Elevated Spleen Tyrosine Kinase in Low-Density Neutrophils During Bacterial Sepsis in a Nonhuman Primate Model

Abstract

Background

Sepsis is a leading cause of death worldwide. Identifying novel host-directed therapeutic targets may improve sepsis outcomes.

Methods

Six nonhuman primates were infected with Klebsiella pneumoniae to induce septic shock and provided supportive care for up to 72 hours. Flow cytometry was used to characterize whole-blood neutrophils (WBNs) and low-density neutrophils (LDNs) at time 0 (T0), T6, T24, and T48 hours postinfection, and postmortem examination (ie, necropsy). Dimensional reduction with clustering via FlowSOM and traditional gating strategies were used to compare WBNs to LDNs and delineate spleen tyrosine kinase (SYK) expression across neutrophils subsets. We measured soluble biomarkers of end-organ dysfunction and neutrophil activation, and quantified SYK and myeloperoxidase in tissue.

Results

At T6, we identified populations of active immature WBNs and a population of LDNs not detected at baseline. At T24, neutrophil heterogeneity increased across WBNs and LDNs with differential expression of myeloperoxidase (MPO). Compared to WBNs, LDNs were more activated with increased MPO expression. At T6, SYK expression surged in WBNs and LDNs and SYK⁺ WBNs and LDNs expressed higher levels of MPO and lactoferrin compared to SYK⁻ neutrophils. Circulating levels of SYK⁺ LDNs significantly correlated with serum creatinine levels, indicative of acute kidney injury; prolonged prothrombin time and decreased fibrinogen, indicative of consumptive coagulopathy; and SYK expression in tissues.

Conclusions

Bacterial sepsis leads to heterogenous populations of circulating neutrophils, including LDNs. Elevated SYK expression in WBNs and LDNs correlates with end-organ dysfunction, highlighting SYK as a potential therapeutic target in bacterial sepsis.

Graphical Abstract

Graphical Abstract.

In a nonhuman primate model of bacterial sepsis, low-density neutrophils are elevated with increased heterogeneity over time and express high levels of spleen tyrosine kinase (SYK) that associate with end-organ dysfunction, indicating SYK inhibition as a novel therapeutic approach.

Sepsis leads to tremendous worldwide morbidity and mortality making it a global health problem [1]. Decades of research have focused on developing therapeutics targeting the dysregulated immune response that defines sepsis, with little success [2]. However, the success of immunomodulators for the treatment of hospitalized patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has resulted in renewed interest in identifying novel therapeutics for patients with sepsis [3].

Neutrophils play an integral role in the innate immune response during sepsis by serving to clear invading pathogens [4, 5]. When neutrophils become overactivated they can contribute to end-organ dysfunction through excessive degranulation, release of neutrophil extracellular traps (NETs), and generation of reactive oxygen species, making them a therapeutic target [4–6]. Historically, neutrophils were thought to be a homogeneous population; however, in recent years, it has become increasingly recognized that neutrophils are diverse, containing subpopulations with differing phenotypic and functional characteristics [7]. One such subset of neutrophils are low-density neutrophils (LDNs), which were first identified as neutrophils that localize to the peripheral blood mononuclear cell (PBMC) fraction of gradient density separation [8]. While LDNs are characterized in chronic diseases such as systemic lupus erythematous and psoriasis and critical illness caused by SARS-CoV-2 infection, their role in bacterial sepsis is less understood [9–13]. A prior evaluation of LDNs in bacterial sepsis demonstrated variability in markers of maturation, myeloperoxidase (MPO) granule content, and CXC chemokine receptor-4 (CXCR-4) expression; diminished phagocytosis and chemotaxis; and a longer lifespan compared with normal-density neutrophils [12].

Spleen tyrosine kinase (SYK) is a nonreceptor tyrosine kinase present within neutrophils [14]. We previously demonstrated that SYK inhibition decreased neutrophil activation in coronavirus disease 2019 (COVID-19) and in a phase 2 clinical trial SYK inhibition was associated with reduced circulating LDNs in treated patients compared with those randomized to placebo [15–17]. Additionally, we previously reported that R406, the in vitro active component of the United States Food and Drug Administration-approved SYK inhibitor, fostamatinib, impeded lipopolysaccharide (LPS)-induced neutrophil end-effector functions that contribute to poor outcomes in bacterial sepsis [18]. Therefore, we aimed to characterize whole-blood neutrophils (WBNs) and LDNs, along with SYK in a nonhuman primate (NHP) model of bacterial sepsis.

