Between November, 2020 and December, 2021, we recruited 64 patients who tested positive for SARS-CoV-2 by nasopharyngeal swab and were assessed to have severe or critical COVID-19, as defined by the WHO criteria (38). The study participant demographics and summary of clinical information are shown in Table 1. The mean age was 59.8 ± 14.6 years. ICU admission was required in 59 of the 64 patients, 57 required mechanical ventilatory support, and 6 were supported with ECMO. The PaO2:FiO2 was below 100 in 25 patients, between 100 and 200 in 29, between 200 and 300 in 7, and above 300 in 3 who required mechanical ventilation. Comorbidities known to be risk factors for severe COVID-19 were present in 56 of the 64 patients, the most common of which were hypertension (n=39, 61%) and obesity (n=42, 66%). Elevated neutrophil to lymphocyte ratio (NLR), D-dimer, ferritin, and C-reactive protein were common, as previously reported (16, 39–42). The mean/median hospital day on which neutrophils used in this study were obtained was 8.5/6.5. Treatment with corticosteroids, remdesivir, and monoclonal antibodies or convalescent plasma occurred in 59 (92%), 48 (75%), and 29 (45%) patients, respectively. Outcome was poor, as 32 (50%) patients died during hospitalization and 19 (30%) were discharged to a long-term care or rehabilitation facility. A total of 26 healthy donors (HD) were recruited from the students, staff, and faculty at the University of Louisville. Demographic data was not recorded for HD, but the pool of subjects eligible for recruitment was younger than the patient population.

Neutrophil phagocytosis and respiratory burst activity are critical for host defense function (43), and release of ROS is suggested as a contributing mechanism of lung injury leading to ARDS in COVID-19 (44, 45). Previous reports presented contradictory results for both neutrophil phagocytosis and respiratory burst activity (8, 25), and the response of LDN has not been reported. To address this gap in knowledge, we measured bacterial phagocytosis and phagocytic stimulation of H₂O₂ production in LDN and NDN neutrophils isolated from 11 COVID-19 patients and compared their responses to those of NDN from 6 HD. Figure 1A shows phagocytosis before and after in vitro priming with TNFα. Phagocytosis by unprimed cells was significantly greater in COVID NDN, compared to HD NDN. COVID LDN showed a level of phagocytosis intermediate between the other 2 groups and was not significantly different from either. TNFα priming significantly enhanced phagocytosis of both HD NDN and COVID NDN, with a significantly greater response by COVID NDN. Phagocytosis by COVID LDN did not increase with in vitro priming. Figure 1B shows phagocytosis-stimulated H₂O₂ production in the three groups of neutrophils. H₂O₂ production by COVID NDN, but not COVID LDN, was significantly greater than that by HD NDN. TNFα priming significantly enhanced H₂O₂ production to an equivalent degree in all three groups, with COVID NDN production remaining significantly greater than HD NDN or COVID LDN. We interpret these results to indicate that NDN from severe COVID-19 patients are primed for enhanced phagocytosis and respiratory burst. Despite existing in the same milieu, LDN are not primed for either enhanced phagocytosis or respiratory burst activity. Respiratory burst activity by COVID LDN was enhanced to the same degree as HD NDN in response to in vitro priming, indicating LDN are responsive to some priming agents.

