Introduction

Each year, at least 1.7 million Americans develop sepsis, of which 270,000 will die1,2. It is present in 6% of adult hospitalizations and causes 1 in 3 in-hospital deaths1. Globally, sepsis afflicts 19 million people and leads to 5 million deaths annually3. Those patients who survive to be discharged from the hospital are not cured4,5,6,7,8. Two- and 5-year mortality after treatment for community acquired pneumonia is upwards of 9 times higher than the baseline age-specific mortality and higher than mortality after an initial hospitalization for heart failure, stroke, or major fracture5,7,8. This elevated risk is not explained by a higher burden of chronic diseases prior to the infection4,5,7,8. Recent studies of sepsis survivors draw attention to persistent elevations in systemic markers of inflammation (e.g., interleukin (IL)-6), lending to a theory of inflammation causing their disabilities and shortened lifespan9,10,11.

Eukaryotic life depends upon ATP and calcium (Ca2+); the two are intimately entwined within the mitochondrion. Ca2+ is a ubiquitous signaling molecule involved in the regulation of many ATP-consuming cellular functions; thus, it makes teleologic sense that Ca2+ serve as the signaling molecule to modulate bioenergetics and ATP synthesis12. Indeed, elevations in mitochondrial [Ca2+]m augment the activity of three dehydrogenases of the tricarboxylic acid (TCA) cycle13,14,15. The reciprocal is also true; mitochondria influence Ca2+ signaling12,16,17. Mitochondria provide the power to generate Ca2+ gradients that drive the Ca2+ release from intracellular and extracellular stores. They serve as a large capacity buffering system that sequesters Ca2+ predominantly through the negative electrochemical gradient generated by the mitochondrial membrane potential (ΔΨm)12. In doing so, mitochondria influence the spatiotemporal profile of cellular Ca2+ signals.

Mitochondrial Ca2+ uptake occurs through the heteromeric mitochondrial calcium uniporter (MCU) complex positioned in the inner membrane and comprised of a channel and several regulatory subunits18,19,20,21,22,23. MCU is the channel that mediates the electrophoretic uptake of Ca2+ into the matrix. MICU1 (mitochondrial calcium uptake 1) and MICU2 function as gatekeepers of the MCU channel. High-affinity cooperative Ca2+ binding by MICU1 and MICU2 renders the MICU1-MICU2 complex capable of responding directly to Ca2+cyt. MICU1/2 inhibit Ca2+ entry through the MCU channel at low [Ca2+]cyt; with rising and high [Ca2+]cyt, MICU1 facilitates Ca2+ entry into the mitochondrion24,25,26,27,28,29. Cells deficient in MICU1 exhibit impaired bioenergetics and cell function and increased production of reactive oxygen species (ROS) due to unregulated elevations in mitochondrial [Ca2+]mit at physiologic [Ca2+]cyt24,25,26,27. Heritable mutations in neuronal MICU1 causes uncontrolled mitochondrial Ca2+ uptake that underlies a unique human neurodegenerative disease30,31. MCUR1 (mitochondrial calcium uniporter regulator 1) is a direct regulator of MCU channel activity and is required for mitochondrial Ca2+ uptake and maintenance of normal cellular bioenergetics. MCUR1 deletion reduces Ca2+ uptake by mitochondria and disrupts oxidative metabolism32.

Sepsis, particularly when severe, perturbs Ca2+ homeostasis and mitochondrial function. Indeed, nearly 100% of septic patients experience derangements in Ca2+ homeostasis, leading to toxic levels of Ca2+ within the cytosol and mitochondrial matrix, altered oxidative metabolism, and elaboration of ROS occur33,34,35,36,37. These insults may threaten cell survival. They may also serve as sufficient pressure to select for cells uniquely able to adapt and restore Ca2+ homeostasis – i.e., ‘weather the storm.’ For cells that survive, these changes may be long-lasting. In this study we show that the lymphoid tissues of mouse and human sepsis survivors possess a restructured MCU complex. Changes in the expression of its constituent subunits cause long-term alterations in Ca2+ signaling, oxidative metabolism, and the ability of lymphoid tissues to maintain their phenotype – e.g., immune dysfunction. As Ca2+ signaling is essential to monocyte cytokine production, the resultant alteration to Ca2+ homeostasis generates a basal state of heightened systemic cytokine concentrations similar to post-intensive care syndrome (PICS)38,39,40.

Results

Mice surviving sepsis have increased basal plasma cytokine concentrations and an altered lymphoid phenotype

Heightened systemic cytokine concentrations (i.e., plasma IL-6) have been reported in sepsis survivors with post-intensive care syndrome (PICS)38,39. We first tested whether this occurs in mice. We initially compared the plasma concentrations of cytokines in mice that had survived to 7 and 30 days after either 1) cecal ligation puncture (CLP) followed by best clinical practices (i.e., control of the source of infection41, resuscitation with balance crystalloids42, 96 h of antibiotics43) for the treatment of intraabdominal sepsis (CLP30d) or 2) sham laparotomy plus best clinical practices (Sham30d). We observed significantly increased basal plasma IL-6 and IL-10 concentrations in CLP30d mice at 7 days and TNFα and IL-6 at 30 days compared to Sham30d mice (Fig. 1A). With repeat TLR4 (toll-like receptor 4) agonism (i.e., intratracheal instillation of lipopolysaccharide - LPSi.t.), CLP30d mice failed to mount as robust a cytokine response compared to Sham30d mice (Fig. 1B).

