Inefficient Recovery of Repeatedly Stimulated Memory CD8 T Cells after Polymicrobial Sepsis Induction Leads to Changes in Memory CD8 T Cell Pool Composition

Sepsis is a serious, life-threatening medical syndrome defined by a dysregulated host response to disseminated infection (1). Sepsis affects nearly 50 million people annually, representing almost 20% of all global deaths (2), and it is the leading cause of in-hospital deaths in the United States (3). The economic impact of sepsis is remarkable, as it is the most expensive medical condition in the United States, accounting for more than $24 billion in treatment costs annually (4). Sepsis typically follows a pathophysiological progression that begins with a period of hyperinflammation marked by a dysregulated release of both proinflammatory and anti-inflammatory cytokines in the bloodstream, an increase in circulating neutrophils and monocytes, and the apoptosis of lymphocytes (5–9). Advancements in medical technologies and critical care management have increased survival during this hyperinflammatory period, and, as a result, most septic patients progress into a phase of prolonged immunosuppression, termed sepsis-induced immunoparalysis. The immunoparalysis phase is characterized by decreased overall innate and adaptive immune cell function and can extend for years after the initial septic onset (10–12). This chronic immune dysfunction impairs immunity against novel and previously encountered pathogens as the host is unable to mount a proper immune response (8, 13–16). Changes to the number and function of naive and memory CD8 T cells are major contributors to the sepsis-induced immunoparalysis (10, 15, 17, 18). Sepsis causes the apoptosis and transient numerical decline of naive and memory CD8 T cells, leading to long-lasting changes in phenotype, repertoire, and function of the surviving pool of cells (19). Consequently, sepsis survivors experience an increased incidence of opportunistic secondary infections and reactivation of latent viral infections (20–23). Furthermore, 20% of hospitalized patients with a prior sepsis diagnosis are readmitted to the hospital within 30 d of discharge, primarily due to secondary infections (23).

Sepsis impacts vaccine or infection-induced pre-existing memory CD8 T cell subsets in an indiscriminate fashion, with equal loss of central (CD62L⁺CCR7⁺) and effector memory (CD62L⁻CCR7⁻) T cell populations (19). It also impairs the expansion and functionality of memory CD8 T cells, defined by decreased effector cytokine production after cognate Ag encounter or the ability to respond in a bystander, Ag-independent manner, further defining the chronic sepsis-induced immunoparalysis phase (17). Depending on the severity of the septic event, sepsis can also lead to apoptotic loss of resident memory CD8 T cells, resulting in decreased homosubtypic immunity (8). Finally, data in mice and humans have shown that during postsepsis recovery, memory CD8 T cells become compositionally biased toward a central memory phenotype and sepsis induces long-lasting transcriptomic alterations in the memory CD8 T cell compartment (15, 19). Importantly, these and other studies (8, 17, 24–27) mainly focused analysis on primary (1°) memory CD8 T cell responses, thus leaving the influence of sepsis on repeatedly stimulated (termed higher order) memory CD8 T cell subsets undefined.

Upon pathogen re-encounter (or reinfection), 1° memory CD8 T cells will recognize cognate Ag and facilitate pathogen clearance. Because of Ag recognition and proliferation, 1° memory CD8 T cells will develop into 2° memory CD8 T cells. This process can repeat, and tertiary (3°) and quaternary (4°) memory CD8 T cells can be generated. These higher-order memory CD8 T cells differ substantially from 1° memory cells in phenotype, function, and tissue localization (28–36). For instance, upon each subsequent Ag encounter, there is stepwise change in biology of higher-order memory CD8 T cells characterized by changes in contraction, progression (development) to central memory T cells, and basal proliferation when compared with 1° memory CD8 T cells. In addition, higher-order memory CD8 T cells display enhanced cytotoxicity when compared with 1° memory and, depending on the type of pathogens, can provide better protection to rechallenge (30, 37, 38). Furthermore, transcriptomic analysis of memory CD8 T cells with repeated Ag encounters reveals a stepwise change in the genetic signature of memory CD8 T cells, suggesting that each additional Ag stimulation fundamentally alters the characteristics of the memory CD8 T cell (39). These transcriptomic analyses have also shown that higher-order memory CD8 T cells are more primed to develop into tissue-resident memory populations, due, in part, to a propensity to localize to tertiary tissues (30).

