Unmasking Unique Immune Altering Aspects of the Microbiome as a Tool to Correct Sepsis-Induced Immune Dysfunction

Abstract

Background: Sepsis-related mortality is driven by immune dysfunction. A bidirectional micro-organism–immune cell cross talks exists. Gut Bacteroides fragilis–T-cell crosstalk maintains innate immune cell/pathogen homeostasis. Commensal gut Clostridia spp. suppress inflammation and induce gut tolerance. Probiotics are administered to restore immune microbiome homeostasis. Individual microbial components have an immunomodulatory effect. However, probiotic therapies for sepsis-induced immune disruptions are rarely tailored to specific immune responses. Thus, we ask the question as to how components of the intestinal microbiome, often found in probiotic therapies, affect lymphocyte phenotypic profile?

Methods: T-lymphocytes were cultured with either monomicrobial or polymicrobial combinations. Microbes used were Bacteroides fragilis, Clostridium perfringens, or Lactobacillus acidophilus. Cytokines, measured by enzyme-linked immunosorbent assay (ELISA)-included interleukin (IL)-6, IL-10, IL-22, and IL-33. Flow cytometry was used for T-cell phenotyping for program-death receptor-1 (PD-1) and B- and T-lymphocyte attenuator (BTLA). T-cell DNA was extracted to assess global epigenetic changes. For translation, IL-33 was measured from surgical intensive care unit (ICU) patients with sepsis with either monomicrobial or polymicrobial infection.

Results: Lactobacillus consistently induced IL-22 and IL-33. Bacteroides fragilis induced IL-33 only under polymicrobial (pB) conditions. Within surgical ICU patients, IL-33 levels were higher in polymicrobial versus monomicrobial patients. PD-1⁺ expression was lowest with either monomicrobial Bacteroides fragilis or Bacteroides fragilis predominant polymicrobial context. Conversely Bacteroides fragilis exposure induced a distinct PD-1^-high subpopulation. B- and T-lymphocyte attenuator-positive expression did not differ after individual microbes. Among polymicrobial conditions, Bacteroides fragilis predominant (pB) and Lactobacillus acidophilus predominant (pL) increased BTLA⁺ expression. DNA methylation was most increased in response to Clostridium perfringens in monomicrobial and in response to Bacteroides fragilis in polymicrobial conditions.

Conclusion: Unique microbe/lymphocyte interactions occur. Bacteroides fragilis induced a T-cell phenotype consistent with potential long-term immune recovery. This work begins to discover how varying microbes may induce unique functional and phenotypic T-lymphocyte responses.

Sepsis remains a devastating clinical condition with profound short- and long-term effects. The recent re-definition of sepsis emphasizes the often destructive effect of the body's own response to an infecting organism [1]. A dysregulated immune response to an infection is responsible for most of the clinical findings among patients with sepsis. The onset of sepsis induces a profound inflammatory and immunosuppressive response [2,3]. T-lymphocytes play key roles in the outcomes and survival from sepsis seen in both murine models as well as in critically ill surgical patients [4,5]. The lymphocyte response to sepsis is, in part, regulated via a series of checkpoint proteins including program-death receptor-1 (PD-1) and B- and T-lymphocyte attenuator (BTLA) across both murine models and septic adult and neonatal patients [6]. A newly described PD-1^-high lymphocyte subpopulation is recognized as being highly active and associated with a restorative and protective profile across both acute and chronic conditions [7,8].