Fig. 1: Mice surviving sepsis exhibit an altered immune phenotype in vivo.
figure 1

C57Bl/6 mice underwent cecal ligation and puncture (CLP) with best clinical practices or sham surgery with best clinical practices. At day 7 or 30, mice were euthanized, blood was collected, and plasma cytokines concentrations assayed by ELISA. At day 30, a nested, parallel cohort was administered intratracheally either LPS (3 mg/kg; i.t.) or equivolume 0.9% saline (NS). After 24 h, mice were euthanized, blood was collected, and plasma cytokines concentrations assayed by ELISA. A TNFα, IL-6, and IL-10 plasma concentrations – basal at day 7 and day 30. B TNFα, IL-6, and IL-10 plasma concentrations – endotoxemia. Sham (n = 5, 7 mice) CLP (n = 8, 19 mice). Wilcoxon rank sum. bar = median.

Full size image

Mice surviving sepsis exhibit altered MCU complex stoichiometry: reduced MICU1

Immune cell function depends upon Ca2+ signaling and ATP, both of which are regulated by the MCU complex and the mitochondrion. We explored whether alterations to the MCU complex occur in CLP30d mice. Mitochondrial calcium uptake 1 (MICU1) functions as a gatekeeper of the MCU channel, inhibiting Ca2+ entry through the MCU channel at low [Ca2+]cyt and facilitating Ca2+ entry with high [Ca2+]cyt24,25,26,27,28,29. In both the spleen and bone marrow, CLP30d mice exhibited lower expression of MICU1 relative to the general mitochondrial marker: mitochondrial import receptor subunit translocase of the outer membrane (TOM) 20 (Fig. 2A, B, Supplementary Fig. 1A). Expression of the MCU channel relative to TOM20 was altered after CLP, though differently in each tissue, being significantly reduced in the spleen and not reduced in bone marrow (Fig. 2C). Thus, intraabdominal sepsis leads to a significant reduction in the ratio of MICU1 to MCU (MICU1:MCU) expression in the spleen of the host organism (Fig. 2D)44. A similar, though not statistically significant, reduction in MICU1:MCU was observed in the bone marrow of sepsis surviving CLP30d compared to Sham30d mice. MCUR1 is required for mitochondrial Ca2+ uptake and maintenance of normal cellular bioenergetics32. We observed significantly reduced MCUR1 expression relative to TOM20 in the spleen and in the bone marrow, and relative to MCU in the bone marrow, of CLP30d mice (Fig. 2E, F). These data suggest that sepsis perturbs the stoichiometry of the regulatory subunits MICU1 and MCUR1 and the channel MCU of the MCU complex in lymphoid tissues.

Fig. 2: Sepsis induces long-lasting changes to the mitochondrial Ca2+ uniporter (MCU) complex.
figure 2

A–F CLP 30 days. C57Bl/6 mice underwent Sham or CLP and best clinical practices. At 30 days, mice were euthanized, spleen and bone marrow tissues were harvested, and total cell protein lysate isolated and analyzed by immunoblot for MCUR1, MICU1-3, EMRE, MCU, and TOM20. Sham (n = 8 mice) CLP (n = 15 mice) G–I Endotoxemia. C57Bl/6 mice were injected with a single dose of LPS (4 mg/kg; i.p.). At 30 days, spleen and bone marrow tissues were harvested and analyzed for MCUR1, MICU1, MCU, and TOM20. Sham (n = 8 mice) CLP (n = 15 mice) J Mitochondrial protein. Spleen were harvested, and a mitochondria-enriched fraction was isolated and analyzed by immunoblot for MICU1, MCU, and TOM20. Sham (n = 5 mice) CLP (n = 5 mice) K RT-PCR. Densitometry was performed with NIH Image (J). Wilcoxon rank sum. bar = median.

Full size image

As CLP is a model with a predominance of Gram-negative organisms, we asked ‘can a single dose of LPS (4 mg/kg i.p.; LD10) delivered intraperitoneally (LPS30d) induce similar changes?’ We did not observe a significant reduction in MICU1:MCU or MICU1:TOM20 in LPS30d mice compared to mice that received intraperitoneal normal saline (NS30d) (Fig. 2G–I, Supplementary Fig. 1B).

The changes to MICU1:MCU stoichiometry after CLP sepsis appeared to occur as early as 7 days and continued to be observed at 60 days after CLP (Supplementary Fig. 2). The reduced MICU1 appeared to occur at the mitochondrion (Fig. 2J) and was not due to reduced transcription of MICU1 mRNA (Fig. 2K).

Restructuring the MCU complex alters mitochondrial and cellular Ca2+ homeostasis

Alterations to MICU1 and MCUR1 expression may perturb the regulation of Ca2+ uptake through the MCU channel. Hence, we next tested and compared mitochondrial Ca2+ homeostasis in CLP sepsis and sham-surviving mice. Peritoneal macrophages (Mφ) isolated from CLP30d mice had reduced MICU1 expression, which correlated with increased basal mitochondrial [Ca2+]m compared to Sham30d Mφ (Fig. 3A, B). This observation is consistent with a channel lacking the MICU1 regulatory subunit25,27. Guided by these data, we used siRNA to test whether inhibiting MICU1 expression can recapitulate the alterations to mitochondrial Ca2+ regulation imparted by sepsis. MICU1siRNA RAW 264.7 Mφ cells exhibited elevated basal [Ca2+]m relative to non-target (NTsiRNA) cells (Fig. 3C). LPS increased [Ca2+]m in NTsiRNA RAW cells but not in MICU1siRNA RAW cells, for which basal levels were already elevated. Thus, the reduced MICU1 seen in the lymphoid tissues of sepsis survivors alters mitochondrial Ca2+ uptake; there is greater uptake with low extramitochondrial [Ca2+] and at resting basal states.

