Aged Mice Are Unable To Mount an Effective Myeloid Response to Sepsis Free

Sepsis remains a significant problem throughout the world. Infections remain one of the top causes of morbidity and mortality in the elderly (1), and sepsis has been labeled a disease of the aged (2), as 60% of septic patients are older than 65 y (2, 3). Severe sepsis and septic shock have estimated in-hospital mortalities of 29–40% and >50%, respectively (4–6). Of these patients, >80% of the deaths are in the elderly, and age is an independent predictor of mortality in sepsis (2, 7). Even with improvements in patient outcomes due to efforts to standardize initial patient care (8), the total number of deaths due to sepsis is growing because of its increasing incidence (9). In addition, as the elderly population steadily increases, so has the average age of the septic patient (2). Thus, sepsis has become particularly relevant in the aged as compared with other pathologies. For example, in the general surgery population, the incidence of sepsis is greater than the incidence of pulmonary embolism and myocardial infarction combined (8). Ten years ago, it was estimated that septic patients in the United States alone have an annual cost of $17 billion (7), and, to date, immune modulation therapy and pharmacotherapeutic agents have proven disappointing in regard to modifying outcome (10, 11).

Although much research has examined the immune system of the aged, it remains unclear why age is associated with worse outcomes in infection and sepsis. Murine research has demonstrated that aged mice are more susceptible to the same insult of polymicrobial sepsis and that older rodents do not respond as well to antibiotic therapy (12). Several explanations have been identified that may explain these results, including “inflamm-aging” (13), the low-grade chronic proinflammation present in the elderly, as well as “immunosenescence,” the inability of aged immune system to mount as an effective response to an infectious pathogen as the young (14). However, the role of inflammation, and whether the aged response to sepsis is proinflammatory or immunosuppressive, has not been well delineated (1, 2, 15–18). In addition, whereas aged defects in adaptive immunity have been well studied, the impact of aging on innate immunity has been underinvestigated (19).

Sepsis is associated with the rapid release of mature and immature myeloid cell populations from the bone marrow (BM) in response to endogenous and exogenous danger signals (20, 21). We have demonstrated that this evacuation of BM cells creates niches in the BM that stimulate emergency myelopoiesis, an endogenous effort to restore adequate numbers of myeloid populations to inflammatory stress (22). Myelopoiesis is clearly driven at the expense of lymphopoiesis and erythropoiesis (20, 22). The factors driving this process are not completely known, although we have demonstrated that emergency myelopoiesis, in response to polymicrobial or Gram-positive sepsis, is not dependent on either TLR signaling, type I IFNs, or Toll/IL-1R domain-containing adapter inducing IFN-β/MyD88 pathways (22, 23). Regardless, the process results in expansion of both long-term (LT) and short-term (ST) hematopoietic stem cells (HSCs), as well as common myeloid progenitors (22).

We hypothesize that the increased mortality to severe sepsis in the aged can be explained, at least in part, by differences in the early myeloid response of innate immunity. In this work, we tested the specific hypothesis whether the increased mortality in the aged was secondary to an exaggerated inflammatory response or to defects in protective innate immunity.

All experiments were approved by the Institutional Animal Care and Use Committee at the University of Florida. Specific pathogen-free male C57BL/6 (B6) mice were purchased from The Jackson Laboratory (Bar Harbor, ME) at 6–7 wk or from the National Institute of Aging at 20–24 mo of age, and allowed to acclimatize for 1 wk before being used for experimental procedures. Mice were maintained on standard rodent food and water ad libitum.

For induction of polymicrobial sepsis, cecal ligation and puncture (CLP) was performed under isoflurane anesthesia, as previously described (24, 25). Briefly, the cecum was exposed after a laparotomy, ligated with 2-0 silk suture, and punctured through and through with a 25-gauge needle. The cecum was returned in the abdomen, and the incision was closed using surgical clips. After the procedure, the mice were administered 0.05–0.20 mg/kg buprenorphine in 1 ml 0.9% saline, returned to their respective cages, and closely monitored for any signs of distress. The Institutional Animal Care and Use Committee requires euthanasia for moribund mice that are then considered as nonsurvivors.