METHODS

Nonhuman Primate Model of Bacterial Sepsis

Approval was received from the National Institute of Allergy and Infectious Diseases (NIAID) Animal Care and Use Committee (ACUC) and Scientific Advisory Board prior to initiation of this study (see Supplementary Materials for more details). We previously developed an NHP model of Klebsiella pneumoniae-induced septic shock that closely emulates the pathogenesis and supportive care provided to humans in modern intensive care units [19, 20]. In this study, 6 NHPs were infected with K. pneumoniae, 2 of which were randomized to fostamatinib, a SYK inhibitor, 3 hours after the onset of infection, as part of an evaluation of the efficacy of fostamatinib in this model (authors remain blinded to study allocation at the time of manuscript submission). Blood was collected in EDTA tubes, whole blood and PBMCs were analyzed by flow cytometry, and plasma was collected at time 0 (T0) (preinfection), and at 6 hours (T6), 24 hours (T24), 48 hours (T48), and at 72 hours, or at time of necropsy. See Supplementary Material for blood processing and flow cytometry methods (Supplementary Table 1). Physiologic, clinical, and immunological data were previously reported for 2 animals (NHP-1 and NHP-2) [19].

Statistical Analysis

Data were analyzed using GraphPad Prism (version 10.3) or Microsoft Excel (version 16.9) and are reported as the mean ± standard error of the mean (SEM) or Pearson correlation coefficient. Statistical significance between groups was determined using a 1-way ANOVA or repeated measures ANOVA. Post hoc analysis for multiple comparisons were completed using Tukey multiple comparisons test. Statistical significance between 2 groups was determined using paired or unpaired Student t test and 2-tailed significance test for Pearson correlation coefficients. P values < .05 were considered statistically significant.

RESULTS

Klebsiella pneumoniae Infusion Leads to Clinical Sepsis

Six NHPs had septic shock induced by intravenous infusion of K. pneumoniae with doses ranging from 1.01E+8 to 6.31E+8 colony-forming units/kg (Table 1). All animals developed tachycardia and hypotension requiring vasopressor therapy; however, with supportive care survived infection (Supplementary Figure 1). Consistent with acute onset bacteremia, all animals developed initial neutropenia according to the absolute neutrophil count followed by neutrophilia, thrombocytopenia, prolonged prothrombin time (PT), and increases in serum creatinine (SCr), aspartate aminotransferase (AST), and alanine aminotransferase (ALT), indicative of coagulopathy, renal failure, and hepatic injury (Table 1 and Supplementary Figures 2 and 3). Additionally, all animals had histopathological evidence of end-organ injury consistent with sepsis (Supplementary Figure 4).

Table 1.

Clinical Characteristics of Nonhuman Primates

Clinical Variables	NHP-1	NHP-2	NHP-3	NHP-4	NHP-5	NHP-6
Demographics
Age, y	11	4	13	7	12	5
Sex	F	M	F	M	F	M
Weight, kg	9.15	6.23	9.38	11.95	7.32	9.46
Bacterial dose, CFU/kg	2.12E+08	2.95E+08	1.01E+08	1.01E+08	6.31E+08	6.31E+08
Hematologic characteristics
BL ANC, μL	8378	7708	9760	9044	6237	6853
Min ANC, μL	432	342	1600	2336	684	441
BL platelets, ×1000 μL	250	196	327	236	277	264
Min platelets, ×1000 μL	27	27	120	47	103	164
BL PT, s	9.3	9.8	9.2	9.8	10.3	9.3
Max PT, s	12.5	17.8	11.5	15.1	14.4	13.1
BL fibrinogen, mg/dL	203	295	246	160	283	205
Fibrinogen range, mg/dL	177–300	177–295	153–442	89.9–342	210–307	205–391
Markers of organ function
BL creatinine, mg/dL	0.9	0.6	0.5	0.7	0.5	0.9
Max creatinine, mg/dL	1.3	1.3	0.9	1.4	0.7	1.4
BL AST, IU/L	74	41	26	53	43	54
Max AST, IU/L	131	137	59	185	75	200
BL ALT, IU/L	43	24	18	30	39	41
Max ALT, IU/L	105	84	47	147	66	116
BL ALP, IU/L	102	380	121	177	137	120
Max ALP, IU/L	263	380	315	344	194	155
Mortality
Lived or died	Lived	Lived	Lived	Lived	Lived	Lived
Time to necropsy, h	54	72	72	72	72	72

Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; ANC, absolute neutrophil count; AST, aspartate aminotransferase; BL, baseline; CFU, colony-forming units; F, female; M, male; Max, maximum; Min, minimum; NHP, nonhuman primate; PT, prothrombin time.