The intestinal microbiome, comprising the collection of micro-organisms including bacteria, viruses, phages, and fungi as well as the associated genes and gene products found in the intestine is both beneficial and essential to maintaining health homeostasis. The dominant phyla within the human intestinal microbiota are Firmicutes, Bacteroidetes, Actinobacteria, Proteobacteria, and Verrucomicrobia. Among the microbiome, two of the most important species are the Bacteroides species and Clostridial species from the Firmicutes phylum [9]. The microbiome has often been considered to exert a passive influence on the immune system. An altered immune system may allow the emergence of virulent organisms or allow the transition of opportunistic to pathogenic micro-organisms. Recently, however, the microbiome has been shown to have an active impact on the immune system, and many believe that the microbiome has a greater influence on the adaptive [10] rather than the innate immune system. Crosstalk has been shown to exist between the microbiome and the adaptive immune system with respect to cytokine production, cell phenotype, and potentially long-term effects. Specifically, Bacteroides fragilis has been shown to be capable of engaging in a microbe/immune system crosstalk with intestinal lymphocytes mediated in part via interleukin (IL)-13 and IL-33 [11,12]. Commensal gut Clostridia spp. may suppress inflammation and induce gut tolerance via IL-22 [13–15]. Lactobacillus spp. induces differing helper T cell (T_H) T_H1/T_H2 responses from lymphocytes derived from healthy versus unhealthy individuals [16].

Probiotics, specifically Lactobacillus spp. dominant [16], are increasingly being administered to patients before expected major surgical procedures or to patients recovering from septic events with presumed altered microbiomes. However, probiotic therapies are neither patient- nor immune-dysfunction specific. Our understanding of how specific components of the microbiome or probiotic therapies affect lymphocyte phenotype and function remain largely unknown. Given the important role for lymphocytes as well as the microbiome in sepsis recovery, we therefore ask how components of the intestinal microbiome, often found in probiotic therapies, affect lymphocyte phenotypic profile?

Materials and Methods

Cells, microbes, and culture technique

All cells and micro-organisms were obtained from American Tissue Culture Collection (ATCC, Manassas, VA). All microbe work was undertaken in a BSL2 laboratory. Micro-organisms included Bacteroides fragilis, Clostridium perfringens, and Lactobacillus acidophilus (ATCC). These microbes were selected for several reasons. These pathogens best reflect the predominant phyla within the intestinal microbiome wherein Firmicutes and Bacteroidetes constitute 90% of the gut microbiota. Broad-spectrum antibiotic agents lead to an imbalance between Firmicutes and Bacteroidetes. It has been demonstrated that these microbes play an immunoregulatory role on T-lymphocytes, specifically Bacteroides. Last, Lactobacillus is among the probiotics most commonly used. Microbial culture methods were adapted from polymicrobial and anaerobic methods described by Gabrilska et al. [17] and others [18,19]. Two sets of experiments, monomicrobial and polymicrobial, were conducted. For monomicrobial experiments, 10⁵ colony forming units (CFU) of individual bacteria were used. For polymicrobial experiments, three separate environments were created, either Bacteroides dominant (pB), Clostridium dominant (pC), or Lactobacillus dominant (pL), To create the polymicrobial cultures, 10⁵ CFU of the dominant organism were combined with 10³ of the other two organisms to create the final microbial mixture. T-lymphocytes were obtained from ATCC (HB-11052). All cell line experiments were undertaken within seven passages of the original cell line; 10⁵ T-cells were used for all experiments.

Cytokine production

After experimentation, the supernatant was extracted for cytokine analysis to include IL-6, IL-10, IL-22, and IL-33. Measurement was conducted using enzyme-linked immunosorbent assay (ELISA) kits obtained from BD Pharmingen (San Diego, CA; IL-6 and IL-10) or from Abcam (Cambridge, MA; IL-22 and IL-33). Enzyme-linked immunosorbent assay kits were used according to the manufacturer's recommendations.

Antibody phenotyping of T-cells

Phenotying of the cells was assessed using flow cytometry. Monoclonal antibodies were used according to the manufacturer's recommendations and as described previously [20,21]. After experimentation, samples were centrifuged and washed initially at 1,400 rpm. The resultant pellet was re-suspended and re-centrifuged at 400 rpm for five minutes to separate T-cells from bacteria. This was repeated and the resultant T-cell pellet was then re-suspended and stained with monoclonal antibodies (eBioscience, San Diego, CA) to FITC-CD3 with APC-CD69 and APC-BTLA or antibodies to APC-CD3 and PE-PD-1 (eBioscience). Gating for live cells was undertaken using forward scatter (FSC) and side scatter (SSC). Gating was then undertaken using appropriate isotype controls, and data was analyzed using FlowJo V10 software (FlowJo LLc, Ashland, OR).