Memory CD8 T cells in humans are constantly reactivated in response to reinfection by pathogens, such as seasonal respiratory viruses (e.g., influenza virus, respiratory syncytial virus, and SARS-CoV2) and parasites (e.g., Plasmodium) (40–42). DNA viruses such as HSV and CMV undergo cycles of latency and reactivation, resulting in repeated cognate Ag exposure and reactivation of memory CD8 T cells (43–46). Memory CD8 T cells make up the bulk (∼70%) of CD8 T cells within human hosts (19), and it is plausible to presume that the human memory CD8 T cell compartment consists of cells with various histories of Ag encounters. Finally, vaccines that employ a “prime-and-boost” protocol have the capacity to generate memory CD8 T cells that were reactivated with cognate Ag multiple times. Because these higher-order memory CD8 T cells are crucial for mediating protection against pathogens that commonly reinfect the host, understanding how sepsis may alter these populations is crucial to defining the chronic immunoparalysis phase that septic patients face.

Experimental procedures using mice were approved by the University of Iowa Animal Care and Use Committee under protocol number 9101915. The experiments performed followed the Office of Laboratory Animal Welfare guidelines and Public Health Service Policy on Humane Care and Use of Laboratory Animals. Inbred C57BL/6 (B6; Thy 1.2/1.2) mice were purchased from the National Cancer Institute (Frederick, MD) and maintained in the animal facilities at the University of Iowa at the appropriate biosafety level. P14 TCR-transgenic mice (Thy 1.1/1.1 or Thy 1.1/1.2) were bred and maintained at the University of Iowa (Iowa City, IA).

Naive P14 CD8 T cells obtained from splenocytes of naive P14 mice (Thy 1.1/1.1 or Thy 1.1/1.2) and adoptively transferred into naive B6 recipients (Thy 1.2; 1 × 10⁴ per mouse) i.v., followed by lymphocytic choriomeningitis virus (LCMV)–Armstrong infection (2 × 10⁵ PFU) i.p. For generation of higher-order memory, P14 CD8 T cells (Thy 1.1/1.2) were obtained from splenocytes of LCMV-immune B6 mice bearing memory P14 CD8 T cells 45 d postinfection and were adoptively transferred into naive B6 recipients (Thy 1.2, 10⁵ per mouse) followed by LCMV infection (2 × 10⁵ PFU) i.p. For mixed transfer experiments, naive P14 (Thy 1.1/1.1, 1 × 10⁴) and 3° P14 (Thy 1.1/1.2, 1 × 10⁵) T cells were coadoptively transferred into naive B6 i.v., followed by LCMV-Armstrong infection (2 × 10⁵ PFU) via i.p. Of note – 10-fold more 3° memory P14 T cells were coadoptively transferred due to their on per-cell basis decreased proliferative ability compared with naive CD8 T cells of the same specificity (39).

Virulent Listeria monocytogenes strain 10403s was grown and 1 × 10⁵ CFU were injected i.v. into sham or cecal ligation and puncture (CLP) hosts at 5 d postsurgery.

PBLs were collected by retro-orbital bleeding. Single-cell suspensions from spleen were generated after mashing tissue through a 70-µm cell strainer without enzymatic digestion. Flow cytometry data were acquired on an LSRFortessa II (BD Biosciences, San Diego, CA) and analyzed with FlowJo software (Tree Star, Ashland, OR). To determine expression of cell surface proteins, mAbs were incubated at 4°C for 20–30 min, and cells were fixed using Cytofix/Cytoperm solution (BD Biosciences) and, in some instances, followed by mAb incubation to detect intracellular proteins. The following mAb clones were used in murine samples: CD8 (53-6.7, eBioscience), Thy 1.1 (HIS51, eBioscience), Thy 1.2 (30-H12, eBioscience), IFN-γ (XMG1.2, eBioscience), TNF-α (MP6-XT22, eBioscience), IL-2 (JES6-5H4, eBioscience), CD62L (MEL-14, eBioscience), CD27 (LG.7F9, eBioscience), KLRG-1 (2F1, eBioscience), CD127 (A7R34, eBioscience), CD122 (5H4, eBioscience), and CX3CR1 (SA011F11, BioLegend). For human samples, the following mAb clones were used: CD3 (OKT3, Tonbo Biosciences), CD8 (HIT8a, Tonbo Biosciences), CCR7 (G043H7, BioLegend), CD45RO (UCHL1, Tonbo Biosciences), CD45RA (HI100, Tonbo Biosciences), and VTE tetramer (MBL International). Ki67 (B56, MOPC-21, BD Biosciences) mAb was used in both human and murine samples. Peptide stimulation was performed by incubating splenocytes in media with 200 nM LCMV peptide gp_33–41 and brefeldin A followed by surface staining, fixation, and permeabilization of the cell membrane using Cytofix/Cytoperm solution (BD Biosciences) and staining for intracellular cytokines. Fixation with Foxp3 fixation/permeabilization (eBioscience) buffer was used to stain Ki67 and BrdU. For BrdU staining, following fixation and permeabilization, cells were treated with DNase I for 1 h at 37°C, then stained for intracellular BrdU (BU20A, eBioscience).