Whole-Blood Neutrophils and Low-Density Neutrophils Are Activated During Sepsis

We performed high-dimensional flow cytometry on whole blood and PBMCs at T0, T6, T24, T48, and necropsy to capture WBNs and LDNs, respectively (Supplementary Figure 5A and 5B). Dimensional reduction of CD66⁺ WBNs combined with clustering by flow cytometry-specific self-organizing map (FlowSOM) identified 19 metaclusters (MC) within the WBN compartment (Figure 1A–C). At baseline, MC07 represented 79.7% of WBNs (Figure 1B). At T6, MC04 dominated the neutrophil subsets, comprising 95.3% of the population. MC04 was characterized as an immature (increased CD33 and decreased CD10) and activated (increased CD11b and decreased CD62L) population, with elevated levels of primary and secondary neutrophil granule components, MPO, and lactoferrin (LTF), respectively (Figure 1C). Appreciable biological diversity was observed at T24 compared to T6, which was composed of MC16, MC15, MC18, and MC19, representing 23.7%, 18.3%, 8.9%, and 5.4% of the WBNs, respectively. Compared to MC04, the MCs from T24 expressed similar levels of CD33 and CD10, indicating they are an immature population, although they had lower CD11b and CD66 and higher CD62L and CD64 (Figure 1C). Additionally, MC16 and MC18 expressed relatively low levels of MPO, potentially indicating they had degranulated, while MC15 and MC19 expressed high levels of MPO. Statistically relevant variations of individual surface and intracellular markers on WBNs over time are shown in Figure 1D, highlighting the changes in neutrophil subsets in circulation in response to bacterial sepsis throughout the duration of the study.

Figure 1.

Characterization of WBNs. A, optSNE plot of flow cytometry data overlaid with FlowSOM MCs from WBNs at all time points and by day. B, The proportion of each MC in parent gate over time determined by FlowSOM. C, Heatmap of the relative expression of surface and intracellular markers between MCs. D, Bar graphs representing the percentage of WBNs as a proportion of all CD45⁺ white blood cells over time and the expression of individual extracellular and intracellular markers over time expressed as MFI. Data are represented as means ± SEM. Significance established by repeated measures 1-way ANOVA and Tukey multiple comparisons test. Significance set at *P < .05, **P < .01, ***P < .001, ****P < .0001. Abbreviations: FlowSOM, flow cytometry-specific self-organizing map; HLA-DR, human leukocyte antigen-DR; LTF, lactoferrin; MC, metacluster; MFI, mean fluorescence intensity; MPO, myeloperoxidase; NX, necropsy; optSNE, optimized t-distributed stochastic neighbor embedding; SYK, spleen tyrosine kinase; T, time; WBN, whole-blood neutrophil.

Evaluation of the PBMC fraction using traditional flow cytometry gating identified a significant increase in the proportion of CD66⁺ LDNs at T6 and T24 (Figure 2A). Therefore, we used dimensional reduction methods (optimized t-distributed stochastic neighbor embedding [optSNE]) combined with FlowSOM for clustering to characterize LDNs at T6 and T24. We identified 12 MCs that varied over time (Figure 2B). Similar to the WBNs, at T6 the LDNs were dominated by a single MC08, which comprised 88.5% of the LDN fraction (Figure 2C). MC08 were phenotypically immature (increased CD33 and decreased CD10), activated (increased CD11b and decreased CD62L), expressed high levels of MPO and LTF, and had high levels of CD87 (urokinase plasminogen activator receptor, associated with neutrophil migration) (Figure 2D). T24 was predominately characterized by 2 populations, MC02 and MC01, which made up 45.1% and 36.1% of LDNs, respectively. MC01 was a more mature population (increased CD10 and decreased CD33) that remained activated (increased CD11b and decreased CD62L) and had low levels of MPO and LTF. On the contrary, MC02 was more immature (increased CD33 and decreased CD10), had decreased CD11b and increased CD62L, and was associated with high levels of MPO (Figure 2D). Similar to WBNs, notable alterations in surface markers and granule proteins were observed over time in LDNs, emphasizing that the subset of neutrophils detected in circulation may be dependent upon the time point following the onset of bacterial sepsis (Figure 2E).