Previous neutrophil degranulation studies using expression of plasma membrane markers showed enhanced azurophilic granule exocytosis, measured by CD63, in COVID-19 (28, 29, 31). Contradictory results were reported for exocytosis of specific (CD66b) and gelatinase (CD11b) granules (9, 28, 29, 31, 32, 34, 35, 46). To evaluate degranulation of neutrophil populations from COVID-19 patients, we compared expression of plasma membrane markers of the 4 neutrophil granule subsets, CD35 (secretory vesicles), CD11b (gelatinase granules), CD66b (specific granules), and CD63 (azurophilic granules), among HD NDN, COVID NDN, and COVID LDN before and after stimulation (Figure 2A). Basal CD35 expression was significantly greater in COVID NDN and COVID LDN than in HD NDN, indicating that similar secretory vesicle degranulation occurred in vivo in LDN and NDN in COVID-19 patients. Stimulation with TNFα and fMLF increased CD35 expression in HD NDN and COVID NDN, but not COVID LDN, suggesting LDN underwent maximal release of secretory vesicles in vivo. Basal CD11b expression by COVID NDN was significantly greater than expression by HD NDN or COVID LDN. Additionally, TNFα and fMLF stimulation increased CD11b expression in HD NDN and COVID NDN, but not in COVID LDN. Basal CD66b expression was significantly greater in COVID NDN, but not COVID LDN, than in HD NDN. In vitro stimulation resulted in a significant increase in CD66b in all 3 groups of cells. CD63 expression did not differ between HD NDN and COVID NDN. CD63 expression by COVID LDN was significantly lower than either group of NDN. The response of CD63 to in vitro stimulation was not tested, as pharmacologic disruption of the actin cytoskeleton is required (47). Thus, markers for the different neutrophil granule subsets showed distinct patterns of expression, suggesting differences in granule marker expression or in granule mobilization both among granule subsets and between NDN and LDN in COVID-19. We interpret these results to indicate that severe COVID-19 is associated with enhanced granule mobilization in NDN, except for the granule subset requiring the most potent stimulus, azurophilic granules. LDN demonstrate an impaired in vivo response to degranulation, except for the granule subset most easily induced to undergo exocytosis, secretory vesicles.

We previously reported that an intermediate level of CD16 expression (CD16^Int) identifies a distinct subpopulation of LDN that emerges in severe COVID-19 (16). CD16^Int LDN have an activated genetic and proteomic phenotype, and their prevalence in the circulation predicts disease severity. CD16^Int and CD16^Hi LDN obtained by cell sorting and NDN isolated by gradient centrifugation were available from 7 COVID-19 patients to compare expression of neutrophil granule markers. Figure 2B shows CD16^Hi LDN express CD35, CD11b, CD66b, and CD63 at the same levels as COVID NDN. On the other hand, CD16^Int LDN showed reduced CD35, CD11b, and CD63 expression, while CD66b expression was increased. We interpret these data to indicate that CD16^Int LDN represent a unique subpopulation of LDN with impaired mobilization of secretory vesicles, gelatinase granules and azurophilic granules, while specific granule mobilization is enhanced.

Previous studies found elevated levels of granule constituents in the plasma of COVID-19 patients (21, 31, 48–51). To confirm our granule marker results, constituents of gelatinase granules (MMP9), specific granules (lactoferrin), and azurophilic granules (human neutrophil elastase, HNE) were quantified in plasma from HD and severe COVID-19 patients. Figure 2C shows that all three markers were significantly increased in plasma from COVID-19 patients. Thus, plasma concentrations of granule contents and neutrophil plasma membrane expression of granule markers present consistent findings for gelatinase and specific granules, while CD63 expression failed to conform with azurophilic granule mobilization detected by HNE release into plasma.