Peritoneal bacterial counts were determined by culturing 100 μl serially diluted peritoneal washings on sheep’s blood agar plates (Thermo Fisher Scientific) at 37°C in 5% CO₂. Plates were counted after 24 h of culture (23).

Spleens, whole blood, and BM were harvested, and single-cell suspensions were created by passing the cells through 70-μm pore-sized cell strainers (BD Falcon, Durham, NC). Erythrocytes were then lysed using ammonium chloride lysis buffer and washed twice using PBS without calcium, phenol red, or magnesium. Cells were stained with the following Abs for flow cytometric studies: PE Cy7 anti-CD11b, allophycocyanin anti–Gr-1, and Pacific Blue anti-Ly6G (BD Pharmingen, Billerica, MA). Additional Abs used were antilineage mixture (BD Biosciences, San Jose, CA), anti-ckit, anti–Sca-1, anti-CD135, and anti-CD150 (eBioscience, San Diego, CA). Sytox Blue (Invitrogen, Carlsbad, CA) was used for cell viability analysis, and samples were acquired and analyzed using a LSRII flow cytometer (BD Biosciences) and FACSDiva (BD Biosciences) (20, 26).

Blood was harvested by intracardiac puncture at 2 h, 1 d, or 3 d after CLP. Plasma was collected and stored at −80°C until the time of analysis. Plasma cytokine concentrations were determined using a commercially available multiplexed Luminex kit (MILLIPLEX MAP, mouse cytokine/chemokine panel; Millipore, Bellirica, MA). Cytokines evaluated included IL-1β, IL-6, IL-12 (p70), IFN-inducible protein-10, keratinocyte-derived chemokine (KC), MCP-1, MIP-1α, and TNF-α. All assays were performed according to the manufacturer’s protocols. Cytokine concentrations were determined using BeadView software (Millipore).

Spleen and BM cells were prepared using a Histopaque density gradient (1.119 specific gravity) and washed using PBS without calcium, phenol red, or magnesium. Cells were then labeled for surface markers, as described, and washed twice with PBS. Reactive oxygen species (ROS) production was determined using dihydrorhodamine 123 (Invitrogen, Carlsbad, CA). Subsequently, cells were stimulated with PMA at 37°C and evaluated by flow cytometry analysis every 10 min for a 30-min period. A minimum of 1 × 10⁴ live, nondebris cells was collected for analysis.

Spleen and BM cells (10⁵) were incubated with 10⁶ yellow-green polystyrene microspheres (FluoSpheres: Invitrogen) in 37°C water bath for 10 min; washed with PBS containing 0.1% BSA; stained with anti-Ly6G, anti-CD11b, and anti-Ly6C; and analyzed by flow cytometry.

Peritoneal cells were collected 1 d after CLP, prepared using a Ficoll gradient, and resuspended in medium containing 0.5% FBS at a concentration of 1 × 10⁷ cells/ml. Medium containing MCP-1, KC (30 ng/ml; BioLegend, San Diego, CA), or medium alone as a control was added to the lower chambers of a 24-well Costar Transwell plate (Corning, Corning, NY). The cell suspension (100 μl) was added to the upper chamber, which was separated from the lower chamber by a polycarbonate membrane (5.0-μm pores). After incubation for 2 h at 37°C, cells in the lower chamber were collected; stained with anti-Ly6G, anti-CD11b, and anti-Ly6C; and analyzed by flow cytometry. Results are presented as a migration index calculated by dividing the number of cells that migrated toward MCP-1 or KC by the number of cells that migrated to medium alone (27).