Figure 2.

Characterization of LDNs. A, Bar graph representing the percentage of LDNs as a proportion of CD45⁺ white blood cells over time. B, optSNE plot of flow cytometry data overlaid with MC determined by FlowSOM from pooled LDNs, T6, and T24. C, The proportion of each LDN MC in parent gate at T6 and T24 determined by FlowSOM. D, Heatmap showing the relative surface expression of each LDN surface marker and intracellular marker by MC. E, Bar graphs representing the expression of individual extracellular and intracellular makers on LDNs over time expressed as MFI. Data are represented as means ± SEM. Significance established by 1-way ANOVA and a Tukey multiple comparisons test. Significance set at *P < .05, **P < .01, ***P < .001, ****P < .0001. Abbreviations: FlowSOM, flow cytometry-specific self-organizing map; HLA-DR, human leukocyte antigen-DR; LDN, low-density neutrophil; LTF, lactoferrin; MC, metacluster; MFI, mean fluorescence intensity; MPO, myeloperoxidase; NX, necropsy; optSNE, optimized t-distributed stochastic neighbor embedding; SYK, spleen tyrosine kinase; T, time.

LDNs have been proposed to be a hyperactive, pathogenic subset of neutrophils; therefore, we compared WBNs to LDNs at T6 and T24 (Figure 3A) [9, 10]. At the population level, the LDNs at T6 appeared to be the most immature (increased CD33 and decreased CD10) and expressed the highest levels of MPO and LTF (Figure 3A and 3B). Comparing individual markers, we found that LDNs were phenotypically more divergent than WBNs at T24 compared to T6 (Figure 3C and 3D). At T24, LDNs had higher CD11b (P = .02) and lower CD62L (P = .04), consistent with a classic activated neutrophil phenotype observed in bacterial sepsis, but no difference in CD87 (Figure 3D and Supplementary Figure 6A). Furthermore, T6 and T24 LDNs expressed higher levels of the β1-integrin CD49d (P < .0001 and P = .001) (Figure 3C and 3D). While no differences were seen at T6 in the Fc receptors CD32, CD64, and CD16, we found lower CD32 (P = .002), higher CD64 (P = .005), and higher CD16 (P = .03) at T24 in LDNs compared to WBNs (Figure 3C and 3D). MPO was significantly elevated in LDNs at both T6 and T24 (P = .0007 and P = .01) (Figure 3C and 3D). Similarly, LTF was numerically higher in LDNs at T6 and T24, but did not reach statistical significance (Supplementary Figure 6B).

Figure 3.

Characterization of WBNs versus LDNs. A, optSNE plot and heatmap representing surface and intracellular relative expression levels in WBNs and LDNs at T6 and T24. B, optSNE plot representing the expression of CD11b, CD62L, CD49d, CD32, CD64, CD16, and MPO across WBNs and LDNs at T6 and T24. C, Bar graphs representing differences in the expression of CD11b, CD62L, CD49d, CD32, CD64, CD16, and MPO in WBNs and LDNs at T6. D, Bar graphs representing differences in the expression of CD11b, CD62L, CD49d, CD32, CD64, CD16, and MPO in WBNs and LDNs at T24. Data are represented as means ± SEM. Significance established by Student t test. Significance set at *P < .05, **P < .01, ***P < .001, ****P < .0001. Abbreviations: HLA-DR, human leukocyte antigen-DR; LDN, low-density neutrophil; LTF, lactoferrin; MFI, mean fluorescence intensity. MPO, myeloperoxidase; optSNE, optimized t-distributed stochastic neighbor embedding; SYK, spleen tyrosine kinase; T, time; WBN, whole-blood neutrophil.