Enhanced NET formation by circulating neutrophils in COVID-19 is well established, and NETs are observed in the lungs of patients with COVID-19 ARDS and in areas of microthrombosis (10–17). We compared spontaneous and bacterially-stimulated NET formation by isolated HD NDN, COVID NDN and COVID LDN by confocal microscopy. Figure 3A shows a representative merged confocal microscopic image of lactoferrin and DAPI staining of NETs under basal and stimulated conditions, and individual images of each stain are shown in Supplemental Figures 1, 2. Minimal extracellular lactoferrin was observed directly adjacent to unstimulated HD NDN and COVID NDN. On the other hand, NETs were observed emanating from many unstimulated COVID LDN. Incubation with the emerging oral pathogen, Peptoanaerobacter stomatis, for 3 hr resulted in a greater number and larger patches of nets in COVID LDN, compared to either COVID NDN or HD NDN. Enlarged images confirming co-localization of extracellular staining for lactoferrin and DAPI, indicating lactoferrin and dsDNA co-localize outside of cells, are shown in Supplemental Figures 3, 4. To quantify NET release, extracellular dsDNA was measured in the same groups of cells. Figure 3B shows that although NET formation in unstimulated COVID LDN and COVID NDN was numerically greater than in HD LDN, those differences were not statistically significant. All three neutrophil groups showed a significant increase in NET formation following incubation with P. stomatis for 3 hr. The release of dsDNA was significantly greater following bacterial stimulation in COVID LDN, compared to either COVID NDN or HD NDN. Three sets of sorted CD16^Hi and CD16^Int LDN were available to compare NET formation in LDN subsets. As shown in Figure 3C, NET formation occurred in a greater number of CD16^Int LDN for each set, however, those differences were not statistically different in this limited data set. Taken together, our data suggest that COVID LDN possess a higher capacity for stimulated NET formation, and NET formation may be greater in CD16^Int LDN than CD16^Hi LDN.

The genetic signature of LDN has been interpreted to indicate at least a portion of this neutrophil population has immunosuppressive activity (23, 25). Evaluation of neutrophil supernatants or neutrophil-T cell interaction showed impairment of T cell proliferation (52, 53). To determine the effect of different neutrophil populations on T cell activation, we determined the effect of NDN and CD16^Hi/Int LDN from 6 COVID-19 patients or HD NDN on T cell activation. T cells from a single donor were labeled with CellTracker dye CFSE and then incubated with NDN, CD16^Hi LDN, or CD16^Int LDN from COVID-19 patients or HD NDN. The distribution of those 3 neutrophil populations in 12 COVID-19 patients is shown in Figure 4A. NDN accounted for 37.4 ± 8.9% of total neutrophils, CD16^Hi LDN for 54.7 ± 8.6, and CD16^Int LDN for 7.4 ± 2.0%. T cell proliferation was assessed based on CFSE dilution, and cell culture supernatants were harvested to measure IFN-γ levels. While co-culture of CD16^Int LDN from COVID-19 patients stimulated proliferation of CD4⁺ and CD8⁺ T cells to the same degree as HD NDN, CD16^Hi LDN and COVID NDN failed to stimulate CD8⁺ and CD4⁺ T cell proliferation (Figure 4B). Co-culture with HD NDN significantly stimulated IFN-γ production. On the other hand, COVID NDN, CD16^Int LDN, and CD16^Hi LDN all failed to stimulate IFN-γ production (Figure 4C).

The Institutional Review Board at University of Louisville approved the present study and written informed consent was obtained from subjects or their legal authorized representatives (IRB No. 20. 0321). Patients were recruited between November, 2020 and December, 2021. Inclusion criteria were all hospitalized adults (older than 18) who have positive SARS-CoV-2 test results and were consented to this study. Exclusion criteria included age younger than 18 and refusal to participate. COVID-19 patients enrolled in this study were diagnosed with a 2019-CoV detection kit using real-time reverse transcriptase–polymerase chain reaction performed at the University of Louisville Hospital Laboratory from nasal pharyngeal swab samples obtained from patients. The identification of COVID-19 patients with severe/critical disease was based on the requirement for high-flow oxygen support or the need for mechanical ventilation and this group had blood draw along with their standard laboratory work. All COVID-19 patients were followed by the research team daily and the clinical team was blinded to findings of the research analysis to avoid potential bias.

The demographic characteristics (age, sex, height, weight, Body Mass Index (BMI) and clinical data (symptoms, comorbidities, laboratory findings, treatments, complications, and outcomes) were collected prospectively. All data were independently reviewed and entered into a computer database. For hospital laboratory complete blood count (CBC) tests, normal values are the following: white blood cell (4.1-10.8 x10³/μL); hemoglobin (13.7-17.5g/dL); platelet (140-370 x10³/μL). For hospital laboratory inflammatory and coagulation markers, normal values are the following: D-dimer (0.19-0.74 μg/ml Fibrinogen Equivalent Units); ferritin (7-350 ng/ml); lactate dehydrogenase (LDH) (100-242 Units/Liter); C-reactive protein (<3.0 mg/L).