Fig. 3: Immune cells of mice surviving sepsis exhibit alterations in mitochondrial and cellular Ca2+ regulation.
figure 3

C57Bl/6 mice underwent Sham or CLP and best clinical practices. A Macrophage (Mφ) MCU complex expression. At day 30, peritoneal Mφ were isolated, and total cell protein lysate was analyzed by immunoblot for MICU1, MICU2, MCUR1, MCU, TOM20, and actin. Sham (n = 2 mice). CLP (n = 5 mice) B Mφ mitochondrial [Ca2+]m ex vivo. At 30 days, peritoneal Mφ were isolated from Sham and CLP mice, plated, and then loaded with Rhod-2 am for 30 min at 37 °C. Fluorescence intensity as a marker of [Ca2+]m was imaged by spectrophotometry: ex/em 552 nm/581 nm. Sham (n = 4 mice) CLP (n = 5 mice) C siRNA inhibition of MICU1 and MCUR1. RAW 264.7 Mφ cells were transfected with Non-target (NT), MICU1, MCUR1, or both MICU1 and MCUR1 siRNA. After 72 h, cells were loaded with Rhod-2 am at 37 °C. Fluorescence intensity was imaged by spectrophotometry: ex/em 552 nm/581 nm. In parallel cell populations, total cell protein lysate was isolated and analyzed by immunoblot for MICU1-3, MCUR1, MCU, TOM20, and actin. RAW 264.7 (4 separate experiments; n = 3 and 4 cell groups per experiment) Wilcoxon rank sum. bar = median.

Full size image

Reduced MICU1:MCU of sepsis survivors correlates with altered metabolism

Elevations in [Ca2+]m activate several dehydrogenases of the tricarboxylic acid (TCA) cycle including pyruvate dehydrogenase (PDH). Upon elevation in [Ca2+]m, PDH phosphatase (PDP1/2) dephosphorylates Serine 293 in the E1α subunit of PDH (pSer293-PDH-E1α), thereby activating PDH45,46. Active PDH-E1α then catalyzes the oxidative decarboxylation of pyruvate to produce acetyl coenzyme A (acetyl-CoA), nicotinamide adenine dinucleotide (NADH), and CO2. These reduced equivalents (i.e., NADH) enter the electron transport chain (ETC) to increase ATP production. Whether these changes in the MCU complex alter the response of tissues to subsequent metabolic stress was studied by subjecting Sham30d and CLP30d mice to a second insult of intratracheal LPS (LPSi.t.). Again, we observed reduced MICU1:MCU in CLP30d vs. Sham30d mice, and in Sham30d mice exposure to LPSi.t. appeared to reduce MICU1:MCU further (Fig. 4A, B, Supplementary Figs. 3A and B). We did not observe appreciable differences between CLP30d mice and Sham30d mice in p-PDHE in lymphoid tissues (Fig. 4C, D). Though not statistically significant, basal NAD+ and NAD total were lower in the spleens of CLP30d compared to Sham30d mice (Fig. 4E–H). With a second stress of LPSi.t., NAD+ and NAD total levels significantly decreased in Sham30d mice. This did not occur in CLP30d, which after LPSi.t. had higher spleen concentrations of NAD+ and NAD total compared to Sham30d + LPSi.t. mice and a significant increase in tissue ATP concentration (Fig. 4I). In vitro RNAi of MICU1 and MCUR1 compared to Non-target siRNA in RAW 264.7 Mφ (Fig. 3B) did not significantly alter resting or LPS-stimulated NADH, NAD+, NAD+/NADH, or NAD total concentrations (Fig. 4J). These data support that lymphoid tissue metabolism may be altered in vivo in animal sepsis survivors.

Fig. 4: Reduced MICU1:MCU of sepsis survivors correlates with altered metabolism.
figure 4

C57Bl/6 mice underwent Sham30d or CLP30d and best practices and on day 30 were administered intratracheally either lipopolysaccharide (3 mg/kg; i.t.) or equivolume 0.9% normal saline (NS). Twenty-four hours later spleen and bone marrow tissues were harvested and analyzed. A, B MICU1 and MCU expression. C, D p-PDHE and PDHE expression. E–H NADH, NAD+, NAD+/NADH, NAD total. I ATP concentration. Sham (n = 5, 7 mice) CLP (n = 8, 19 mice) J siRNA inhibition of MICU1 and MCUR1. RAW 264.7 Mφ cells were transfected with Non-target (NT), MICU1, MCUR1, or both MICU1 and MCUR1 siRNA. After 72 h cells were exposed to LPS (1 ug/mL) or equivolume normal (0.9%) saline. Twenty-four hours later total cell lysate was harvested and analyzed. (n = 6 experiments) Densitometry was performed with NIH Image (J). Wilcoxon rank sum. bar = median.

Full size image

Reduced MICU1:MCU is associated with altered apoptosis and pyroptosis in lymphoid tissues

Calcium is critical for the induction of apoptosis and pyroptosis47,48,49,50. The degree to which a restructured MCU complex alters lymphoid tissue apoptosis and pyroptosis was tested in CLP30d and Sham30d mice at baseline and after LPSi.t.. Caspase 3 is a protein that plays a central role in apoptosis; its activation is heralded by cleavage, yielding cleaved caspase 3 (cl-cas3). The spleen tissues of CLP30d mice exhibited significantly elevated basal cl-cas3 expression compared to Sham30d mice; for both groups of mice, cl-cas 3 was increased by LPSi.t., more so in CLP30d mice (Fig. 5A, Supplementary Fig. 3A). The bone marrow tissues of CLP30d and Sham30d mice exhibited similar levels of cl-cas 3 at baseline, though significant increases with LPSi.t. that were greater in CLP30d mice (Fig. 5B, Supplementary Figs. 3B). Pyroptosis is triggered by Caspase-1, and its induction detected by the presence of cleaved caspase-1 (cl-cas1). The lymphoid tissues of CLP30d mice exhibited similar cleaved-caspase 1 (cl-cas1) at baseline compared to Sham30d mice (Fig. 5A, B, Supplementary Figs. 3A, B). A second insult of LPSi.t. reduced cl-cas1 expression in the splenic tissue of CLP30d mice, which was not observed in Sham30d exposed to LPSi.t.. Thus, in response to subsequent endotoxemia, the lymphoid tissues of sepsis surviving mice undergo heightened apoptosis and, in the spleen, less pyroptosis.