Whole-Blood and Low-Density Neutrophils Express SYK During Bacterial Sepsis

Based on our prior work in COVID-19, we sought to determine if SYK inhibition could represent a therapeutic approach in bacterial sepsis [15]. Dimensional reduction demonstrated that SYK expression is highest at T6 in WBNs, which was confirmed by quantifying SYK MFI over time (Figure 4A). The percentage of SYK⁺ WBNs at T6 increased in response to infection, although did not reach statistical significance, and significantly decreased over time (Figure 4A). Similar to WBNs, LDNs had the highest amount of SYK expression at T6 compared to T0, T24, T48, and necropsy (Figure 4B). Furthermore, the percentage of SYK⁺ LDNs was elevated at T6 and T24 compared to T0 and significantly decreased at T48 and necropsy (Figure 4B).

Figure 4.

Characterization of SYK in whole blood and LDNs. A, optSNE plot of SYK expression in all samples and T6 and T24 hours in WBNs; bar graphs representing the MFI of SYK in WBNs across time and the percent of SYK⁺ WBNs as a proportion of CD45⁺ white blood cells and CD66⁺ WBNs. B, optSNE plot of SYK expression in all samples and T6 and T24 hours individually in LDNs; bar graphs representing the MFI of SYK in LDNs across time and the percent of SYK⁺ LDNs as a proportion of CD45⁺ white blood cells and CD66⁺ LDNs. C, Histograms and bar graph of SYK expression in WBNs and LDNs at T6 and T24, and bar graphs of SYK MFI in SYK⁺ WBNs and SYK⁺ LDNs at T6 and T24. D, Correlation of SYK⁺ LDNs AUC with bacterial dose. Data are represented as means ± SEM. Significance established by repeated measures 1-way ANOVA or 1-way ANOVA and Tukey multiple comparisons test or Student t test. Significance set at *P < .05, **P < .01, ***P < .001, ****P < .0001. Abbreviations: AUC, area under the curve; CFU, colony forming unit; LDNs, low-density neutrophils; MFI, mean fluorescence intensity; NX, necropsy; optSNE, optimized t-distributed stochastic neighbor embedding; SYK, spleen tyrosine kinase; T, time; WBNs, whole-blood neutrophils.

We then analyzed SYK expression between WBNs and LDNs because our COVID-19 clinical trial showed a reduction in frequency of LDNs in response to SYK inhibition [21]. We determined that LDNs have heightened levels of SYK compared to WBNs at T6 (P = .02), although the increase did not reach statistical significance at T24 (Figure 4C). Interestingly, the highest expression of SYK overall was in LDNs at T6 (Figure 4C). Within SYK⁺ neutrophils, SYK⁺ LDNs expressed higher levels of SYK than SYK⁺ WBNs at both T6 (P = .02) and T24 (P = .02) (Figure 4C). Additionally, SYK⁺ LDNs expressed vast amounts of MPO at T6 and T24 compared to SYK⁺ WBNs (P = .0008 and P = .01) (Supplementary Figure 7A).

We next compared expression of intracellular and surface markers amongst the SYK⁺ WBNs and SYK⁺ LDNs at T6 and T24 (Supplementary Figure 7B). SYK⁺ LDNs at T6, the population that expressed the highest levels of SYK overall, expressed higher levels of MPO and LTF and appeared to have a more immature phenotype (increased CD33 and decreased CD10) compared to other SYK⁺ populations (Supplementary Figure 7B). When comparing SYK⁺ LDNs to SYK⁺ WBNs at T6 and T24, SYK⁺ neutrophils expressed higher CD11b, CD66, CD87, and lower CD62L indicating they were activated and immature (decreased CD33), consistent with a population of neutrophils that was recently released from the bone marrow due to emergency granulopoiesis [22]. SYK⁺ LDNs uniformly (T6 and T24) expressed higher MPO and CD49d compared to the SYK⁺ WBN populations (Supplementary Figure 7B). Lastly, we determined that the area under the curve (AUC) for SYK⁺ LDNs over time correlated with bacterial dose, suggesting that SYK⁺ LDNs correlate with disease severity (Figure 4D).