For the detection of in vivo cellular division, BrdU (Sigma-Aldrich, 2 mg/mouse) was injected i.p. daily for 3 consecutive days. Detection of BrdU incorporation was preformed per the manufacturer’s protocol (BD Biosciences) and stained with anti-BrdU as described above.

The CLP procedure was performed on mice that were anesthetized with ketamine/xylazine (University of Iowa, Office of Animal Resources) as previously described (47). Briefly, the abdomen was shaved and disinfected with Betadine (povidone iodine; Purdue Products), and a midline incision was made. The distal third of the cecum was ligated with Perma-Hand silk (Ethicon) and one puncture was made with a 25G needle, and a small amount of fecal matter was extruded. The cecum was returned to the abdomen, the peritoneum was closed with 641G Perma-Hand silk (Ethicon), and skin was sealed using surgical Vetbond (3M). Following surgery, 1 ml of PBS was administered s.c. to provide postsurgery fluid resuscitation. Bupivacaine (Hospira) was administered at the incision site, and flunixin meglumine (Phoenix Pharmaceuticals) was administered for postoperative analgesia. This procedure created a septic state characterized by loss of appetite and body weight, ruffled hair, shivering, diarrhea, and/or periorbital exudates with 0–10% mortality rate for moderate sepsis similar to previous reports (8, 47). Sham mice underwent identical surgery excluding CLP.

Patients were recruited at the University of Iowa Hospitals and Clinics, an 811-bed academic tertiary care center. Blood sample acquisition, patient data collection, and analysis were approved by the University of Iowa Institutional Review Board (ID no. 201804822). Informed consent was obtained from patients or their legally authorized representatives.

Human cell isolation was adjusted from previously described methodology (48). Briefly, whole blood was centrifuged and plasma was removed. ACK (ammonium-chloride-potassium) RBC lysis buffer was then added to the cell pellet and rested for 5 min at room temperature. Cells were again centrifuged, and supernatant was removed. Lysis and centrifugation were repeated one to two additional times. Cells were then washed with PBS three times before being counted and resuspended in cell freeze media (90% FCS [HyClone]/10% DMSO [Fischer Scientific]). Isolated cells were stored at –80°C until use. When used in vitro, PBLs were rapidly thawed and placed into warmed complete media. Cells were then washed three times with warmed media and aggregates were filtered prior to use.

Subjects 18 y of age or older meeting Sepsis-3 criteria for sepsis or septic shock secondary to intra-abdominal infection, soft tissue infection, bloodstream infection, or pneumonia were enrolled. Exclusion criteria were infection requiring antibiotics in the past month, hospitalization for infection in the past year, and chemotherapy or radiation within the past year. Demographics and baseline characteristics including age, sex, race, APACHE II (Acute Physiology and Chronic Health Evaluation II) score, SOFA (Sequential Organ Failure Assessment) score, and presence of septic shock were collected. EDTA-treated blood samples were collected within 24 h of presentation.

Recombinant human (rh)IL-7 (catalog no. 207-IL, R&D Systems, Minneapolis, MN) was stabilized with an anti–IL-7 mAb (M25, Dr. John Harty, Iowa City, IA): 1.3 μg of rhIL-7 was mixed with 6.25 μg of M25 per mouse and incubated at room temperature for 15 min. Sterile saline was added to increase the total volume to 200 μl per mouse and injected i.v. at 24 and 48 h postsurgery.

Unless stated otherwise, data were analyzed using Prism 6 software (GraphPad) using a two-tailed Student t test, one-way ANOVA, and two-way ANOVA with a confidence interval of >95% to determine significance (*p ≤ 0.05 and **p ≤ 0.01). Data are presented as SEM.

Patients who survived the initial hyperinflammatory phase of sepsis typically progress into a state of prolonged immunosuppression, a phase known as immunoparalysis (10, 49–52). Septic survivors who experience immunoparalysis are susceptible to infections that would otherwise be controlled by a healthy immune system and have an increased risk of cancer and viral reactivation, due in part to apoptosis and functional deficits of memory CD8 T cells (20, 21, 23). Understanding the impact of sepsis on memory CD8 T cell subsets is crucial to further defining the chronic immunoparalysis that septic survivors experience. Previous studies have demonstrated the numerical, phenotypic, and functional changes memory CD8 T cells undergo after the initial septic event (8, 10, 17, 27, 53). Interestingly, a study from our laboratory observed that septic patients have increased proliferation in their naive and memory CD8 T cell subsets, presumably in response to the sepsis-induced lymphopenic environment (19). However, these CD8 T cells subsets were examined at a bulk population level, failing to delineate how memory CD8 T cells with defined Ag specificities are influenced by sepsis.