Epigenetic analysis

After culturing and cell isolation as described above, T-cells were prepped for DNA extraction using the Quick-DNA miniprep plus kit (Zymo Research, Irvine, CA) according to the manufacturer's recommendations. Three milliliters of BioFluid and cell buffer along with 100 mcL of proteinase K were added to the re-suspended cell sample and incubated at 55°C for two hours. Next, genomic lysis buffer was added. The lysate was centrifuged for five minutes at 1,000g and washed with g-DNA wash buffer. Two hundred microliters of DNA elution buffer were added and incubated for five minutes, then centrifuged to elute the DNA and stored at −20°C for future analysis.

To assess degree of DNA methylation the MethylFlash™ Global DNA Methylation (5-mC) ELISA Easy Kit (Epigentek Group, Farmingdale, NY) was used according to the manufacturer's recommendations. Positive and negative controls were used to generate standard curves. For samples, 100 ng of DNA from each T-cell sample was added to the 96-plate pre-coated wells and incubated together for one hour. After washing, 50 mcL of 5-mC detection complex solution was added to each well and incubated. One hundred microliters of detection solution were added and the plate was read at 450 nm.

To calculate the percentage of methylated DNA, the slope of the standard curve was calculated. The percentage of methylated DNA/5-methylated cytosine (5-mC) in total DNA was calculated using the following formula:

5-mC% = ([Sample optical density – Negative control optical density]/(standard curve slope * 100 [concentration of DNA added]) * (100%). Given that this kit measures methylated cytosine only and that cytosine constitutes 21% of human DNA, the resultant value is then divided by 0.21 to give the total DNA methylation.

Patients

Critically ill surgical patients with sepsis were enrolled prospectively from the intensive care unit (IRB #211087-6). Exclusion criteria included patients who died within 48 hours of presentation, pregnancy, patients receiving steroid treatment at the time of blood draw, or any patient with a known history of lymphoma or leukemia. All blood draws occurred between 12 and 24 hours after presentation to the intensive care unit (ICU). Charts were reviewed for patient characteristics including age, gender, and primary diagnoses, associated medical comorbidities and mortality, and primary source of sepsis. Charts were reviewed for microbiology culture data and patients were grouped into those with monomicrobial infection versus those with polymicrobial infection. Charts were also reviewed for appropriate clinical and biochemical factors that would be required to calculate Acute Physiology and Chronic Health Evaluation (APACHE) II score. The APACHE II calculation occurred at the same time as the blood was drawn for each patient. For patients with abdominal or necrotizing soft tissue infection as the source of sepsis, blood draws occurred within 24 hours of initial presentation to the hospital. For patients with pneumonia, blood draws occurred within 24 hours of conducting the bronchoalveolar lavage for sample collection for diagnosis of pneumonia occurring at an average of sixth hospital day. Blood was drawn and stored at −80°C for later analysis of IL-33 level via ELISA technique as descrived previously [21].

Statistics

Chi-squared analysis was used for categorical data. Student t-test was used for continuous variables across two groups. Analysis of variance (ANOVA) was used for comparison of continuous variables across multiple groups with Holm-Sidak post hoc analysis. Continuous data are reported as mean ± standard error of the mean (SEM). Data was analyzed using SigmaPlot 12.5 (Systat Software, San Jose, CA). Significance was set at p < 0.05.

Results

Cytokine response

We reviewed IL-6 as a marker of a proinflammatory response and IL-10 as the anti-inflammatory response. Interleukin-22 is a cytokine involved in mediating the destructive component of a lymphocyte response to stress/infection. Interleukin-33 plays a role in restoring intestinal epithelium though key crosstalk interaction with intestinal lymphocytes after an infection.