Differential Expression of Surface Markers in SYK⁺ Versus SYK⁻ Neutrophils

After determining SYK is elevated during bacterial sepsis, we assessed differences between SYK⁺ versus SYK⁻ neutrophil populations. The clustered heatmap of SYK⁺ and SYK⁻ WBNs showed WBNs more closely clustered according to time point than to SYK expression (Supplementary Figure 8A). This may be due to a small percentage of SYK⁻ cells present at T6 (Supplementary Figure 8B and 8C). Differences in SYK⁺ versus SYK⁻ at T6 included a trend towards higher CD11b (P = .07), a finding that was statistically significant at T24 (P = .04), and higher CD32 at T6 and T24 (P = .02 and P = .03) (Supplementary Figure 8D–F). SYK⁺ WBNs had lower CD64 (P = .002) and CD16 (P = .06) compared to SYK⁻ WBNs that became equivalent by T24 (Supplementary Figure 8D–F). LTF had higher expression in SYK⁺ WBNs at 24 hours (P = .01) (Supplementary Figure 8G and 8H) and MPO trended towards higher levels in SYK⁺ WBNs versus SYK⁻ WBNs at both T6 and T24 (P = .07 and P = .06, respectively) (Supplementary Figure 8I and 8J).

Contrary to WBNs, SYK⁺ LDNs and SYK⁻ LDNs clustered according to SYK expression rather than time point with a small SYK⁻ population observed at T6 (Supplementary Figure 9A–C). Relative expression of surface markers suggests that SYK⁺ LDNs are more immature (higher CD33 and lower CD10) and contain more LTF and MPO compared to SYK⁻ LDNs (Supplementary Figure 9A). Similar to WBNs, SYK⁺ LDNs trended towards lower CD11b at T6 (P = .09) and higher levels of CD62L at T6 and T24 (P = .02 and P = .04) (Supplementary Figure 9D–F). Interestingly, the major proportion of CD62L at T24 is expressed on SYK⁺ LDNs (Supplementary Figure 9D–F). CD87 directionality was dependent on timing as T6 SYK⁺ LDNs expressed lower CD87 (P = .02) and T24 had higher CD87 expression (P = .0001) compared to SYK⁻ LDNs. SYK⁺ LDNs expressed higher CD32 (P = .03) and lower CD64 (P = .02) at T6, while there were no differences in expression of Fc receptors at T24 (Supplementary Figure 9E and 9F). LTF was increased at both T6 (P = .003) and T24 (P = .01) in SYK⁺ LDNs, while more MPO was detected at T24 (P = .03) compared to SYK⁻ LDNs (Supplementary Figure 9G–J).

SYK Expression on Neutrophils Is Associated With Renal Failure and Coagulopathy

Because hyperactive neutrophils are suggested to contribute to end-organ damage through NET release and excessive degranulation [6], we next sought to determine if SYK⁺ neutrophil subsets associated with end-organ dysfunction and how this related to soluble MPO and MPO-DNA complexes (Figure 5). The percentage of SYK⁺ WBNs/CD66 had a positive correlation with PT (r = 0.42, P = .02) and a negative correlation with fibrinogen (r = −0.68, P = .00004), findings consistent with disseminated intravascular coagulopathy. Furthermore, SYK⁺ WBNs/CD66 had positive correlation with SCr (r = 0.50, P = .004) and a negative correlation with alkaline phosphatase (r = −0.39, P = .03) (Figure 5A). Similarly, the proportion of SYK⁺ LDNs/CD66 correlated with PT (r = 0.67, P = .0002), fibrinogen (r = −0.40, P = .04), SCr (r = 0.44, P = .02), and alkaline phosphatase (r = −0.65, P = .0004) (Figure 5B). Because SYK inhibition has been demonstrated to inhibit MPO and NETs release in vitro, we measured soluble MPO and MPO-DNA complexes over time (Supplementary Figure 10A and 10B) and found MPO had a strong correlation with SCr (r = 0.48, P = .007), while MPO-DNA complexes associated with fibrinogen (r = 0.43, P = .02) (Figure 5C and 5D). Neither SYK⁺ population had a correlation with serum AST or ALT.

Figure 5.

Correlations with neutrophil subsets and end-organ dysfunction. A, Heatmap representing the correlations of WBNs and WBN subsets with clinical makers of end-organ dysfunction. Size and color of the bubbles indicate strength and directionality of the correlation. B, Heatmap representing the correlations of LDNs and LDN subsets with clinical makers of end-organ dysfunction. C, Scatter plot of the correlation of soluble MPO with creatinine. D, Scatter plot of the correlation of soluble MPO-DNA complexes with fibrinogen. Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; ANC, absolute neutrophil count; AST, aspartate aminotransferase; FIB, fibrinogen; LDN, low-density neutrophil; MPO, myeloperoxidase; MPO-DNA, myeloperoxidase-deoxyribonucleic acid; PLT, platelet; PT, prothrombin time; SCr, serum creatinine; WBN, whole-blood neutrophil.

