Clinical deterioration of GAS-induced sepsis was influenced by the route of infection

To determine whether different infection routes elicit distinct immune responses, C57BL/6J mice were infected with GAS via subcutaneous, intravenous, or intranasal routes. Corresponding control animals received PBS. Peripheral blood samples from all groups were taken at 24 and 48 h to assess bacterial load. At 42 h (6 h prior to experimental end point), mice were intraperitoneally administered a weight-adapted dose of BFA. Animals were then euthanized, and their organs were harvested for subsequent analyses (Fig. 1A).

In order to investigate whether the different infection routes resulted in varying clinical progressions, we scored sepsis severity via clinical manifestations, such as weight loss and abnormal behaviour or appearance. Indeed, when comparing sepsis scores between infection groups and over time, significant differences were observed (p < 0.001, two-way ANOVA, Fig. 1B). In detail, subcutaneously infected animals exhibited immediate weight loss and clinical signs of infection, plateauing at 24 h. In contrast, intravenously infected mice showed a delayed progression without any clinical manifestations for the first 24 h, with a steep increase in disease activity thereafter. At the final observation point at 48 h, disease activity was comparable between the subcutaneous and intravenous routes (p = 0.179, two-sided t-test). Surprisingly, intranasally infected mice did not display significant signs of infection at any time.

Fig. 1

Clinical disease burden and bacterial dissemination in Group A Streptococcus (GAS)-infected mice. (A) Experimental design depicting the infection protocol of C57BL/6J mice with GAS via subcutaneous (SC, n = 8), intravenous (IV, n = 9), and intranasal (IN, n = 8) routes with respective control groups receiving PBS. (B) Disease activity scores over 48 h post infection for SC, IV, and IN routes. Dots and error bars display mean sepsis scores and standard error of the mean, respectively. (C) Bacterial dissemination in blood and organs quantified by colony-forming units (CFU) per ml or per organ. Dots indicate individual counts of CFU per mouse. Bars represent median values. Groups were compared using the Kruskal-Wallis test with Dunn’s multiple comparisons test (**p < 0.01, ***p < 0.001)

Bacterial dissemination was assessed in the peripheral blood at 24 and 48 h and in homogenized organs and BAL fluids at endpoint. Despite the differences in clinical progression, bacterial load in the blood was similar across all infection routes at both time points (Fig. 1C). As for peripheral organs, intravenously infected mice exhibited the highest bacterial dissemination, while intranasally infected animals had only minimal bacterial colonization.

In summary, subcutaneous and intravenous infections culminated in comparable sepsis manifestation at the experimental endpoint, even though differences occurred in kinetics and bacterial disseminations. In contrast, intranasal infection did not elicit sufficient disease progression to effectively model sepsis. These findings indicate that while bacteraemia was present across all infection routes, the primary site of GAS infection impacted on clinical progression.

Increased granulocyte-to-lymphocyte ratio in the blood of septic animals correlated with high bacterial burden in the liver

After establishing that different infection routes caused distinct clinical progressions, we investigated the underlying immune responses by assessing the respective peripheral blood immune cell landscapes. GAS infection in general led to reduced leukocyte counts across all groups, with significant leukopenia observed in subcutaneous and intravenous, but not intranasal infections (Fig. 2A).

To identify cell types affected by leukopenia, we next analyzed immune cell compositions using flow cytometry, applying the gating strategy for blood samples shown in Supplementary Fig. 1A. Subcutaneously and intravenously infected mice showed a relative increase of the neutrophil population at the expense of the B cell fraction (Fig. 2B). Absolute counts confirmed reductions in B cells, T cells, and innate lymphoid cells (ILCs, Fig. 2C), suggesting that lymphopenia was the main cause of leukocyte composition alterations. Although there was also a decrease in macrophage counts, the alterations in this low-frequency population affected the peripheral immune landscape only marginally (Supplementary Fig. 1B). Conversely, neutrophil counts remained stable during this phase of infection, as well as natural killer (NK) cells, natural killer T cells (NKTs), dendritic cells (DCs) and CD11b+B220+ B1 cells. Correlation analyses between leukocyte population counts and sepsis scores revealed a negative correlation of B cell, T cell, and ILC numbers with sepsis severity (Fig. 2D), underscoring the strong interrelationship between lymphocyte depletion and disease progression. Finally, we asked whether the granulocyte-to-lymphocyte ratio could be linked to the bacterial load in the liver as an indicator of organ stress and indeed, an elevated ratio (> 1) correlated significantly with a higher bacterial burden (Fig. 2E). These increased ratios were almost exclusively observed after intravenous infection, hinting at a link between reduced circulating lymphocytes and peripheral bacterial dissemination.