BM cells from young and aged mice were aseptically collected 1 d after CLP. Single-cell suspensions were created by passing the cells through 70-μm pore-sized cell strainers (BD Falcon, Durham, NC). Erythrocytes were lysed using ammonium chloride lysis buffer and washed with PBS. Cells were stained with antibiotin lineage mixture (BD Biosciences), anti-ckit, and anti–Sca-1 (eBioscience). Lineage⁻ Sca-1⁺ ckit⁺ cells (LSKs) were sorted using FACSAria (BD Biosciences). Five hundred LSKs were cultured in methylcellulose media (R&D Systems, Minneapolis, MN) supplemented with either GM-CSF, G-CSF, M-CSF, or IL-7 (R&D Systems). Colonies were counted after 10- to 14-d incubation at 37°C (22).

Blood was collected by intracardiac puncture at 2 h, 1 d, or 3 d after CLP using 1-ml syringes containing 100 μl 169 mM EDTA. RBCs were lysed using Buffer EL (Qiagen, Valencia, CA). The supernatant was decanted after centrifugation, and the cell pellet was homogenized in RLT buffer (Qiagen) supplemented with 2-ME and passed through Qiashredder (Qiagen). Total RNA was isolated using RNeasy kit (Qiagen, Valencia, CA), and the quality and quantity were assessed using Agilent Bioanalyzer 2000. Nucleic acids were labeled using the 3′ IVT Express Kit, and 15 μg labeled cRNA was hybridized to mouse genome 430 2.0 arrays (Affymetrix, Santa Clara, CA). Arrays were hybridized for 16 h at 45°C. Following hybridization, arrays were stained and washed using a FS450 Affymetrix fluidics station and Affymetrix FlexFS 450-0004 protocol. Arrays were then scanned in an Affymetrix GeneChip scanner 7G Plus. Genome-wide expression was performed on total blood leukocytes.

Continuous nongenomic variables were first tested for normality and equality of variances. Differences among groups in flow cytometric analyses were evaluated using Student t test. Additional statistics were performed using one-way ANOVA and two-way ANOVA. Post hoc comparisons were performed using Student Neuman-Keuls test. Significance was determined at the 95% confidence interval using a two-sided test. Blood leukocyte genome-wide expression patterns were compared between healthy and young/aged CLP mice using a false discovery adjusted F test (p < 0.001) with BRB Tools. We also calculated the distance from reference (DFR) based on the studies of Warren et al. (28). The DFR calculation derives a single value for the overall differences in gene expression calculated as the natural log of the sum of the differences in gene expression (between healthy and septic animals) for each probe set divided by the pooled variance for that individual probe set.

We initially examined whether elderly mice were more susceptible to polymicrobial sepsis, as previously reported (12). Indeed, aged mice have significantly increased mortality to polymicrobial sepsis (CLP) as compared with young mice when both sets of mice received the same insult (Fig. 1A). However, it would be expected that the majority of the deaths would occur early if this was due to an exaggerated systemic inflammatory response syndrome and shock or acute organ failure. Rather, the kinetics of the increased mortality was gradual over the 7-d observation period.

We also investigated whether an inability to control infection was associated with the observed adverse outcomes. One day after polymicrobial sepsis, we lavaged the peritoneum from young and aged mice and determined the bacterial CFUs from each mouse. As illustrated in Fig. 1B, aged mice had a 3 log-fold increase in the number of bacterial CFUs as compared with young mice (p < 0.004).

Although there is no doubt that inflamm-aging exists (13), the role of increased inflammatory cytokine expression after sepsis in the elderly remains unclear. We found no significant difference in the plasma concentration of inflammatory cytokines (KC, MIP-1α, MCP-1, IFN-inducible protein-10, IL-6, and TNF-α) at either 1 or 3 d after CLP (Fig. 2) between aged and young mice. Aged mice did produce significantly less IL-10, an anti-inflammatory cytokine, 1 d after CLP (Fig. 2).