Tissue MPO Levels Associate With LDNs and Tissue SYK Levels Associate With SYK⁺ LDNs

Examination of tissue from lung, kidney, and liver identified an increase in SYK⁺ and MPO⁺ cells compared to uninfected healthy controls (Supplementary Table 2). SYK⁺ cells were most noted in alveolar capillaries and large pulmonary vessels of the lung, in peritubular vessels, glomerular capillaries and apical portion of cortical tubular epithelial cells, and the hepatic vasculature (sinusoids, central veins, and portal veins) (Figure 6A). Interestingly, the proportion of circulating SYK⁺ LDNs correlated with the amount of SYK in tissue from the lung (r = 0.94, P = .006) and kidney (r = 0.81, P = .049) (Figure 6B). Similarly, MPO⁺ cells were most noted in alveolar capillaries, larger pulmonary vessels, and occasionally in alveolar lumina of the lung, circulating capillaries of the kidney, and sinusoids of the liver (Figure 6C). Interestingly, the maximum percentage of circulating LDNs in the peripheral blood strongly correlated with tissue MPO levels measured in the lung (r = 0.90, P = .01) and liver (r = 0.85, P = .03) (Figure 6D).

Figure 6.

MPO and SYK are found in tissue and correlate with LDNs and SYK⁺ LDNs. A, Representative images of SYK staining in the lung, liver, and kidney of nonchallenged controls and Klebsiella pneumoniae-infected NHPs. B, Correlation of percent SYK⁺ LDNs (maximum) with the percentage SYK positive cells in lung and kidney tissue. C, Representative images of MPO staining in the lung, liver and kidney of nonchallenged controls and K. pneumoniae infected. D, Correlation of percent LDNs (maximum) with percentage MPO-positive cells in lung and liver tissue. Each data point represents 1 NHP. Abbreviations: LDN, low-density neutrophil; MPO, myeloperoxidase; NHP, nonhuman primate; SYK, spleen tyrosine kinase.

DISCUSSION

Due to high morbidity and mortality, there is significant interest in gaining further insight into the pathogenesis of bacterial sepsis and identifying novel therapeutic targets [23]. Here we demonstrate that bacterial sepsis in a NHP model leads to an elevation in LDN populations with diverse phenotypic characteristics, including increased expression of MPO. We further demonstrate that both WBNs and LDNs express elevated SYK after the onset of infection and SYK expression associated with impaired renal function and coagulation parameters consistent with disseminated intravascular coagulopathy, suggesting SYK as a novel therapeutic target in bacterial sepsis.

We demonstrate that in our NHP model, LDNs are rapidly induced within 6 hours of infection onset. These LDNs persist beyond 6 hours and appear to be more activated than WBNs 24 hours after the onset of infection, with increased expression of CD11b and CD64 and decreased CD62L. We also noted LDNs express higher levels of MPO than WBNs, indicating that excessive degranulation may be a mechanism by which LDNs contribute to sepsis pathogenesis [21]. LDNs, compared to WBNs consistently expressed elevated CD49d, a β1-integrin that binds to CD106 and fibronectin on activated endothelium [24]. The phenotype of increased CD11b (β2-integrin) and CD49d could indicate that these neutrophils are primed to attach to endothelium and migrate into the tissue, where they can contribute to end-organ dysfunction. Interestingly, it is noted that integrin signaling in neutrophils is mediated through SYK [25].

The origin of LDNs is not explicitly known but 2 main hypotheses exist. The first is that LDNs are a population of immature neutrophils released during the process of emergency granulopoiesis and the second is that they are neutrophils that have degranulated resulting in a buoyancy change [26–28]. Our data support both of these hypotheses. At the T24 time point we observed 2 different populations of LDNs, one with an immature phenotype and high MPO (MC02), suggestive of a neutrophil population released from the bone marrow as a result of emergency granulopoiesis, and another mature population with low levels of MPO (MC01), indicating that these neutrophils have potentially degranulated and released MPO. These divergent neutrophil populations based on MPO expression were also observed in our analysis of WBNs (MC16 and MC18 vs MC15 and MC19). It is also noted that, traditionally, polymorphonuclear-myeloid derived suppressor cells, a subset of neutrophils that were demonstrated to inhibit T-cell activation and proliferation during critical illness, are purified following gradient density separation, providing additional evidence that LDNs are a diverse population of neutrophils [15, 29, 30]. Further phenotypic and functional characterization of LDNs will help elucidate the many different subsets that comprise LDNs.