In summary, the early phase of clinically apparent sepsis was characterized by peripheral blood lymphoid cell depletion, resulting in an elevated granulocyte-to-lymphocyte ratio, which in turn showed a significant association with both disease severity and bacterial burden in the liver.

Fig. 2

Immune cell alterations in the blood after GAS infection. (A) Absolute leukocyte counts of infected animals (INF) as a fraction of counts in control animals (CON). Interleaved bars indicate mean values. (B) Stacked bar plots depict immune cell compositions. Columns represent individual samples. (C) Absolute counts of B cell, T cell, ILC, and neutrophil populations are shown. Data of control and infected animals are presented for each infection route. Box plots show median values and interquartile range. Mann-Whitney test was used for comparison between control and infected animals (#p < 0.05). Kruskal-Wallis test with Dunn’s multiple comparisons test was applied for comparisons between infection routes (*p < 0.05, ****p < 0.0001). (D) Correlation analysis of leukocyte population counts with sepsis score is presented as a linear regression graph with a 0.95 confidence interval. Spearman correlation coefficient (r) is indicated. Dots represent individual animals. (E) Granulocyte-to-lymphocyte ratio of the peripheral blood was correlated with bacterial load in the liver using Spearman correlation

Intravenous infection resulted in increased granulocyte infiltration and B lymphopenia in the liver

Next, we explored the effects of GAS infection on the immune cell landscapes in the liver, spleen and lung (i.e. bronchoalveolar lavage, for intranasal infection only). For spleen and liver, we implemented the gating strategy outlined in Supplementary Fig. 2A to assess both absolute immune cell numbers as well as their relative distribution. Notably, intravenous infection resulted in a significant increase in hepatic leukocyte infiltration, while the spleen remained unaffected (Fig. 3A and B).

In intranasally infected animals, using the gating strategy shown in Supplementary Fig. 3A, bronchoalveolar lavage revealed significant leukocyte migration into the lower respiratory tract despite the lack of clinical symptoms, confirming an adequate immune response following mucosal infection. In detail, significant increases in relative cell numbers were noted for B cells, T cells, ILCs, and NK cells, while the fraction of neutrophils was reduced (Supplementary Fig. 3B).

In the livers of subcutaneously and intravenously infected mice, we observed immune cell alterations similar to those in the blood. Especially in intravenous infection, there was a relative decrease in B cells and an increase in neutrophils (Fig. 3C). Again, the percentages of B cells correlated negatively with the respective sepsis score (Supplementary Fig. 3B and C). In contrast, the fraction of DCs increased after infection and positively correlated with disease activity. In the spleens however, the immune cell compositions remained largely stable post infection (Fig. 3D), with significant alterations restricted to low-frequency populations. In detail, ILCs were significantly depleted, while CD11b+B220+ B1 cell fractions increased in subcutaneous and intravenous infection (Supplementary Fig. 3E). Both alterations correlated strongly with the sepsis score (Supplementary Fig. 3F).

These findings suggest that during the early phase of sepsis, splenic immune cell proportions remain largely unaffected by both subcutaneous and intravenous GAS infections. In contrast, the liver exhibits distinct immune alterations, with intravenous infection leading to pronounced hepatic leukocyte infiltration.

Fig. 3

Immune cell alterations in peripheral organs after GAS infection. Absolute leukocyte counts in liver (A) and spleen (B). Data of control and infected animals are presented for each infection route. Dot plots represent individual samples and lines depict mean values. Mann-Whitney test was used for comparisons between control and infected animals (#p < 0.05). Immune cell compositions of the liver (C) and spleen (D) are depicted in stacked bar plots for various cell types. Columns represent individual samples

Early sepsis induced depletion of common myeloid and lymphoid progenitor cells in the bone marrow

In order to link the peripheral immune response to the stem and progenitor cell landscape, we analyzed the bone marrow compartment of infected mice using the gating strategy shown in Supplementary Fig. 4A. To that end, we examined the main lineage leukocyte populations in the bone marrow in absolute (Fig. 4A) and relative cell counts (Supplementary Fig. 4B). Interestingly, B cell reduction was observed across all infection routes, yet reached significance only following subcutaneous and intranasal infection. Similarly, neutrophil depletion occurred and was paralleled by a negative correlation of sepsis activity with Gr-1 expression on neutrophils (Supplementary Fig. 4C). Since Gr-1 serves as a marker of neutrophil maturity [8], we further analyzed the population of immature Gr-1− neutrophils, applying the gating strategy shown in Supplementary Fig. 5A. Compared to mature polymorphonuclear Gr-1+ neutrophils, immature neutrophils were significantly enriched in the blood of septic animals, while their increase in the spleen and liver was less pronounced (Supplementary Fig. 5B). Cytokine analysis further indicated reduced functionality, as immature neutrophils exhibited a universal decrease in IL-6 expression and partial decline in TNFα and IFNγ levels in the bloodstream and liver (Supplementary Fig. 5C).