We determined the relative and absolute numbers of neutrophils (PMNs; CD11b⁺Ly6G⁺), monocytes/macrophages (MOs; CD11b⁺Ly6G⁻), dendritic cells (CD11c⁺), and immature myeloid-derived suppressor cells (MDSCs; CD11b⁺GR-1⁺) in the spleen, BM, and circulation at 1 and 3 d after the induction of polymicrobial sepsis. Despite a 3-log increase in the number of peritoneal bacteria in aged mice, there was no significant difference in the dendritic cell populations (data not shown), PMNs, and MOs; MDSCs were mostly found to be increased or trending toward increased levels in both the spleen and BM of aged mice after CLP at specific time points (Fig. 3). The circulating number of mature and immature myeloid cells, including MDSCs, was unchanged compared with the young mice, at both 1 and 3 d after CLP, indicating that the increased mortality of aged mice after sepsis and the increased number of bacteria could not be easily explained by any deficit in the numbers of myeloid cell populations.

One day after CLP, splenic, BM, and peritoneal PMNs and MOs were tested for their ability to produce ROS and to phagocytose, and for their capacity for chemotaxis in response to the chemokines, KC and MCP-1, respectively. Although not universal, specific compartments of myeloid cells displayed cellular dysfunction. Both splenic monocytes and BM PMNs from aged mice had significantly less ROS production as compared with young septic mice (Fig. 4A). Splenic and BM monocytes, as well as BM PMNs, also had decreased phagocytic function (Fig. 4B). However, myeloid cells from the spleen and BM of aged mice did not exhibit significant differences in chemotaxis as compared with young mice (data not shown). In contrast, peritoneal monocytes and PMNs from aged mice had significantly or trended toward decreased chemotaxis ability (Fig. 4C). These cells had no difference in their ROS production or phagocytic ability (data not shown).

We examined the different subsets of HSCs and progenitor cells in the BM of naive and post-CLP mice. BM LSK, LT-HSC (CD150⁺CD135⁻LSK), and ST-HSC (CD150⁻CD135⁺LSK) were analyzed phenotypically. Interestingly, prior to polymicrobial sepsis, as well as 1 and 3 d afterward, the relative numbers of ST-HSCs were reduced in the elderly (Fig. 5A). LT-HSCs can reconstitute hematopoiesis long term at very low numbers, but more recent data from the transplantation literature suggest that ST-HSCs, although more limited in their self-renewing potential, are more vital for appropriate, rapid myelopoiesis after BM loss (28). In addition, the elderly murine BM response to severe sepsis differs to young mice, as LSKs from aged mice appear limited in their capacity to proliferate along lymphoid and myeloid pathways in response to certain growth factors (Fig. 5B).

Initial analysis determined that the genome-wide expression pattern of circulating leukocytes from healthy young and aged mice could not be readily differentiated (data not shown). Therefore, expression patterns from healthy, young, and aged mice were used as a single control group for analyzing the response to sepsis. In addition, we determined the relative and absolute differentials of the circulating leukocytes of naive and septic mice 1 d after CLP in both young and aged mice. The makeup of the circulating WBCs was not significantly different, and thus could not explain the differences found in the genomic response (Fig. 6A).

The expression of 28,464 probe sets representing 15,804 genes was significant in differentiating the effect of age and time after sepsis (at p < 0.001). Looking specifically at 2 h, 1 d, and 3 d after sepsis, 6,203 probe sets (4,324 genes), 7,975 probe sets (5,374 genes), and 15,254 probes sets (9,911 genes) were differentially expressed, respectively. A DFR metric was calculated for the treatment groups at each time point as a natural log estimate of the global aberration in gene expression (29). After sepsis, young mice have a significantly different leukocyte genomic response as compared with aged mice (Fig. 6B). As illustrated in Fig. 6B, 1 d after sepsis, the DFR of young mice is significantly greater than that of aged mice, indicative of an attenuated early genomic response in aged mice. We then used Gene Ontology and Biocarta to identify specific pathways of genes differentially expressed that varied between the young and aged animals. This included, but was not limited to, pathways representing neutrophil-mediated immunity, neutrophil chemotaxis, and CC chemokine binding, all of which were more significantly upregulated in the young as compared with the aged mice at 1 d (Fig. 7). These data suggest that the failure of myeloid cells from aged mice to exhibit normal neutrophil functions can be explained at the level of the murine transcriptome.