The response within a monomicrobial environment (cells alone versus T-cells combined with Lactobacillus acidophilus versus Bacteroides fragilis versus Clostridium perfringens) was first reviewed. Compared with T-cells alone, IL-6 levels were elevated after exposure to both Lactobacillus acidophilus and Clostridium perfringens but were not elevated in response to Bacteroides fragilis (29 vs 310 vs 130 vs 335 pg/mL; p < 0.05). Interleukin-10 levels were uniformly elevated in response with no difference across microbial exposure (3.8 vs 12.8 vs 10.8 vs 15.6 pg/mL; p > 0.05 between microbe groups). Interleukin-22 increased in response to all micro-organisms but was most markedly elevated in response to Lactobacillus acidophilus (10.5 vs 43.6 vs 22.9 vs 24.9 pg/mL; p < 0.05). Interleukin-33 was elevated with both Lactobacillus acidophilus and most notably in response to Clostridium perfringens but was not elevated in response to Bacteroides fragilis (8.5 vs 28.7 vs 10.3 vs 47.9 pg/mL; p < 0.05; Fig. 1).

FIG. 1.

Distinct cytokine production, IL-6 (A), IL-10 (B), IL-22 (C), and IL-33 (D) from T-cells cultured with monomicrobial microbiome components. IL = interleukin; Lac = Lactobacillus acidophilus; Bact = Bacteroides fragilis; Clos = Clostridium perfringens.

Next, the response within polymicrobial environments was assessed comparing cells alone versus Lactobacillus acidophilus dominant (pL) versus Bacteroides fragilis dominant (pB) versus Clostridium perfringens dominant (pC). With respect to both IL-6 and IL-10, rather than any individualized cytokine response to differing polymicrobial environments, it was noted that each polymicrobial condition induced uniform elevations in IL-6 and IL-10 (Fig. 2A and 2B). Both pL and pB induced elevated IL-22 levels, with levels being highest in an pL environment; IL-22 levels were not elevated in the pC environment (10.5 vs 64.6 vs 35.1 vs 20.7 pg/mL; p < 0.05 pL and pB vs controls; Fig. 2C). With respect to IL-33 production, although levels were increased across all groups, it was noted that IL-33 was most markedly elevated when Bacteroides fragilis predominated (7.9 vs 31.2 vs 67 vs 27 pg/mL; p < 0.05 all groups vs control; Fig. 2D).

FIG. 2.

Among polymicrobial environments no difference was noted in IL-6 (A) or IL-10 (B) production, however, IL-22 (C) was most elevated from pL and IL-33 (D) most elevated from pB environments. IL = interleukin; pL = Lactobacillus dominant; pB = Bacteroides dominant; pC = Clostridium dominant.

To assess for a potential translatable association, critically ill patients with sepsis for whom source and microbiologic data were identified were enrolled from the surgical ICU. Patients were divided into those with known monomicrobial versus polymicrobial infections. Nine patients were enrolled, four with monomicrobial infections and five with polymicrobial infections. Although a small cohort, patients were matched with respect to age (57.8 ± 6.9 vs 51 ± 6.4 years; p = 0.5), female sex (50% vs 40%; p = 1), and APACHE II score (11.8 ± 2.5 vs 14 ± 2.4; p = 0.54). Pneumonia was the most common source of infection among monomicrobial patients (Fig. 3A). Among patients with monomicrobial infection, the microbiology included Clostridium difficile (abdominal), Haemophilus influenza and methicillin-sensitive Staphylococcus aureus (MSSA; pneumonia) and methicillin-resistant Staphylococcus aureus (MRSA; abscess). Among patients with polymicrobial infections, the microbiology included polymicrobial from perforated viscus (abdominal), Escherichia coli/MSSA and Haemophilus influenza/MRSA (pneumonia) and Streptococcus/Proteus (abscess/necrotizing soft tissue infection). Compared with monomicrobial patients, IL-33 levels were higher in patients with polymicrobial infection (28.1 vs 50.2 pg/mL; p = 0.017; Fig. 3B).

FIG. 3.

(A) Characteristics of critically ill surgical ICU patients with monomicrobial versus polymicrobial infections. (B) IL-33 levels were higher in patients with polymicrobial infections. ICU = intensive care unit; IL = interleukin; APACHE II = Acute Physiology and Chronic Health Evaluation.