To address whether and to what extent bona fide Ag-specific memory CD8 T cells undergo proliferation after sepsis, we recruited a small number of intensive care unit patients admitted with confirmed sepsis. Peripheral blood of these patients was sampled, and lymphocytes were subjected to flow cytometric analysis. In order to identify virus-specific memory CD8 T cells, lymphocytes were stained with a tetramer specific for the VTE epitope of CMV, a common latent virus that infects up to 50–75% of the population in developed countries (54, 55), and proliferation was assessed via Ki67 expression. We observed proliferation in naive (CCR7⁺CD45RA⁺CD45RO⁻), effector (T_EFF, CCR7⁻CD45RA⁺CD45RO⁻), and memory subsets of CD8 T cells (Fig. 1A), with effector memory (T_EM; CCR7⁻CD45RA⁻CD45RO⁺), central memory (T_CM; CCR7⁺CD45RA⁻CD45RO⁺), and stem cell memory (T_SCM; CCR7⁺CD45RA⁺CD45RO⁺) T cell subsets expressing higher levels of Ki67 compared with naive and effector subsets (Supplemental Fig. 1B), which is consistent with previously published reports (19). Furthermore, we detected CMV-specific memory CD8 T cells within these patients and observed Ki67 expression on these cells (Fig. 1B, 1C). Therefore, virus-specific memory CD8 T cells undergo proliferation after sepsis, presumably in response to the transient lymphopenic environment.

Interestingly, the proliferation of CMV-specific memory CD8 T cells early after sepsis induction was somewhat reduced when compared with other memory CD8 T cell populations (Fig. 1, Supplemental Fig. 1B) (19). This reduced proliferation could be due to the fact that the memory CD8 T cells we examined are specific for CMV, a herpesvirus that undergoes periods of latency and reactivation, resulting in memory CD8 T cell populations that have encountered cognate Ag multiple times (44, 46, 56). It is noteworthy that murine memory CD8 T cell populations from Ag-driven and homeostatic proliferation are diminished in multiple stimulated compared with 1° memory CD8 T cells (37, 39). Thus, these data suggest that sepsis has a capacity to differentially influence recovery of memory CD8 T cell subsets, leading to changes in composition of the memory CD8 T cell pool in sepsis survivors.

Higher-order memory CD8 T cells are present in the memory CD8 T cell compartment as humans are repeatedly infected by common pathogens in which humoral immunity fails and are infected by viruses that undergo periods of latency and reactivation. Furthermore, vaccinations that include secondary and tertiary boosting (e.g., SARS-CoV2 vaccination strategies) induce higher-order memory CD8 T cell populations that provide protection upon pathogen encounter. However, no combination of surface markers can precisely distinguish higher-order memory CD8 T cells from 1° memory CD8 T cells.

To circumvent this problem, we employed a mouse model of serial adoptive transfers of allelically marked virus-specific TCR-transgenic CD8 T cells that has been used by our laboratory and others where memory CD8 T cell populations with different numbers of Ag encounters can be defined (Fig. 2A) (30, 31, 57–60). To generate 1° and 4° memory CD8 T cells within the same host, naive (Thy 1.1/1.1) and tertiary (3°, Thy 1.1/1.2) memory P14 TCR-transgenic CD8 T cells were coadoptively transferred at a 1:10 ratio into congenically distinct naive B6 Thy 1.2/1.2 mice and subsequently infected with LCMV-Armstrong, a pathogen that expresses the cognate gp_33–41 epitope of the P14 TCR. This ratio compensated for the decreased proliferation and survival of 4° compared with 1° memory T cells and allowed enhanced detection at the time of analyses. At 30 d postinfection, 1° and 4° memory CD8 T cells could be identified via Thy allele expression in the circulation of recipient mice (Fig. 2B). Our laboratory and others have described the characteristic phenotypic, functional, and transcriptomic changes that higher-order memory CD8 T cells experience, and in line with these reports, 4° memory CD8 T cells expressed lower levels of CD122, CD27, CD62L, and CD127 and higher levels of KLRG1 and CX3CR1 when compared with 1° memory CD8 T cells (Fig. 2C, 2D) (28, 31, 37, 39). Therefore, generating hosts that contain memory CD8 T cell subsets with a defined history of Ag encounters enables in-depth analyses of the role that sepsis might play in shaping the pre-existing memory CD8 T cell pool.

We and others have shown that pre-existing 1° memory CD8 T cells are impacted numerically, phenotypically, and functionally after the initial septic event (10, 16, 17, 19, 24, 27, 61). One of the main clinical signs of sepsis in human patients is the profound lymphopenia observed in the acute phase of disease (62, 63). To address whether 1° and 4° memory T cells are equally susceptible to sepsis-induced lymphopenia, recipient LCMV-immune B6 mice containing 1° and 4° memory P14 T cells underwent CLP or sham (control) surgery, and blood was sampled 3 d postsurgery. Polymicrobial sepsis induced from CLP resulted in significant morbidity and mortality in comparison with sham surgery (Fig. 3B, 3C). As previously described (15, 17, 62–64), sepsis induced the apoptosis and subsequent numerical decline of circulating lymphocytes and total CD8 T cells (Fig. 3D). Interestingly, we observed an equal fold loss between 1° and 4° memory P14 T cells in the blood (Fig. 3D) and spleen (data not shown). Thus, these data suggest a nondiscriminate nature of sepsis-induced lymphopenia as susceptibilities of 1° and 4° memory CD8 T cell populations were similar.