We also analyzed the fold changes in individual genes found to have significantly altered expression from baseline. Comparison of the probe sets representing IL genes illustrated that there was a failure to increase upregulation of proinflammatory cytokines at 2 h and 1 d after CLP in mouse leukocytes from aged, as compared with young, mice (Table I). However, there was also a lack of appropriate upregulation of other genes important to innate immunity and myeloid cell function in the elderly mice as compared with young mice at 2 h and 1 d after polymicrobial sepsis (Table II).

At 3 d after CLP, surviving young mice have gene expression patterns that are more similar to healthy, control mice, indicating a return to homeostasis. In contrast, the overall transcriptome of aged mice continues to be aberrant with time. In addition, some cytokines (such as IL-6, Table I) and innate immunity genes (data not shown) continue to exhibit increased expression in the elderly mice at this late time point. Rather than being a delayed response, this is more likely a reflection of the aged mouse’s inefficient/suboptimal ability to clear the bacteria appropriately and resolve the infection (Fig. 3). Thus, the aged mice appear to fail to clear their microbial challenge and have an ongoing infectious insult, leading to continued stimulation and upregulation of specific genes, whereas young mice, who apparently have successfully contained the septic insult, exhibit gene expression patterns returning to baseline values. It should also be noted that even though some of these genes related to innate immunity have increased expression in the aged mice 3 d after sepsis as compared with young mice, most of the genes in the aged mice never reach expression levels initially demonstrated by younger mice at 2 h and 1 d after polymicrobial sepsis (Table II). Finally, the expression of MHC genes in aged mice after CLP remains much more downregulated at 3 d as compared with juvenile mice (Table III).

Our data indicate that elderly mice are more susceptible to the same model of polymicrobial sepsis as their juvenile counterparts. This increased mortality is associated with a failure of protective immunity, rather than exaggerated inflammation. In aged mice, there appears to be a failure of myelopoiesis to generate myeloid cells that can perform appropriate protective immune function. This is reflected in the leukocyte transcriptome, in which aged mice were not able to initially respond to sepsis in the same manner as young mice. In addition, because of the failure to control infection in aged mice, expression of specific genes related to innate immunity cannot return to baseline expression values as seen in young rodents.

Our understanding of the septic response in the elderly population is still quite limited. It has been previously demonstrated that elderly rodents have specific leukocyte deficits as well as increased mortality to polymicrobial sepsis (14, 30–33). Some sentinel work in aged mice indicated increased serum cytokines in elderly mice after CLP may be causative for this effect (2, 12). Although there were some differences in our model of CLP from this previous research, our results would indicate that an overwhelming inflammatory response does not appear to be the primary cause of the increased susceptibility of the aged. Conversely, our data are more consistent with the findings of others who have also demonstrated an inadequate cytokine upregulation in the acute and subacute time periods after pneumonia infection in elderly (1, 16). Interestingly, one laboratory implanted pumps that chronically released low levels of TNF-α into young mice to try to recreate inflamm-aging (16). Postinfection, these mice acted more like aged mice than the young controls and did not have an increase in proinflammatory cytokine secretion (16). Also, TLR1, 2, and 4 receptors in the lung were decreased (16), similar to the reduced expression demonstrated by circulating WBCs in aged mice after CLP from our work (Table II).

Previous work from our laboratory would indicate a lack of an appropriate myeloid response can explain increased mortality in young mice (23). However, aged hematopoietic stem cells have a predilection for myelopoiesis (34, 35). Data from our laboratory and from others would indicate that aged mice have no difficulty engendering myeloid cells in response to severe injury or infection. Yet, myeloid cells from aged animals have clear transcriptomic and phenotypic differences from young animals, and these older mice remain remarkably more susceptible to mortality after an infectious challenge (2, 12, 16, 17, 34–39).