Table 1 shows that blood for neutrophil isolation was collected an average of 8.5 days (median 6.5 days) after hospitalization (range 1 to 36 days). Blood was collected into citrate. Neutrophils were isolated from the blood of healthy donors using plasma-Percoll gradients as previously described (76, 77). Microscopic evaluation of isolated cells showed that >92% of cells were neutrophils. Trypan blue exclusion indicated that >95% of cells were viable. Neutrophils isolated by gradient centrifugation were used in flow cytometric experiments examining phagocytosis, H₂O₂ generation, and exocytosis, as neutrophils were specifically gated. The Institutional Review Board of the University of Louisville approved the use of human donors who provided informed consent (IRB No. 96.0191).

Alternatively, neutrophils were isolated from whole blood of healthy donors and COVID-19 patients using an EasySep Direct Human Neutrophil Isolation Kit (STEMCELL Technologies, Cat# 19666) accordingly to the manufacturer’s instructions and >95% purity was obtained. The purified PMNs were resuspended in colorless RPMI and experiments were performed at 37°C in 5% CO₂. Neutrophils isolated by this method were used in NET assays.

PBMCs were isolated from peripheral blood using a Ficoll gradient centrifugation (Lympholyte^®-H solution, CEDARLANE) followed by isolation of the buffy-coat layer. The cells were then washed using complete RPMI-1640 (cRPMI) (Sigma). The cell pellet was washed with sterile, cold PBS and added with Fc block (Human TruStain FcX™, Biolegend) to incubate at 4°C for 10 minutes followed by staining with anti-human monoclonal antibodies (mAb) for CD66b-PE (Clone G10F5, Biolegend) and CD16-APC (Clone 3G8, Biolegend) and with the viability dye eFluor 780 (eBioscience), then incubated at 4°C for 20 minutes in dark. The cells were washed with 3-5 ml of MACs buffer (Miltenyi Biotech) by centrifugation at 500 g for 5 minutes. The pellet was then suspended in MACs buffer, filtered, and the cells sorted with a BD FACsAria III sorter to collect the CD16^hi (gated as viable, CD66b⁺CD16^hi) and CD16^int LDNs (gated as viable, CD66b⁺CD16^int) using separate collection tubes containing a mixture of 50% Fetal Bovine Serum (Atlanta Biologicals), 40% PBS (Sigma) and 10% HEPES buffer (Corning). After collection, the collection tubes were centrifuged at 500 g at 4°C for 8 minutes, washed once with cRPMI, counted and resuspended in required volumes of cRPMI.

To measure H₂O₂ production, isolated neutrophils or PBMCs (4 x 10⁶ cells/ml) were incubated with 2’,7’-dichlorofluorescin diacetate (final concentration 0.5 mM; Molecular Probes/Invitrogen, Carlsbad, CA, USA) for 10 min at 37°C. Fifty microliters of cell suspension was sampled before, and 10 min after, the addition of 50 ml of opsonized, propidium iodide-labeled Staphylococcus aureus (final concentration ~10⁸ bacteria/ml). Samples were analyzed for uptake of labeled bacteria and oxidation of 2’,7’-dichlorofluorescin diacetate to fluorescent 2’,7’-dichlorofluorescein by flow cytometry as previously described (57, 76). The LDN population in the PBMC fraction was defined for analysis by gating on CD66b⁺ cells.