Next, we investigated alterations of the stem and progenitor cells. Numbers of long-term hematopoietic stem cells (LT-HSCs) were unaffected by infection, but short-term hematopoietic stem cells (ST-HSCs), mesenchymal stem/stromal cells (MSCs), and multipotent progenitor cells (MPPs) increased significantly following subcutaneous and intravenous infections (Fig. 4B). These effects were more pronounced in intravenously infected mice compared to the other infection routes. Conversely, common myeloid progenitors (CMPs) and common lymphoid progenitors (CLPs) significantly decreased after both subcutaneous and intravenous infection.

To link alterations in the bone marrow compartment to shifts in the peripheral immune cell landscape, we performed correlation analyses. Supplementary Fig. 6 illustrates that the sepsis-induced accumulation of bone marrow MPPs and ST-HSCs significantly correlated with a decrease in CMPs, CLPs, and bone marrow neutrophils and was also associated with peripheral lymphocyte depletion. Furthermore, an increase in peripheral immature neutrophils was significantly associated with this disruption in hematopoiesis.

Our data suggest that ongoing emergency hematopoiesis was characterized by an exhausted reservoir of progenitor cells from both myeloid and lymphoid lineages, which coincided with a failure to replenish depleted peripheral lymphocytes while leading to the release of immature neutrophils with reduced effector functions.

Fig. 4

Changes in leukocytes (A) and their progenitor cells (B) in the bone marrow after GAS infection. Data of control and infected animals are presented for each infection route. Box plots show median values and interquartile range. Mann-Whitney test was used for comparison between control and infected animals (#p < 0.05). Kruskal-Wallis test with Dunn’s multiple comparisons test was applied for comparisons between infection routes (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

B cells were key drivers of cytokine production and IL-6 overexpression in peripheral blood indicated hepatic stress in sepsis

Next, we examined the cellular origins of cytokines during early sepsis. To this end, we generated broad cytokine profiles at a single cell level and concentrated on a selection of pro- and anti-inflammatory cytokines as well as chemokines. Analyzing general cytokine expressions by all peripheral blood leukocytes, we found only a non-specific upregulation of CCL3 in both intranasally infected and their respective control mice (Fig. 5A and Supplementary Fig. 7A). Due to this overall inconspicuous cytokine profile, we thus focused on subcutaneous and intravenous infections. Here, we observed elevated expressions of IL-6, IFNγ, CCL2, TNFα, and IL-10 in blood leukocytes. Notably, IFNγ-expressing cells were more prevalent in intravenous infection.

In order to filter for clinically relevant cytokine dynamics, we performed correlation analyses between significant changes in cell type-specific cytokine expressions and the infection route-specific sepsis scores. This way, we identified B cells, T cells, ILCs, and MHCII+ macrophages as key contributors to cytokine production (Supplementary Table 1). No relevant changes were observed for IL-10, IL-17A, and CCL3 in the cell type-specific analyses. However, IL-6+ cells were significantly enriched in infected animals across nearly all cell types and regardless of infection route, and their frequency correlated positively with the sepsis score (Fig. 5B). Of note, the highest proportion of IL-6 expressing cells was found in the B cell population. IFNγ expression in subcutaneous infection was significantly increased in B cells only, while intravenously infected animals demonstrated elevated IFNγ levels in B cells, T cells, ILCs, and macrophages (Fig. 5C). Consequently, intravenous infection was characterized by a strong correlation between IFNγ expression and the sepsis score for the mentioned cell types. CCL2 levels were enriched in B cells of both subcutaneously and intravenously infected animals, while upregulation of CCL2 in ILCs and macrophages was significant only following subcutaneous infection (Supplementary Fig. 7B). Finally, TNFα overexpression was restricted to B cells and ILCs following subcutaneous infection, while intravenously infected animals remained inconspicuous (Supplementary Fig. 7C).