SYK is a novel therapeutic target that was recently suggested for the treatment of patients with COVID-19 and mechanistically was associated with decreased neutrophil activation and lower proportions of circulating LDNs [15, 16]. We found that SYK is elevated in WBNs and LDNs 6 hours after the onset of infection and that LDNs express more SYK than WBNs. Furthermore, SYK⁺ WBNs and SYK⁺ LDNs both appear to express increased levels of MPO and LTF compared to their SYK⁻ counterparts.

Sepsis is defined as an infection with end-organ dysfunction, indicated by a change in sequential organ failure score [31]. We found that both SYK⁺ WBNs and SYK⁺ LDNs associated with serum creatinine levels. Furthermore, SYK⁺ WBNs and SYK⁺ LDNs both had a positive correlation with PT and negative correlation with fibrinogen, indicating an association with disseminated intravascular coagulopathy. Lastly, we found that SYK⁺ LDNs associated with the intensity of SYK staining in tissue of the lung and kidney at the time of necropsy. The observation that SYK⁺ LDNs decrease over time and that the maximum proportion of SYK⁺ LDNs early in the study correlates with tissue levels of SYK suggests that SYK⁺ LDNs migrate through the endothelium into tissue over time, contributing to pathogenesis. Additionally, elevated SYK expression has been noted in circulating CD34⁺ progenitor cells during COVID-19 infection, indicating that SYK inhibition could impact the release of immature neutrophils during sepsis [32]. Taken together, elevated SYK, particularly in LDNs, the association with end-organ dysfunction, and the presence of SYK in tissue at necropsy suggests SYK inhibition may be a potential therapeutic target in bacterial sepsis.

In vitro studies using LPS stimulation of healthy donor neutrophils have demonstrated that SYK inhibition is able to decrease the release of NETs, reactive oxygen species, degranulation, and adhesion of neutrophils to endothelial cells, while other important end effector functions such as phagocytosis, migration, and release of proinflammatory cytokines remain intact, furthering the mechanistic link for SYK inhibition in bacterial sepsis [18]. We noted that SYK⁺ LDNs express higher levels of MPO than SYK⁻LDNs and that LDNs express more MPO than WBNs, indicating that degranulation may be an important mechanism of SYK inhibition. The impact of SYK inhibition on degranulation is of particular interest because excessive degranulation and release of MPO has been associated with disease severity in sepsis [21]. Furthermore, SYK inhibition was demonstrated to decrease mortality across multiple mouse models of sepsis [33–36]. Mechanistically, in an LPS-induced lung injury model, SYK inhibition resulted in decreased neutrophilic lung inflammation, vascular permeability, and MPO activity. Similar effects on neutrophil activation were observed in a model of LPS-induced acute kidney injury [33, 34].

Limitations of our study include the small sample size of 6 NHPs and varying doses of K. pneumoniae infusion, limiting some conclusions, including associations with end-organ dysfunction, which will need to be confirmed in larger human cohorts. However, the consistent findings we observed using unbiased dimensional reduction and traditional gating strategies across a biologically diverse group of animals makes our findings closely applicable to human disease. Additionally, all animals survived the infection, limiting our ability to evaluate LDNs and SYK expression through the entire course of sepsis. However, the observation that LDNs and SYK expression normalize over time strengthens their association with sepsis disease severity. Lastly, 2 of the animals in this study are part of an ongoing randomized and blinded study evaluating the efficacy of SYK inhibition in our NHP model. The authors remain blinded at the time of this study, and thus are not able to determine if study drug allocation impacted neutrophil populations or the association of SYK-positive neutrophils with end-organ dysfunction.

In conclusion, we performed a detailed characterization of WBNs, LDNs, and SYK expression providing further insight into the role of the innate immune responses during bacterial sepsis. Specifically, we identified that WBNs and LDNs are phenotypically diverse populations of innate immune cells that express elevated SYK and that SYK expression associates with end-organ dysfunction. Future studies to evaluate the efficacy of SYK inhibition in NHPs with bacterial sepsis are warranted.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.