The immunoparalysis phase of sepsis can be also defined by a change in the functional capacity of memory CD8 T cells, and recently we showed that sepsis diminishes the capacity of 1° memory CD8 T cells to respond to ex vivo cognate Ag stimulation (17). To interrogate whether the function of 4° memory CD8 T cells is influenced to a similar degree as 1° memory CD8 T cells, immune mice bearing both 1° and 4° memory CD8 T cells underwent sham or CLP surgery. At 2 d postsurgery, spleens were harvested for ex vivo gp_33–41 peptide stimulation, and the frequency of responding 1° and 4° memory T cells was assessed. Overall, the response to cognate Ag stimulation was diminished after CLP induction, with a decreased frequency of IFN-γ–producing 1° and 4° memory CD8 T cells in CLP hosts when compared with sham mice (Fig. 4B, Supplemental Fig. 2A). Furthermore, the frequency of 1° and 4° memory CD8 T cells that produced both IFN-γ and TNF (Fig. 4B, Supplemental Fig. 2A) was significantly reduced in CLP hosts. Interestingly, the peptide-stimulated production of IL-2, although compromised in multiple stimulated memory CD8 T cells compared with 1° memory Cd8 T cells (28, 29, 37), was not compromised immediately after sepsis. Finally, polymicrobial sepsis induced the loss of polyfunctionality in 1° and 4° memory CD8 T cells, as the frequency of IFN-γ⁺TNF⁺IL-2⁺ was reduced in both CD8 T cell populations (Fig. 4C). In summary, sepsis equally reduces the Ag-dependent effector functions of both 1° and 4° memory CD8 T cells.

Memory CD8 T cells can also respond to inflammatory cytokines in the absence of cognate Ag by producing effector cytokines (IFN-γ and cytolytic granules), known as the bystander response (65). These bystander responses can mediate immunity to pathogens as well as tumors (27, 33, 34, 66, 67). Following our observations of impaired Ag-dependent memory CD8 T cell function after sepsis, we next asked how the Ag-independent activation and bystander function of higher-order memory CD8 T cells were altered by the septic event. To address this, we infected mice containing both 1° and 4° memory P14 T cells with a virulent strain of L. monocytogenes (strain 10403s, does not express the cognate gp_33-41 Ag) 5 d after sham or CLP surgery. At 1 d postinfection with virulent L. monocytogenes strain 10403s, mice were sacrificed and the frequency of memory CD8 T cells expressing bystander activation markers CD25 and CD69 were measured via flow cytometry. Sepsis resulted in a lower frequency of activated 1° and 4° memory P14 T cells as measured by frequency of cells expressing CD25 and CD69 compared with sham hosts (Fig. 5B, 5C). Furthermore, the ability of 1° and 4° memory P14 cells to produce cytokines was impaired in septic hosts, as both the frequency and number of 1° and 4° memory P14 T cells producing IFN-γ were decreased in CLP mice when compared with sham mice (Fig. 5D, 5E). In summary, sepsis equally reduces the ability of 1° and 4° memory CD8 T cells to respond to infection-induced inflammation.

Sepsis results in a massive release of both proinflammatory and anti-inflammatory cytokines, which ultimately leads to a period of lymphopenia (10, 62). This period is transient, as once the cytokine storm has subsided, T cells (and other immune cell populations) will begin to proliferate in an attempt to refill the empty environment (10). This notion was validated in both human septic patients and septic mice, where proliferation in both naive and 1° memory CD8 T cells occurred after the septic event (19). Importantly, basal proliferation of higher-order pathogen-specific memory CD8 T cells is reduced when compared with 1° memory CD8 T cells (30, 39), and it is unknown whether this difference is also observed in the postsepsis lymphopenic environment (Fig. 1, Supplemental Fig. 1) and how this will influence long-term numerical recovery of various memory CD8 T cell subsets.

