient of L[avg] for the tissue
nutrient y divided by L[avg] for the infused blood nutrient x,
[MATH: Ly<−x=Lav
g,yL
avg,x :MATH]
[3]
4.15. Statistics and figures
All t-tests are Student’s two-tailed t-tests, and details of
experimental sample size and t tests are described in each legend.
T-tests on metabolite levels were performed on log-transformed
metabolite intensities, except for the comparisons between day 8 of
acute and day 8 of chronic infection serum and tissues, in which case
t-tests were performed on the log[2] fold change of infected normalized
to uninfected. Some schematics were created using BioRender.
CRediT authorship contribution statement
Caroline R. Bartman: Writing – original draft, Investigation, Writing –
review & editing, Methodology, Visualization, Formal analysis,
Validation, Data curation, Conceptualization. Shengqi Hou: Writing –
review & editing, Methodology, Investigation. Fabian Correa:
Investigation. Yihui Shen: Investigation. Victoria da Silva-Diz:
Investigation. Maya Aleksandrova: Investigation. Daniel Herranz:
Supervision. Joshua D. Rabinowitz: Methodology, Funding acquisition,
Writing – review & editing, Conceptualization, Supervision. Andrew M.
Intlekofer: Funding acquisition, Conceptualization, Supervision,
Writing – review & editing.
Declaration of competing interest
The authors declare the following financial interests/personal
relationships which may be considered as potential competing interests:
Joshua D. Rabinowitz reports a relationship with Bantam Pharmaceutical
LLC that includes: consulting or advisory and equity or stocks. Joshua
D. Rabinowitz reports a relationship with Pfizer Inc that includes:
consulting or advisory. Joshua D. Rabinowitz reports a relationship
with Third Rock Ventures LLC that includes: consulting or advisory.
Joshua D. Rabinowitz reports a relationship with Empress Therapeutics,
Inc. that includes: equity or stocks. Joshua D. Rabinowitz is an
advisor and stockholder in Colorado Research Partners, L.E.A.F.
Pharmaceuticals, Barer Institute, and Rafael Pharmaceuticals; a
founder, director, and stockholder of Farber Partners, Serien
Therapeutics, and Sofro Pharmaceuticals; and a director of the
Princeton University-PKU Shenzhen collaboration. - JDR.
If there are other authors, they declare that they have no known
competing financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
Acknowledgements