Lymphocyte phenotype

Flow cytometry assessed lymphocyte phenotype including CD69 (lymphocyte activation), as well as checkpoint proteins BTLA (a negative regulator inducing anergy) and PD-1 (coinhibitory/costimulatory receptor). The PD-1^-high lymphocytes are highly activated, non-anergic cells. With respect to T-cell activation, all microbial exposures, within both monomicrobial and polymicrobial conditions, uniformly elevated CD69⁺/CD3⁺ expression ranging between 75% and 85% with no differences between groups (data not shown). Within monomicrobial environments, all individual microbes induced increased BTLA⁺/CD3⁺ expression, with no difference between groups (1.9% vs 7.9% vs 10.8% vs 7.4%; p > 0.05 between groups; Fig. 4B). However, within polymicrobial environments, pL failed to induce an increase in BTLA expression, whereas both Bacteroides fragilis- and Clostridium perfringens-predominant polymicrobial environments increased BTLA⁺/CD3⁺ expression (1.9% vs 5.6% vs 23.4% vs 21.5%; p < 0.05 for pB and pC vs control; Fig. 4C).

FIG. 4.

(A) Representative flow cytometry plots of BTLA⁺/CD3⁺ cells with varieties of polymicrobial cultures. (B) After culture with individual microbes (monomicrobial) there was no difference in BTLA⁺/CD3⁺ expression between groups. (C) CD3⁺ lymphocytes after culture with combinations that predominated in either Lactobacillus (pL), Bacteroides (pB), or Clostridium (pC), it was noted that both pB and pC led to increased BTLA⁺/CD3⁺ expression, but not pL. BTLA⁺ = B- and T-lymphocyte attenuator-positive.

Expression of PD-1 increased in response to all microbial environments, whether monomicrobial (16.4% vs 83.4% vs 64.6% vs 86.8%; p < 0.05 all groups vs control), or polymicrobial environments (16.4% vs 86.1% vs 93.4%; p < 0.05 all groups vs control). However, it was noted that compared with either Lactobacillus acidophilus or Clostridium perfringens, PD-1⁺ expression was lower in Bacteroides fragilis conditions whether those were monomicrobial or polymicrobial (Fig. 5A and 5B). Among PD-1⁺/CD3⁺ cells, most interestingly, Bacteroides fragilis induced the emergence of a distinct PD-1^-high subpopulation as demonstrated in the representative flow cytometry plots (Fig. 6A). Bacteroides fragilis induced the PD-1^-high response to both monomicrobial Bacteroides fragilis (20.6%) as well as Bacteroides fragilis poly-microbial (pB) conditions (16.3%; Fig. 6B) compared with other populations (∼2%).

FIG. 5.

PD-1 = program-death receptor-1 (PD-1) expression increased in response to all microbial environments. However, the increased PD-1 expression on T-cells was lowest in (A) monomicrobial Bacteroides fragilis or (B) Bacteroides fragilis predominant (pB) polymicrobial conditions.

FIG. 6.

(A) Representative flow cytometry plots of PD-1^-high/CD3⁺ cells with varieties of polymicrobial cultures. (B) PD-1^-high expression occurred in response to Bacteroides fragilis in both monomibrobial and polymicrobial (pB) conditions.

Epigenetic changes

We reviewed DNA methylation given recent evidence for a its role in chromatin modification in repeated infected T-lymphocytes [22] as well as in regulating checkpoint proteins [23]. We examined global T-cell DNA methylation via ELISA with antibodies directed at 5-mC. Within monomicrobial conditions, DNA methylation was noted to be increased with all microbe exposures. However, this increase was most marked in response to Clostridium perfringens (0.4% vs 1.5% vs 1.4% vs 5.2%; p < 0.05 Clostridium perfringens to all others; Fig. 7A). Within polymicrobial conditions, although percent methylation was again noted to be increased in response to Clostridium perfringens, methylation changes were most marked in Bacteroides fragilis-predominant polymicrobial conditions (pB) (0.4% vs 1.4% vs 4.5% vs 3.1% ; p < 0.05 Bacteroides fragilis and Clostridium perfringens vs control; Fig. 7B).

FIG. 7.