To start addressing this, we performed CLP or sham surgery, and proliferation and cycling of 1° and 4° memory CD8 T cells were assessed by measuring BrdU incorporation (68) and Ki67 expression at the indicated days postsurgery (Fig. 6A). In line with previous reports, 1° memory CD8 T cells in both groups incorporated more BrdU and expressed more Ki67 than did their 4° memory counterparts (39). Importantly, we observed that a greater frequency of 1M CD8 T cells from CLP mice had incorporated BrdU and expressed Ki67 when compared to 1M CD8 T cells from sham mice (Fig. 6B–D). However, 4° memory CD8 T cells in the blood and spleen of CLP mice did not uptake significantly more BrdU or express significantly more Ki67 when compared with 4° memory CD8 T cells in sham mice. These data suggest that the decreased intrinsic proliferative capacity of higher-order memory CD8 T cells cannot be reversed after sepsis. In total, sepsis induces proliferation of 1° memory but not 4° memory CD8 T cells, suggesting a diverse capacity of various memory CD8 T cell subsets to repopulate the memory CD8 T cell pool in sepsis survivors.

We next investigated how the disparity in proliferation between 1° and 4° memory CD8 T cells changes the overall composition of the circulating memory CD8 T cell pool after sepsis. To this end, we examined how the representation of 1° and 4° memory CD8 T cells in the circulation changes over time after the initial septic event. Mice containing 1° and 4° memory CD8 T cells underwent sham or CLP surgery with peripheral blood being sampled prior to surgery as well as at 10 and 60 d postsurgery. Although the difference in basal proliferative capacity between 1° and 4° memory T cells resulted in an increase in the 1°/4° memory T cell ratio in both sham and CLP hosts, this difference was amplified following sepsis, presumably due to the increased homeostatic proliferation of 1° memory CD8 T cells in response to sepsis-induced lymphopenia. Compared to mice that underwent sham surgery, CLP mice displayed a significant increase in the circulating 1°/4° memory T cell ratio at day 10, which was further amplified by day 60 (Fig. 7B, 7C, Supplemental Fig. 3B). This shift in the 1°/4° memory T cell ratio reflects an increase in the overall frequency of 1° CD8 T cells after the septic event and an enhanced ability to repopulate the septic-induced lymphopenic environment, due to their intrinsic proliferative capacity. Furthermore, this increased 1°/4° memory T cell ratio was also observed in the spleens of CLP mice when compared with sham mice (Supplemental Fig. 3B). Importantly, the function of 1° and 4° memory CD8 T cells, as measured by production of IFN-γ, TNF, and IL-2 after direct ex vivo peptide stimulation, was restored, pointing to the functional recovery of memory CD8 T cells (Supplemental Fig. 3C). Of note, and in line with our recent study suggesting a sepsis-induced shift toward the central memory T cell phenotype (19), the frequency of 1° memory CD8 T cells producing IL-2 was increased in CLP mice compared with sham controls (Supplemental Fig. 3C).

We next probed other contributing factors to the overrepresentation of 1° memory CD8 T cells in septic mice. Homeostatic cytokines, such as IL-15 and IL-7, are crucial for the long-term survival of both naive and memory CD8 T cells (69–72), so we assessed the expression of CD122 in 1° and 4° memory P14 CD8 T cells. A higher frequency of 1° memory CD8 T cells expressed CD122 than did 4° memory CD8 T cells in both groups, in line with previous reports describing decreased CD122 expression on memory CD8 T cells undergoing Ag stimulation multiple times (Fig. 7D). Similar data are shown for CD127 (IL-7Rα; (Fig. 2C) (37, 39). These data confirm that 1° memory CD8 T cells are better prepared to sense prosurvival cytokine signals than are multiple stimulated memory CD8 T cells, show that sepsis did not change the relative expression of those receptors on 1° and 4° memory P14 CD8 T cells, and suggest underlying mechanisms that control enhanced recovery of 1° over 4° memory CD8 T cells after sepsis.

Finally, we examined the numerical recovery of 1° and 4° memory CD8 T at a late time point after sham or CLP surgery. Quaternary memory CD8 T cells in sham hosts underwent an ∼2-fold numerical loss 60 d after surgery (Fig. 7E), showcasing the reduced ability of 4° memory CD8 T cells to be maintained through basal proliferation induced by prosurvival cytokines signaling, a finding supported by prolonged contraction (or numerical loss) observed in memory CD8 T cells generated in response to multiple stimulations (39). This maintenance deficiency was exaggerated in CLP hosts, as 4° memory CD8 T cells underwent nearly a 5-fold drop in numbers 60 d postsurgery, which is more than double the loss observed in sham mice (Fig. 7E).

In summary, these data show that 4° memory CD8 T cells are unable to numerically recover after the septic event due in combination to the intrinsic proliferative capacity deficiency of repeatedly stimulated memory CD8 T cells and decreased expression of prosurvival cytokine receptors. This, in turn, favors 1° memory CD8 T cells to more efficiently sample homeostatic cytokines available in the lymphopenic environment present during a septic event. Therefore, the composition of the memory CD8 T cell pool after sepsis can be enriched by 1° memory CD8 T cells, potentially resulting in loss of immunity more efficiently provided by higher memory CD8 T cells (32, 38).