Fig. 5: Reduced MICU1:MCU is associated with altered apoptosis and pyroptosis in immune tissues.
figure 5

C57Bl/6 mice underwent Sham or CLP and best practices. On day 30, mice were administered intratracheally lipopolysaccharide (4 mg/kg, i.t.) or equivolume 0.9% normal saline (NS). Twenty-four hours later A Spleen and B Bone Marrow tissues were harvested and analyzed by immunoblot for cleaved caspase 3 (cl-cas3), caspase 3 (cas 3), cleaved caspase 1 (cl-cas 1), caspase 1 (cas 1), and actin. Densitometry was performed with NIH ImageJ. Sham (n = 4, 6 mice) CLP (n = 5, 11 mice) Wilcoxon rank sum. bar = median.

Full size image

MICU1 is degraded by heightened chaperone-mediated autophagy (CMA) and the lysosome after sepsis

We tested potential cell mechanisms of degrading MICU1 protein, focusing upon the lysosome, as it is a common terminal degradative pathway for macroautophagy, microautophagy, and chaperone-mediated autophagy (CMA)50,51,52. We observed increased lysosome-associated membrane protein type 2 (LAMP2) and microtubule-associated proteins 1 light chain 3B (LC3B-II) in the lymphoid tissues of CLP30d mice compared to Sham30d mice at baseline (Fig. 6A, B). MICU1 possesses canonical and non-canonical KFERQ sequences. During CMA, the chaperone protein heat shock 70 kDa protein 8 (HSPA8) recognizes KFERQ sequences within target proteins; HSPA8-bound proteins are then directed to the lysosome. Though not statistically significant, we did perceive an interaction between MICU1, HSPA8, and LAMP2a in the spleens of CLP30d mice compared to Sham30d mice; this interaction is considered a sine qua non of CMA (Fig. 6C)51,52. In controlled in vitro experiments, exposing HEK cells to the potent mitochondrial uncoupling agent carbonyl cyanide m-chlorophenyl hydrazone (CCCP) induced interactions between MICU1 and HSPA8 (Fig. 6D). Alternatively, HSPA8 siRNA inhibited the expression of HSPA8 in RAW 264.7 macrophages and led to increased expression of MICU1 (Fig. 6E).

Fig. 6: MICU1 is degraded by heightened chaperone-mediated autophagy and lysosomal degradation after sepsis.
figure 6

C57Bl/6 mice underwent Sham or CLP and best practices. On day 30, mice were administered intratracheally LPS (4 mg/kg, i.t.) or equivolume 0.9% normal saline (NS). Twenty-four hours later spleen and bone marrow tissues were harvested and analyzed by immunoblot for LAMP2, LC3-I, LC3-II, and actin. Sham (4 mice) CLP (5, 8 mice) Densitometry was performed using NIH ImageJ. A, B LAMP2 and LC3B-II expression. C MICU1 coimmunoprecipitates with HSPA8 and LAMP2a in sepsis survivors. Spleen tissue was immunoprecipitated with anti-HSPA8 or isotype IgG control and analyzed by immunoblot for HSPA8, MICU1, LAMP2a, and MCU. D Mitochondrial depolarization induces HSPA8, MICU1, and LAMP2a interaction in vitro. HEK cells were incubated with CCCP, and at the times indicated, cell lysate was immunoprecipitated with anti-HSPA8 and analyzed for HSPA8, MICU1, and LAMP2a. (n = 5 separate experiments) E HSPA8 siRNA increases MICU1 expression. RAW 264.7 Mφ cells were transfected with Non-target (NT) or HSPA8 siRNA. After 72 h, cells were stimulated with LPS (1ug/mL). After 24 h, total cell lysate was harvested and analyzed by immunoblot for MICU1, MCU, and actin. (n = 2 experiments) F, G Lysosomal inhibition increases MICU1:MCU. C57Bl/6 mice underwent Sham30d or CLP30d. At days 2, 3, and 28 mice were administered intraperitoneally chloroquine and bafilomycin A1 or equivolume DMSO vehicle control. At day 30 spleen and bone marrow tissues were isolated and analyzed by immunoblot for MICU1, MCU, and TOM20. (n = 7, 18 mice) H NAD+/NADH (n = 9–15 mice) I IL-6 and IL-10 plasma concentration – basal. (n = 10, 12 mice) Wilcoxon rank sum. bar = median.

Full size image

To test the hypothesis that MICU1 is constitutively reduced via heightened lysosomal degradation in sepsis survivors, we treated CLP30d mice with the lysosome inhibitors chloroquine and bafilomycin A1 (CQ+BafA1)53,54; the first dose was delivered after operative resection of the cecum (i.e., source control) and administration of best practices (i.e., volume resuscitation, antibiotics). CLP30d mice exhibited reduced spleen and bone marrow MICU1:MCU and MICU1:TOM20 relative to Sham30d mice; however, CLP30d mice administered CQ+BafA1 had elevated tissue expression of MICU1:MCU and MICU1:TOM20 that more closely approximated the expression in sepsis naïve Sham30d mice (Fig. 6F, G, Supplementary Fig. 4). We observed similar NAD+/NADH across the three groups, though the NAD+/NADH of CLP30d mice administered CQ+BafA1 more closely approximated that of Sham30d mice (Fig. 6H). CQ+BafA1 treatment was associated with a partial restoration of a sepsis naïve lymphoid phenotype (Fig. 6I).