Detailed analysis of the BM response of young mice to polymicrobial sepsis in our laboratory clearly illustrated that in response to sepsis there was a marked expansion in both the relative percentage and absolute number of LSK cells, including both LT- and ST-HSCs (22). Our current work demonstrates that the composition and function of young and aged mouse BM are significantly different in regard to the numbers of LT- and ST-HSCs (Fig. 6). Our data would indicate that aged mice have significantly fewer ST-HSCs at baseline, and that this phenotype continues after sepsis (Fig. 6). In addition, data have demonstrated that nonseptic elderly mice and human HSCs have a reduced repopulating capacity as compared with their younger counterparts (40, 41). Again, our data support this, as LSK cells from septic aged mice are less able to form myeloid colonies in response to certain growth factors (Fig. 6B). This may be due in part to changes in the microenvironment/niche of both immature and mature cells (40), although our data indicate that their direct responsiveness to growth factors is also impaired. Interestingly, experiments that have placed young HSCs into bone from older mice have demonstrated that old stroma is less able to support HSCs (41). In addition, injection of murine HSCs from young mice into septic juvenile mice improved survival, which was associated with an improved response to proinflammatory mediators, enhanced phagocytosis, and a better clearance of bacterial peritonitis (42).

Our microarray analysis indicates that the early elderly response to sepsis is attenuated compared with their younger counterparts, consistent with the phenotype and immune dysfunction of these animals. The genomic response of aged rodents in its entirety is less than that of young mice in the first 24 h (Fig. 7), and aged leukocytes fail to upregulate most genes involved in appropriate myeloid cell activities in response to an infection (Tables I, II). In addition to this inadequate initial response, the elderly are not as capable of a homeostatic return to baseline genomic expression for specific genes related to innate immunity, as seen in the juveniles. Although this may be in part due to the inadequacy of the cells derived from their myelopoietic response, it does reflect human data from the Glue Grant “Inflammation and Host Response to Injury Large Scale Collaborative Research Program,” in which patients with complicated outcomes had a much slower return to the expression profiles of uninjured individuals (43).

One of the main weaknesses of this study is the murine CLP model itself. It should be noted that variations in the technique and species used can vary the subsequent mouse response (24). Thus, it is possible that we missed the actual peak cytokine secretion after CLP in our experiments. However, previous work from our laboratory using the same mouse species and CLP technique demonstrated that cytokine levels peak 1 d after CLP and decline thereafter (20, 24); this played a major role in our selection of the time points for sampling. Another weakness with the murine CLP model is the use of animals to recapitulate the human condition. Recently, it was published that the murine genomic response to LPS, burn, or trauma poorly correlates with similar human inflammation (44). The main concern that persists about the use of animal models of sepsis is that it is not representative of the human condition in terms of the heterogeneity of the type of insult, duration, and supportive therapy such as use of antibiotics and adequate hydration, and thus does not reproduce the whole spectrum of human sepsis. Despite its limitations, the use of animal models such as CLP can still provide useful insight to the understanding of a complex process and remain an important tool in developing and/or improving therapeutic options for sepsis. In fact, work from our laboratory that repeated the experiments of Seok et al. (44) illustrated that, regarding innate immunity, the genomic response of human and murine leukocytes can be quite similar (45). Of all the animal models of sepsis, CLP remains the most frequently used because it more closely resembles human sepsis progression. In addition, in our laboratory, the CLP model closely approximates the mortality outcomes of septic human patients based on their age.

In summary, despite a predilection to myelopoiesis, the initial aged myeloid response can be considered suboptimal as compared with that of younger cohorts, leading to a pronounced susceptibility to polymicrobial sepsis. This does not appear to be related to an overwhelming inflammatory response, but rather an inability of myeloid populations to respond appropriately to bacterial infection, as demonstrated by reduced colony formation and reduced function in some mature myeloid cell populations. Microarray analysis reflects this inadequate response as well, as aged leukocytes demonstrate a failure to mount an appropriate upregulation of gene expression important to innate immunity. This is followed by continued systemic inflammation and immunosuppression after the acute phase of sepsis, as opposed to the movement toward baseline genomic expression displayed by young mice.

The authors have no financial conflicts of interest.