Differential DNA methylation was noted across differing microbial environments. Epigenetic changes were most marked with (A) Clostridium perfringens in monomicrobial (Clos) as well as with (B) Bacteroides fragilis in polymicrobial conditions (pB).

Discussion

Critically ill patients display a profoundly dysfunctional and anergic immune system as well as a disrupted microbiome, both of which contribute to both short- and long-term effects of sepsis [24]. We describe unique immunomodulating features of components of the microbiome. Lactobacillus acidophilus is associated with increased IL-22 response, whereas Bacteroides fragilis induces a marked IL-33 response in polymicrobial conditions. The expression of PD-1 was lowest in response to Bacteroides fragilis. However, the protective and restorative PD-1^-high T-cell profile emerged only in response to Bacteroides fragilis. Last, epigenetic modifications were most pronounced in Bacteroides fragilis environments. We propose a potentially potent immune modulating effect may be occurring in response to specific microbes. T-lymphocytes are imperative to the short- and long-term response to sepsis [25]. Loss of lymphocytes is associated with increased mortality rates [26], in part by regulating intestinal epithelial apoptosis and intestinal destruction after abdominal sepsis [27]. Importantly, the role of lymphocytes in response to acute septic and sterile critical injury has also been demonstrated in patients, wherein lymphocyte dysfunction has been associated with an increased risk of mortality [4].

Both IL-22 and IL-33 play unique roles in immune crosstalk at the intestinal epithelial level. Interleukin-22 is a member of the IL-10 family. T-cells produce IL-22, most markedly by intestinal lymphocytes [28]. Tsai et al. [29] demonstrated that intestinal IL-22 contributes to diarrhea and enteric pathogen clearance. Hou et al. [15] demonstrated that Lactobacillus reuteri was capable of stimulating IL-22 secretion and the Lactobacillus reuteri–IL-22 combination led to increased intestinal stem cell growth, ultimately leading to protection and restoration of the intestinal mucosa. This work, utilizing a monomicrobial infection within an intestinal organoid, is in keeping with our observations wherein in both a monomicrobial and a polymicrobial environment Lactobacillus acidophilus induced elevated levels of IL-22. Conversely, a destructive role for IL-22 has also been identified. Elevated IL-22–producing T-cell subsets have been associated with the development and progression of Crohn inflammatory bowel disease [30].

Interleukin-33 plays critical roles in repairing the epithelium of the intestinal tract [31] as well as mediating the bacteria/immune cell crosstalk. When innate lymphoid cells (ILCs) and Bacteroides fragilis crosstalk, with the interaction of IL-33, T-cells have been shown to be able to direct and organize a variety of both innate and acquired immune cells within the intestinal tract [32]. Several authors speculate that this IL-33–mediated crosstalk functions as a surveillance system wherein Bacteroides spp. may warn intestinal lymphocytes of potentially harmful microbiome alterations.

We have demonstrated that specific bacteria or combinations of bacteria induce unique immune and inflammatory profiles. Among a variety of potential mechanistic steps to explain our findings further, microbial by-products offer some intriguing insights. Short-chain fatty acids (SCFAs) are bacterial fermentation products found in high concentration within the intestinal tract and exhibit immune modulatory effects [33]. They have the potential to regulate T-cells within mucosal populations. One of the most studied SCFA is butyrate and can be produced by both Bacteroides and Clostridium spp. Butyrate has beneficial effects on intestinal homeostasis, mainly through the modulation of immune responses and inflammation. Although butyrate exhibits proinflammatory and anti-inflammatory effects, it is considered to function mostly as an agent of homeostasis [34]. In keeping with our data demonstrating Clostridium perfringenes induced increased T-cells IL-10 production, Singh et al. [35] demonstrated that butyrate promotes activation of inflammatory pathways, resulting in the differentiation of regulatory T-cells and IL-10 producing T-cells. Bacteroides production of butyrate plays a role in controlling cellular process including cellular chemotaxis, differential and proliferation [36], as well as inhibiting histone acetylation and altering gene expression [33,37]. Butyrate acts as a histone deacetylase (HDAC) inhibitor, preventing histone deacetylation and affecting gene expression. Kespohl et al. [38] demonstrated a dose-dependent effect on butyrate on localized versus global methylation within regulatory T-cells. In keeping with our data on the effects induced by Bacteroides on DNA methylation among T-cells, Smith et al., [34] demonstrated that butyrate can induce regulation of colonic T-cells via epigenetic upregulation of the Foxp3 gene. Kim et al. [39] demonstrated recently that the microbes from patients with sepsis may directly suppress an appropriate immune response to infection through host transcriptome alterations required for pathogen clearance. Furthermore, the authors demonstrated that a fecal microbiome transplant was able to restore an immunoprotective profile, in part through the expansion of Bacteroidetes with restoration of normal butyrate levels [39]. Future and ongoing work will target SCFAs, and specifically butyrate, to uncover mechanisms of our current findings.