Human sepsis survivors exhibit reduced MICU1:MCU expression in lymphoid cells

Whether or not a similar restructuring of the MCU complex occurs in human sepsis survivors was studied by comparing MICU1:MCU expression in the peripheral blood mononuclear cells (PBMC) isolated from septic and trauma patients (Supplementary Table 1). As shown in Fig. 7A, the PBMC’s from septic subjects exhibited reduced MICU1:MCU at day 30 compared to comparison trauma subjects, though this did not attain statistical significance. We explored whether similar alterations occur with severe viral infection by measuring the expression of MICU1 and MCU longitudinally in the PBMCs of critically ill SARS-CoV2 ICU patients. As shown in Fig. 7B, MICU1:MCU expression gradually declined in all but one subject; this correlated with elevated [Ca2+]m in CD16+ monocytes (Fig. 7C).

Fig. 7: Human sepsis survivors exhibit reduced MICU1:MCU expression in immune cells.
figure 7

A Bacterial sepsis. Samples of blood were obtained from participants at day 30 after ICU admission and a primary diagnosis of either trauma (n = 7) or sepsis (n = 4). Peripheral blood mononuclear cells (PBMC) were isolated and analyzed by immunoblot for MICU1, MCU, and TOM20. Densitometry was performed using NIH ImageJ. bar = median. B SARS-CoV-2. Samples of blood were obtained from participants at day 7, 28 and 84 after a first sample (Day 0) collected early after ICU admission. Peripheral blood mononuclear cells (PBMC) were isolated and analyzed by immunoblot for MICU1, MCU, and TOM20. C PMBC [Ca2+]m in SARS-CoV-2. PBMC were isolated, analyzed by flow cytometry for immune subtypes and for [Ca2+]m using Rhod-2 (ex/em 552 nm/581 nm) Wilcoxon rank sum.

Full size image

Discussion

Epidemiological studies have consistently revealed that survivors of sepsis experience elevated rates of death after short-term (i.e., <28-day) endpoints and an impaired quality of life55. These data suggest that infection may have a biological role in the long-term adverse immunologic sequelae of sepsis survivors. In this study, we identify that the MCU complex has been restructured in lymphoid tissues to express reduced MICU1:MCU and long after visible recovery from sepsis. The changes to the MCU complex appear to alter mitochondrial and cellular Ca2+ handling, metabolism, and cellular function. The tissue phenotypes of these mice approximate the abnormalities observed in human survivors: elevated systemic concentrations of inflammatory mediators. Transiently inhibiting lysosomal mechanisms early in the evolution of sepsis, but after the initial infection, partially preserves the MCU complex stoichiometry and Ca2+ biology closer to a pre-sepsis state; thus, restoring fitness to sepsis survivors may be possible.

There is considerable interest and a growing need to better define, mechanistically, the entity post-intensive care syndrome (PICS) – a constellation of debilitating neurocognitive, psychiatric, and myopathic disorders persisting after ICU discharge56,57. A hallmark is persistent inflammation, immunosuppression, and catabolism syndrome, and laboratory-based diagnostic criteria (i.e., CRP, lymphocyte counts, and albumin) are being defined39,56,58,59. Our preclinical data align with observational studies that consistently report high circulating concentrations of biomarkers of inflammation (e.g., IL-6), immunosuppression (e.g., IL-10), and stress metabolism (C-reactive protein), particularly in subjects meeting diagnostic criteria of PICS38,39,56,57,58. Though widely appreciated, this immune-based paradigm continues to evolve, and the question “Why does this develop in sepsis survivors?” remains unanswered and a subject of intense study.

A preponderance of evidence confirms that abnormally high intracellular [Ca2+] are sustained in the brains, hearts, immune cells, muscle, and livers of organisms contending with severe infection60,61,62,63,64. We hypothesized that in response, cells sufficiently versatile will ‘learn’. Adaptation will manifest as changes to Ca2+ biology, specifically the stringency of regulation to mitigate toxicity and sensitivity to programmed cell death pathways. Though we did not directly test elevated [Ca2+] as a selection pressure, our data do support that a restructuring of the MCU complex of lymphoid tissues occurs across in response to sepsis. In the lymphoid tissues of recovered animals, exposure to a secondary stimulus of TLR4 agonism increases immunologically quiet apoptosis and less of the more inflammatory pyroptosis; this may underlie the paradoxical reduced IL-6 response to LPS by CLP30d mice. How, why, and the extent to which this imparts a fitness to cells, tissues, or host organism threatened by a subsequent flood of [Ca2+] requires additional investigation.

Ca2+ is central to an organized and effective immune response of leukocyte activation, production of inflammatory and immunosuppressive mediators, and the necessary energy production65,66,67,68,69. Elevations in basal [Ca2+]c enhance monocyte/Mφ cytokine production40,65. By identifying reduced MICU1:MCU and a correlation with elevated [Ca2+]m and IL-6, we provide one possible mechanism. More so, our data are encouraging that restoring MCU stoichiometry and immune cell function to a more sepsis naïve state is feasible. It is difficult, however, to attribute the ‘quieter’ phenotype (i.e., reduced cytokine concentrations) after subsequent LPS exposure to the same singular change in the MCU complex. A shift toward heightened apoptosis and less pyroptosis, as we observed in CLP30d mice, may sufficient. But additional mechanisms warrant study – e.g., enhanced mitochondrial Ca2+ buffering reduces the cytosolic Ca2+ transients (magnitude or duration) regulating mediator production.