Sepsis results in the apoptosis and subsequent numerical loss of T cells, rendering the host susceptible to new and previously encountered infections (50, 52, 73). In an attempt to prevent apoptosis and promote numerical recovery in different clinical settings, rhIL-7 has recently emerged as a promising therapeutic agent (74). IL-7 signals through the heterodimeric IL-7R, which is composed of the IL-7Ra (CD127) and the common γ-chain (CD132) and is expressed by most naive and memory CD8 T cell subsets. IL-7 signals through the Jak3-STAT pathway and results in the upregulation of the anti-apoptotic protein Bcl-2. IL-7 treatment has seen clinical success in patients suffering from idiopathic CD4 T cell lymphopenia, lymphopenia-driven diseases such as AIDS, and is now being employed to treat septic patients (73, 75–79). Previous studies have shown that septic mice treated with IL-7 had significantly improved survival, as well as increased T cell viability and function compared with mock-treated mice (80, 81). Thus, we asked how IL-7 treatment influences the recovery of 1° and 4° CD8 T cells after the septic event.

To address this question, mice containing 1° and 4° memory P14 cells underwent sham or CLP surgery at 30 d postinfection and were treated with rhIL-7/anti–IL-7 mAb complexes (IL-7cs) or PBS at 24 and 48 h postsurgery. Twelve days after surgery, mice were sacrificed and cellular composition in the blood and the spleen was assessed. As expected, IL-7c administration resulted in significantly increased numbers of circulating 1° memory CD8 T cells compared with PBS controls in both sham and CLP hosts, albeit with 1° memory CD8 T cells in CLP mice expanding to a lesser degree than in sham mice. Interestingly, IL-7c treatment did not rescue the numbers of 4° memory CD8 T cells in septic hosts, whereas sham hosts displayed an ∼2-fold increase in circulating 4° memory CD8 T cell numbers (Fig. 8B). When examining the spleen, IL-7c significantly increased the number of 1° memory CD8 T cells in both sham and CLP hosts and 4° memory CD8 T cells in the spleens of sham mice. We observed that 4° memory CD8 T cells in the spleens of CLP hosts had not expanded, recapitulating the results seen in the blood of these mice (Fig. 8C). Thus, these data suggest that 4° memory CD8 T cells are unable to efficiently respond to IL-7c treatment compared with 1° memory CD8 T cells. The failure of 4° memory CD8 T cells to respond to IL-7c treatment in septic animals lends important considerations for septic patients treated with IL-7, as higher-order memory CD8 T cells may not be recovered and alternative approaches (such as revaccination or boosting) might be envisioned to regenerate these populations of protective memory CD8 T cells.

One of the canonical features of sepsis is the dramatic (but transient) reduction of lymphocytes, which is a major factor contributing to the increased susceptibility of sepsis patients to secondary infection. Lymphocyte numbers eventually recover, but there is limited knowledge of the impact of sepsis on naive and memory T cells. In the current study, we first demonstrate that sepsis induces relatively low levels of proliferation of virus-specific memory CD8 T cells in human hosts. The Ag-specific memory CD8 T cells examined were specific for CMV, a virus that undergoes periods of latency and reactivation. Repeated infection by the same pathogen and/or prime/boost vaccination protocols can lead to the generation of higher-order memory T cell subsets, with it being possible for differences in the number of times they have been activated after cognate Ag recognition. Because of the lack of defined phenotypic surface markers that delineate primary and higher-order memory CD8 T cells in humans, we used a murine adoptive transfer system to model and compare the influence of sepsis on memory CD8 T cells with a defined and varied number of cognate Ag encounters. Using this model, we observed that sepsis results in equal loss of 1° and 4° memory CD8 T cells. These data support the notion that the sepsis-induced numerical loss of CD8 T cells is indiscriminate and stochastic. We also found that sepsis induces similar changes in the Ag-dependent and -independent functions of memory CD8 T cell subsets, precisely defining the capacity of sepsis to influence this arm of adaptive immunity. The most interesting finding was the inefficient recovery of multiple-stimulated (in this case, 4°) memory CD8 T cells after sepsis that correlates with decreased intrinsic basal proliferative capacity compared with primary memory CD8 T cells with the same Ag specificity. This differential recovery seen in 1° and 4° memory CD8 T cells led to long-lasting changes in the composition of the total memory CD8 T cell pool in sepsis survivors. Finally, we showed that higher-order (4°) memory CD8 T cells were unable to efficiently respond and numerically recover following IL-7c treatment compared with 1° memory CD8 T cells, an important implication for ongoing trials using IL-7 therapy in septic patients (78, 79).