Checkpoint proteins, including BTLA and PD-1 play important roles in regulating a wide variety of T-cell responses within diverse disease states. Both BTLA and PD-1 have been implicated in driving the lymphocyte anergy and dysfunction seen in both murine models as well as patients with sepsis [40]. Initially considered to primarily function as coinhibitory receptors, there are many co-stimulatory features associated with PD-1. As secondary signaling mechanisms, these checkpoint proteins regulate T-cell responses to acute antigen challenges, including those at the intestinal epithelial level [41] and, in keeping with our data, PD-1 appears to have roles in immune/microbe interactions.

Among PD-1⁺/CD3⁺ cells, it is becoming evident that PD-1^-high lymphocytes constitute a unique PD-1⁺ subpopulation [7]. Whereas the exact nature and function of the PD-1^-high subpopulation remains unclear, this population appears to emerge with age, wherein PD-1^-high lymphocytes are only scantly noted in lymph nodes from neonates, but are noticed to increase with age [42]. Yang et al. [7] demonstrated that the PD-1^-low subpopulation was the predominantly anergic population, as denoted by functional exhaustion with decreases IL-2 and interferon-γ production as well as dysfunctional intracellular signal transduction. Patients with lymphoma with predominantly PD-1^-low /CD4⁺ populations had increased risk of death, whereas patients in whom PD^-high /CD4⁺ population predominated demonstrated markedly improved survival [7].

Given the crosstalk between microbes and lymphocytes, we sought to assess potential epigenetic changes that may explain the long-term effects seen in severe sepsis as well as major microbiome alterations. Although a prolonged immune dysfunction state is now well recognized in survivors of sepsis, many of the mechanisms driving these long-term effects remain unknown. Epigenetics alterations of lymphocytes are often considered in the context of chronic illnesses such as cancer or asthma [43]. Many of the immune dysfunctions seen in such chronic illnesses, such as alterations in PD-1 pathways [44], are noted to persist after acute septic insults. Our data on microbe-induced DNA methylation alterations is in keeping with recently emerging data. Shosaku et al. [22] demonstrated a differential effect of prolonged infection on enhancer versus promoter regions within the genome, with markedly greater suppression of promoter regions with ongoing infectious exposure. The authors believe that such findings may represent the noted development of a senescent state of repeated infected lymphocytes [22].

There are several limitations to this work. These microbe culture environments created here do not reflect a true septic microbiome, however, they do allow focused assessment of commonly used probiotic microbes. The single immune cell model does not reflect any evidence of direct tissue damage or timing of microbe administration, but a future organoid model may better reflect tissue involvement. This work begins to unmask aspects of how varying microbes may induce unique T-lymphocyte immune phenotypic and cytokine profiles.

Conclusion

Lactobacillus acidophilus induced the potential for short-term resolution (IL-22 and IL-33) whereas Bacteroides fragilis induced a T-cell phenotype consistent with potential long-term immune recovery, including methylation effects. This may offer a new direction in treating sepsis-induced immune dysfunction. Focused and tailored microbiota therapies may ultimately obviate the need for prolonged antimicrobial therapy in immune-paralyzed patients with sepsis.

Funding Information

This work was supported by the National Institutes of Health (R35 GM 118097 [AA] and K08-GM110495 [DSH]).

Author Disclosure Statement

No competing financial interests exist.