There are limitations and additional hypotheses that require further exploration to develop a more comprehensive paradigm. Our ‘bulk’ tissue approach does not distinguish the contribution of the distinct cell types comprising the whole organ (e.g., myeloid vs. lymphoid), as revealed in the comparison of the in vivo animal and in vitro RAW cell NAD+/NADH data. Inhibition of MICU1 expression in Mφ in vitro replicated the elevated [Ca2+]m observed in CLP30d peritoneal Mφ but not the changes to NAD+ or NADH. This observation suggests that mechanisms additional to an isolated loss of MICU1 mediate alterations to Mφ metabolism, or a different immune cell type underlies the altered metabolism in the spleens of CLP30d mice. Though MICU1 appears to be degraded via the CMA-lysosomal pathway, the cellular changes (e.g., increased affinity of HSPA8 for MICU1) that render it, rather than other MCU components, more ‘susceptible’ to degradation remain to be elucidated. Certainly, post-translational modifications lead in a list of differential causes. Metabolism (i.e., NAD+/NADH) appeared to be different in sepsis- compared to sham-surviving mice; however, correlative changes to the phosphorylation state of PDHE were not detected and the perceived differences in NAD+, NADH, NAD+/NADH did not attain statistical significance. Confirmatory testing using other platforms – e.g., Seahorse and OROBOROS – are being pursued to corroborate our observations. Other Ca2+ sensitive TCA cycle may be induced by the elevated [Ca2+]m; alternatively, the strain of altered Ca2+ regulation on the mechanisms maintaining phenotype may draw globally upon the metabolic machinery. Unexpectedly, the administration of DMSO to Sham30d mice elevated basal cytokine concentrations, above those of CLP30d mice receiving DMSO. In preclinical models of disease, DMSO has been shown to attenuate tissue and plasma cytokine concentrations70,71. Its effects in models of health (i.e., absence of inflammation) are not reported. We propose a similar mechanism to that observed in CLP30d mice, which mounted a blunted cytokine response to subsequent challenge with LPS compared to Sham30d mice: i.e., administration of DMSO reduces mediator production to a greater degree in CLP30d mice. This result does not detract from our conclusion that administering lysosomal inhibitors to sepsis survivors may restore immunity to a more sepsis naïve phenotype. The effects of sepsis are manifold; occult mechanisms, concomitantly induced by sepsis but epiphenomenal to MICU1 and the lysosome, certainly occur and may be the cause of this discrepancy. The Ca2+ biology of each tissue is unique to its function (e.g., rapidly depolarizing of heart and brain vs. non-depolarizing of liver and kidney). Much work remains in understanding how and to what degree sepsis alters Ca2+ biology integrated across all tissues of the organism. Nonetheless, the collective data underscore the importance of further research.

Ca2+ and its regulation are vital to the fitness and function of every cell comprising a multicellular organism. Our data suggest that in response to sepsis, the cell restructures the MCU complex, which imparts long-lasting alterations to Ca2+ homeostasis, oxidative metabolism, and tissue phenotype. Our data link structural changes in the MCU complex with functional changes in lymphoid tissues and thereby provide a mechanistic paradigm for the development of heightened inflammation characteristic of PICS. It is hoped that this new experimental work will provide foundational knowledge as to how the mechanisms governing mitochondrial Ca2+ and metabolism are restructured during sepsis to underlie a persistent loss of cellular phenotype that leads to a progressive loss of health and shortened survival. In the context of the known reduced health and survival experienced by sepsis survivors, it is imperative that we continue to explore how sepsis changes Ca2+ biology and the resultant alterations to long-term tissue function and health.

Methods

Reagents and antibodies

Ultra Pure lipopolysaccharide (LPS - Escherichia coli 0111:B4) was obtained from List Biologicals (Campbell, CA, USA). Antibodies for MCU, MICU1 (12542), TOM20 (42406), caspase 3 (9662), cleaved caspase 3 (9661), caspase 1 (24232), cleaved caspase 1 (89332), and β-Actin (4967) were obtained from Cell Signaling Technology (Danvers, MA, USA). Anti-MCUR1 (PA5-95628), MCU (MA551430), MICU2 (PA588410), MICU3 (PA5-55177), and EMRE (BS-15136R) were obtained from Thermofisher (Waltham, MA USA). Anti-LAMP2 (ab25631) and LAMP2a (ab18528) antibody were obtained from Abcam (Cambridge, MA, USA). Rat anti-HSPA8 (ADI-SPA-815) antibody was purchased from Enzo Life Sciences (Farmingdale, NY, USA). Anti-CD3 (557757), CD16 (558122), CD19 (563227), CD11b (561001, and CD11c (561352) were purchased from BD Pharmingen (Franklin Lakes, NJ, USA). Chloroquine and CCCP were purchased from Sigma (St. Louis, MO, USA). Bafilomycin A1 was purchased from Selleckchem (Houston, TX, USA).

Animal experimentation

We performed all animal experiments in accordance with the National Institutes of Health guidelines under protocols approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pittsburgh and Washington University. We purchased 8 to 12-week-old C57BL/6J male mice from the Jackson Laboratory (Bar Harbor, ME)67,72,73,74. Mice were housed in specific pathogen-free rooms under 12 h light/12 h dark conditions with an ambient temperature of 23 °C ± 1 °C. Mice were allowed to acclimate to their new surroundings for one week prior to any experimentation. Animals were given ad libitum access to water and LabDiet Prolab Isopro RMH 3000 diet pellets (LabDiet, St. Louis, MO). We randomly grouped mice and assigned them to a specific experiment. An experimental unit (i.e., mouse) for each group was included in each box of n = 4 mice. At various time points, mice were euthanized with isofluorane (5% induction) followed by exsanguination by cardiac puncture. Blood was isolated by cardiac puncture, and the spleen and bone marrow were harvested. Investigators who treated animals knew the treatment groups and collected samples, which were then analyzed by other investigators blinded to the specific treatment.