Next, we investigated how cytokine analysis could aid in predicting organ complications in sepsis. Given the strong correlation between disease activity and IL-6 expression, we further explored the relationship between bulk IL-6 levels and organ stress. Therefore, IL-6 expression was correlated with hepatic bacterial load. We found that intravenously infected animals, which were characterized by increased bacterial accumulation in the liver, exhibited the highest IL-6 levels (Fig. 5D).

Fig. 5

Cytokine expressions in the blood after GAS infection. (A) Heat map of color-coded z-scored mean fluorescence intensity of cytokines in live leukocytes. Percentage of IL-6+ (B) and IFNγ+ (C) cells among immune cell populations are demonstrated for subcutaneous and intravenous infection. Data of control and infected animals are presented for each infection route. Superimposed bar plots show median values with interquartile range. Mann-Whitney test was used for comparison between control and infected animals (*p < 0.05, **p < 0.01, ***p < 0.001). Dots represent Spearman correlation analysis of cytokine expression with sepsis score, with correlation coefficient (r) indicated by color and p > 0.05 marked x. (D) Bulk IL-6 expression was correlated with bacterial load in the liver using Spearman correlation

In summary, especially peripheral blood B cells produced multiple pro-inflammatory cytokines following both subcutaneous and intravenous infection. IL-6 levels were consistently elevated across various blood lymphocyte populations. Notably, following intravenous infection, excessive overexpression of IL-6 was strongly associated with hepatic stress.

Intravenous infection led to liver inflammation with an unbalanced expression of IFNγ and IL-10

Cytokine expressions among leukocytes were also investigated in peripheral organs. In intranasally infected animals, leukocyte populations from BAL samples during intranasal infections showed elevated levels of CCL3 among T cells and macrophages (Supplementary Fig. 8A). Moreover, B cells and CD4+ T cells demonstrated a slight but significant overexpression of IFNγ. Given that there was no discernible cytokine reaction in the lung beyond this (Fig. 6A), these findings suggest a localized mucosal immune reaction in the lower respiratory tract following intranasal infection. In the livers of subcutaneously and intravenously infected animals, bulk MFI levels of CCL3, IFNγ, IL-10, IL-6, CCL2, and TNFα were significantly increased (Fig. 6A, Supplementary Fig. 8B). Interestingly, the tendency towards IL-17A overexpression among liver leukocytes was restricted to intravenously infected mice (Fig. 6A). In contrast, cytokine levels in spleen cells remained unchanged, except for an increase in IFNγ following intravenous infection (Supplementary Fig. 8C). We thus focused further analyses on the liver.

Correlation analyses with the sepsis score again identified lymphocytes and macrophages as key contributors to cytokine overproduction (Supplementary Table 2). When investigating differences between the infection routes, we found an enrichment of CCL3+ T cells and MHCII+ macrophages following intravenous infection (Fig. 6B). Due to their strong correlation with the sepsis score, this indicates increased chemotactic functions during intravenous infection. Intravenous infection also resulted in significantly elevated levels of IFNγ-expressing cells across all lymphocyte populations (Fig. 6C), which again correlated strongly with the sepsis score. Although there was a minor increase in median IFNγ expression by T cell populations following subcutaneous infections, this did not exceed 2%, rendering it less relevant. Strikingly, IL-10 was only marginally expressed by B cells in intravenously infected mice (Fig. 6D). In contrast, subcutaneous infection led to increased IL-10+ B cells, CD4+, CD8+ T cells, and MHCII+ macrophages, all correlating with the sepsis score. This suggests a diminished anti-inflammatory response after intravenous infection.

Collectively, our analysis reveals a strong chemotactic response and a higher pro-inflammatory cytokine burden in the liver following intravenous infection. Increased expression of IFNγ in hepatic lymphocytes was paralleled by insufficient IL-10-mediated anti-inflammatory regulation. This suggests a distinct cytokine imbalance and highlights liver involvement in excessive inflammation.

Fig. 6

Cytokine expressions in the liver after GAS infection. (A) Heat map of color-coded z-scored mean fluorescence intensity of cytokines in hepatic leukocytes. Percentage of CCL3+ (B), IFNγ+ (C), and IL-10+ (D) cells are demonstrated for subcutaneous and intravenous infection. Data of control and infected animals are presented for each infection route. Superimposed bar plots show median values with interquartile range. Mann-Whitney test was used for comparison between control and infected animals (*p < 0.05, **p < 0.01, ***p < 0.001). Dots represent Spearman correlation analysis of cytokine expression with sepsis score, with correlation coefficient (r) indicated by color and p > 0.05 marked x