Our collective data have important implications for further defining the period of sepsis-induced immunoparalysis that sepsis patients experience. Memory CD8 T cells generated after single or multiple Ag encounters (after repeated infections or vaccinations) are susceptible to numerical and functional demise early after sepsis induction to an equal degree, rendering the host diminished protection to repeated pathogen encounter. These results suggest that those patients who recover from a septic event may harbor a higher proportion of primary memory CD8 T cells than higher-order memory CD8 T cells, which can have potentially significant ramifications for future immunity and the extent of protection afforded by different memory CD8 T cell subsets (30, 37, 38). A previous report from our laboratory demonstrated how 1° memory CD8 T cells from CLP mice are impaired in their ability to provide protection upon secondary challenge (19). However, this study analyzed the protective capacity of 1° memory CD8 T cells responding to only one type of bacterial pathogen (specifically, L. monocytogenes) and did not address potential deficits in response to other pathogens. The history of Ag encounters could determine the localization and protective capacity of memory T cells; however, dissecting the protective capacity of higher-order memory CD8 T cells to different pathogens, as standalone subsets or as part of the memory pool that could consist of multiple memory CD8 T cell subsets, remains unknown and represents a subject for future investigation (38). It is interesting to speculate on how higher-order memory CD8 T cells, which may be preferentially localized in peripheral tissues, may be altered in their ability to protect the septic survivor upon infections with various pathogens.

In addition, higher-order memory CD8 T cells have a preponderance to develop into a resident memory T cell population, and the preferential loss of these populations may have implications in the development and maintenance of future and current resident memory CD8 T cell responses (30, 82). CD8 T cell localization into peripheral tissues increases with additional Ag encounter (38, 39); however, we did not analyze the localization of memory populations after sepsis. It is possible that higher-order memory cells are increased in peripheral tissues, contributing to their decreased representation in the circulation. This is something that could be explored in future experiments aimed at dissecting whether and how sepsis alters memory CD8 T cell localization. Finally, we observed a direct correlation between CD122 and CD127 expression on memory CD8 T cells generated in response to one or multiple Ag stimulations and the ability of the different memory CD8 T cell populations to respond to homeostatic cytokines (e.g., IL-7 and IL-15) in the sepsis-induced lymphopenic environment.

Another interesting aspect of our study is the finding that higher-order memory CD8 T cells differentially respond to IL-7c therapy due to baseline differences in their ability to proliferate and sense prosurvival signals. IL-7 treatment induces T cell proliferation, but it also prevents apoptosis through the upregulation of the anti-apoptotic molecule Bcl-2, which sustains the overall number of CD8 T cells (80, 83). However, the effect of IL-7 treatment on higher-order memory CD8 T cells is muted because repeated Ag stimulation results in decreased basal proliferative capacity. Consequently, IL-7 therapy preferentially leads to the recovery of 1° memory CD8 T cells, likely due to increased IL-7R expression and basal proliferative capacity. The preferential rescue of primary memory CD8 T cells may have important implications in host immunity against common pathogens. It is interesting to speculate on the ability of these higher-order memory CD8 T cells to respond to other signals including IL-2 or IL-15 cytokine treatments (84–87), and future studies should examine the extent to which higher-order memory CD8 T cells can be rescued numerically after the septic event. Because IL-7 has been shown to recover memory CD8 T cell numbers in clinical trials treating septic patients (79), our data suggest a preferential recovery of primary memory CD8 T cells in these hosts. However, due to the lack of defined phenotypic markers that differentiate between primary and higher-order memory CD8 T cells, it is difficult to directly compare the effect of rhIL-7 treatment on primary and higher-order memory CD8 T cells in humans. Furthermore, our study focuses on memory CD8 T cell responses in the circulation and does not address how higher-order resident memory CD8 T cell populations may be impacted by sepsis. Many studies have shown how repeated Ag encounters alter the CD8 resident memory T cell pool, especially in the lung and lung-draining lymph nodes (30–32). Severe sepsis can detrimentally impact the number of resident memory CD8 T cells, so it would be interesting to examine whether and how severe sepsis impacts higher-order resident memory CD8 T cell populations.

Overall, this study helps further define the chronic immunoparalysis phase that most septic survivors experience. Moreover, the data presented extend the current analysis to focus on higher-order memory CD8 T cell populations that are more relevant to human populations. Because of the frequent use of prime/boost vaccination approaches and the prevalence of pathogens capable of reinfecting humans, the crucial role played by higher-order memory CD8 T cells in eliciting protection against infection and disease is becoming more appreciated. In this study, we also dissect the ability of memory CD8 T cells to respond to cytokine treatment and add to existing knowledge on which cells are capable of sensing and responding to in vivo therapeutic intervention. In total, this study helps us further define sepsis-induced immunoparalysis and sheds new light into how repeated Ag encounter may influence the ability of the immune system (and ultimately the host) respond to and recover from a septic event.

We thank members of our laboratories for technical assistance and helpful discussions.

The authors have no financial conflicts of interest.