Cecal ligation and puncture (CLP)

A 1 cm ligation, 21-gauge double puncture model of cecal ligation and puncture (CLP) under anesthesia was performed31,32,33. Sham surgery consisted of laparotomy with bowel manipulation. At the conclusion of the procedure all mice received a subcutaneous injection of warmed isotonic balanced crystalloid (30 mL/kg) to compensate for intraoperative fluid losses and recreate the hyperdynamic state associated with early sepsis75. Analgesia was achieved with buprenorphine (0.1 mg/kg SQ q12H). After surgery each animal was placed in an individual cage on a heating pad until full emergence from anesthesia (animal upright, mobile, foraging activity). The cage was then removed from the heating pad.

Twenty-four hours after CLP, the animals were re-explored to attain ‘source control’ and implement best clinical practices42,43. Specifically, animals were anesthetized and under sterile conditions, the prior laparotomy incision was reopened. The necrotic cecum was resected and the colotomy repaired. The peritoneal cavity was irrigated with saline. The abdomen was closed, and the animals were administered warmed isotonic balanced crystalloid (20 mL/kg)42. They were administered the broad-spectrum antibiotic imipenem (0.5 mg/kg q12H, i.p.) for a total of 96 h after surgery, in accordance with evidence-based guidelines43. They continued to receive analgesia during this period as well.

For experiments involving the in vivo inhibition of lysosomes, we administered by intraperitoneal (i.p.) injection chloroquine (CQ, 50 mg/kg i.p.) and bafilomycin A1 (BafA1, 1 mg/kg) at Day 2, 3 and 28. Comparison control animals received equivolume DMSO at the same administration schedule.

Endotoxemia

Mice were anesthetized using a mixture of isoflurane and oxygen bled in at 3.5 L/min. Ultra Pure LPS (Escherichia coli 0111:B4) from LIST Biologicals (Campbell, CA, USA) was dissolved in DNase/RNase free sterile normal saline and injected intraperitoneally (4 mg/kg)30,34.

Intratracheal administration of lipopolysaccharide

Mice were weighed and then anesthetized with 4% isoflurane inhalation with 2 L oxygen bleed in for several minutes. Using a fiberoptic light source, the mouth was opened, the vocal cords directly visualized, and the instillate (LPS 3 mg/kg) administered just above the vocal cords76.

Humane endpoints

Mice were euthanized if they showed signs of severe respiratory distress (i.e., tachypnea with prostration, does not resolve after several minutes, worsening condition) or if weight loss >/=20% body weight (initial). Mice were euthanized if they met at least 3 of the following criteria: dehydration (evaluated by skin tenting test), lethargy and decreased movement, abnormal posture such as hunching with ruffled fur, pale eyes, loose stools.

Cell culture

The human embryonic kidney (HEK) 293 and murine RAW 264.7 macrophage (American Type Culture Collection, Rockville, MD, USA) were grown in DMEM (BioWhittaker, Walkersville, MD, USA) supplemented with 10% fetal calf serum (Sigma, San Diego, CA), 50 U/mL penicillin, and 50 µg/mL streptomycin (Cellgro Mediatech Inc., Kansas City, MO, USA). Both cell lines tested negative for Mycoplasma and authenticated by the manufacturer.

RNAi

RAW 264.7 Mφ were plated in each well of a 6-well plate which resulted in 30% confluence. Non-target (NT), MICU1, MCUR1, and HSPA8 siRNA were obtained from Dharmacon (Lafayette, CO, USA). In a separate tube, 10 µl HiPerfect (Qiagen, Valencia, CA, USA) was added in 500 µl Opti-MEM with siRNA (25 nM for MICU1, 50 nM for MCUR1, 100 nM for HSPA8) and incubated at room temperature for 10 min. This transfection mixture was applied to each well and incubated for 72 h. The day before the Rhod-2am assay, cells were scraped to reseed in 96-well plate at 1 × 10^5/well. Inhibition of targeted protein expression was assessed by immunoblot of total cell lysate isolated from 6-well plates. In vitro experiments to test LPS stimulation were conducted using 6-well plates. All experiments and cell number determinations were performed in triplicate65,69.

Mitochondrial isolation

Mitochondria were isolated from freshly harvested tissue samples by differential centrifugation, using 4 °C isolation buffer IBI (225 mM mannitol, 75 mM sucrose, 10 mM HEPES, 1 mM EGTA and 0.1% (w/v) fatty acid- free bovine serum albumin (BSA) (pH 7.4) with both phosphatase (Cocktail set II and IV; Calbiochem, USA) and protease (Sigma-Aldrich, USA) inhibitors. The spleen tissue was minced, homogenized, and then centrifuged at 1300 × g for 10 min twice. The supernatant was transferred to a new tube and centrifuged at 3000 × g for 10 min. The crude mitochondrial pellets were washed twice with IBII buffer and then lysed with 2% CHAPS in 1 x cell lysis buffer (Cell Signaling Technology, Danvers, MA, USA) with PMSF. The mitochondrial protein concentration was determined using the Pierce bicinchoninic acid (BCA) Protein Assay kit (Pierce Biotechnology).

Calcium measurements

Free mitochondrial Ca2+ (Ca2+m) was measured using the mitochondrial Ca2+ dye Rhod-2. Peritoneal Mφ were seeded with RPMI-1640 without FBS in a 96-well plate for 2 h at a final concentration of 1 × 10^5 /well. Next, the medium was changed to RPMI-1640 with FBS. Cell lines subjected to RNAi were plated for 24 h. LPS (100 ng/mL) was added for 2 h. The cells were pre-washed with HBSS once followed by incubation with 5 mM Rhod-2 AM for 30 min at 37 °C. They were then washed with HBSS and maintained in HBSS for 30 min at 37 °C. Rhod-2 fluorescence (Ca2+m) was measured at 552exc/581em nm on a Synergy Mx (Biotek) plate reader.

Flow cytometry

Human Peripheral Blood Mononuclear Cells (hPBMC) were washed, then counted and blocked with human Fc block (BD biosciences). 1 × 10^5 cells were stained with fluorescence conjugated antibody (BD biosciences) master mix plus Rhod-2 (Ex/Em 552/581, ThermoFisher Scientific) in PBS supplied with 2% FBS in the dark at 4 °C for 30 min. The cells were washed and labeled with LIVE/DEAD Fixable Aqua Dead Cell Stain dye (ThermoFisher Scientific) in PBS in the dark at 4 °C for 30 min. After the wash, the cells were resuspended in 0.2 mL of PBS supplied with 2% FBS and the data acquired using a BD X-20 flow cytometer. Data were analyzed with Flowjo. The live cells were gated from single cell population. CD14 and CD16 were used to gate out mononuclear cells, CD11c was used to identify the dendritic cells (Supplementary Fig. 5). CD3 was used to identify T lymphocytes. CD19 was used to identify B lymphocytes. Fluorescent intensity of Rhod-2 (EX/Em 552/581) was measured in different cell types in each sample.

Cellular protein extraction

Total tissue or cellular protein was extracted at 4 °C in 1 mL of lysis buffer30,31,33,35. Protein concentration was determined using a (BCA) protein assay (Pierce Biotechnology, Rockford, IL, USA).

Immunoprecipitation

500 µg HEK293 cell lysate or 1 mg tissue lysate were incubated overnight with 10 µL anti-HSPA8 antibody at 4 °C. 25 µL of Protein A/G-Agarose, (Santa Cruz Biotechnology. Dallas, TX, USA) was added to each sample and incubated for 3 h at 4 °C. The samples were spun at 3000 rpm and washed 3 times in lysis buffer. Samples were resuspended in 30 µL 2× SDS-PAGE loading buffer for analysis33,35.

Immunoblot

Total tissue lysate, mitochondrial lysate, or immunoprecipitated protein was electrophoresed in a 10% or 12% SDS-PAGE gel (based on protein of interest) and transferred to a Hybond-enhanced chemiluminescence nitrocellulose membrane (Amersham Pharmacia Biotech, Piscataway, NJ, USA) as previously performed30,31,33,35. Unedited and uncropped original blots are displayed in Supplementary Data 1.

Cytokine concentrations

Plasma TNFα, IL-6 and IL-10 concentrations were quantified using enzyme immunoassay kits (R&D Systems, Minneapolis, MN) that are based on a coated-well, sandwich enzyme immunoassay.

NAD/NADH assay

The NAD/NADH assay kit was purchased from Abcam (Cambridge, MA, USA). All procedure were conducted according to the manufacturer’s instructions.

ATP quantification

The ATP determination kit (Invitrogen, Eugene, OR, USA) was used according to the manufacturer’s instructions. Luminescence was measured using the Soft-MaxPro ATPase Assay program on a Synergy Mx (Biotek) plate reader26,33.

Human subjects

Study design

We conducted a prospective study (January 1, 2019 to December 1, 2022) using human blood sampling to test the hypothesis that the MCU complex is restructured to express reduced MICU1:MCU after sepsis.

Participants

The human participants were at least 18 years of age, admitted to the hospital for inpatient care for one of the following diagnoses:

Group 1a: Sepsis - necrotizing soft tissue infection (NSTI) of the pelvis/perineum requiring ICU care and surgical management.

Group 1b: Not sepsis - traumatic injury to the chest and/or pelvis (e.g., fracture of multiple ribs, pelvic fracture) requiring ICU care

Patients who were immunocompromised or immunosuppressed were not eligible for participation due to an inability to determine the integrity of the immune response. Patients with an estimated survival of < 48 hours were not eligible due to the improbability of achieving (i.e., testing) the primary outcomes.

End points

The primary biological outcome of interest was protein expression of components of the MCU complex in blood mononuclear cells isolated from participants at day 30 after diagnosis. This biological endpoint required the acquisition of a small amount (5 mL) of blood.

SARS-CoV-2

We conducted a retrospective analysis of biobanked PBMCs collected prospectively in a cohort of ICU patients with a confirmed diagnosis of SARS-CoV-2 infection. Deidentified samples graciously were obtained from the Washington University School of Medicine’s COVID-19 biorepository. Similar endpoints as described above were tested.

Statistics and reproducibility

Statistical analyses were performed using Stata 18SE software (College Station, TX, USA). For in vivo animal studies, we powered the studies for the main outcome of changes in protein expression. Guided by our prior animal work, a group size of 12 achieved a power of 0.83. Wilcoxon rank sum/Kruskall-Wallis tests were used as in vivo outcomes tend not to be normally distributed. For in vitro experiments we used groups of n = 8, as the variance is smaller. Median values for each group are represented by a horizontal bar. A p < 0.05 was considered statistically significant. For p values > 0.04 but <0.05, we used the latter term due to an acceptance of two significant digits.

Study approval

We performed all animal experiments in accordance with the National Institutes of Health guidelines under protocols approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh (IS00022935) and Washington University (23-0225). We have complied with all relevant ethical regulations for animal use. The human subject study was approved by the Institutional Review Board (PRO#119120152). Written informed consent was obtained from all patients prior to inclusion. All ethical regulations relevant to human research participants were followed.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.