Exocytosis of secretory vesicles, gelatinase granules, specific granules, and azurophilic granules in HD NDN, COVID NDN, and COVID LDN was determined by measuring the increase in plasma membrane binding of FITC-conjugated monoclonal anti-CD35 (clone E11; Pharmingen, San Diego, CA, USA), ALEXA-Fluor 488-conjugated monoclonal anti-CD11b (BD Biosciences, San Jose, CA, USA), FITC-conjugated monoclonal anti-CD66b (clone CLB-B13.9; Accurate Chemical, Westbury, NY, USA), and FITC-conjugated anti-CD63 (clone AHN16.1/46-4-5; Ancell Corporation, Bayport, MN, USA), respectively, on 4 x 10⁶/ml neutrophils using a FACSCalibur flow cytometer (Becton Dickenson, Franklin Lakes, NJ, USA) as previously described (47, 57). The LDN population in the PBMC fraction was defined for analysis by gating on CD66b⁺ cells.

For low density neutrophil subsets, Single cell suspensions of PBMCs/Neutrophils were washed with 1 ml PBS (SIGMA) by centrifugation at 500 g for 5 minutes at 4°C. The pellets were then stained with anti-human CD66b-PE (Clone G10F5, Biolegend), anti-human CD16-APC (Clone 3G8, Biolegend), anti-human CD35-FITC (Clone 9H3, Biolegend), anti-human CD11b-PE-Cy7 (Clone CBRM1/5, Biolegend), anti-human CD63-APC-Cy7 (Clone H5C6, Biolegend) and anti-human CD10-PerCP-Cy5.5 (Clone HI10a, Biolegend) monoclonal antibodies, briefly vortexed and incubated for 30 minutes at 4°C in dark. After incubation, cells were washed with ice-cold PBS by centrifugation at 500 g for 5 minutes at 4°C. The pellets were finally suspended in 250-300 µl of cold PBS, briefly vortexed and acquired by BD FACsCanto for analysis.

Plasma concentrations of constituents of gelatinase granules (MMP9, R&D Systems, Minneapolis, MN, USA), specific granules (lactoferrin, Abcam, Cambridge, MA, USA), and azurophilic granules (HNE, Abcam, Cambridge, MA, USA) were determined by commercial ELISA kits, according to the manufacturers’ instructions.

Peptoanaerobacter stomatis strain CM2 (provided by Dr. Slava Epstein, Northeastern University) was cultivated in tryptic soy broth (TSB) supplemented with 20 g/L yeast extract, 1.0% hemin and 1.0% reducing agent (37.5 g/L NH₄Cl, 25 g/L MgCl₂ x 6H₂O, 5 g/L CaCl₂ x 2H₂O, 50 g/L L-cysteine HCl, 5 g/L FeCl₂ x 4H₂O) overnight at 37°C under anaerobic conditions as previously described (78). Overnight culture of P. stomatis was diluted in fresh growth media and incubated under anaerobic conditions for 4-5 hours until reaching an OD₆₀₀ between 0.4 to 0.6. For confocal assays, P. stomatis was labeled with carboxyfluorescein succinimidyl ester (CFSE, 0.25mg/ml; Invitrogen, USA) for 30 min at room temperature in the dark and washed 3 times with PBS prior to its use.

For confocal imaging, neutrophils 1 × 10⁶ cells/experimental condition in RPMI media without phenol red supplemented with 10mM HEPES were seeded onto poly-L-Lysine-coated 12 mm glass coverslips that were placed in 24-well dishes. Cells were allowed to attach for 30 minutes at 37°C, after which, CFSE-labeled P. stomatis (10 MOI) was added to appropriate wells and the plate centrifuged for 10 min at 600 rcf to synchronize phagocytosis. Untreated and treated cells were incubated for an additional 3h at 37°C. After fixation with paraformaldehyde (2% final concentration) for 30 min at room temperature, cells were blocked in 1% BSA/PBS for 1h at room temperature, followed by incubation with anti-lactoferrin antibody (MP Biomedicals #55040; 1:500), in 1% BSA/PBS, overnight at 4°C. Primary antibody was washed off with three 5 min PBS washes, followed by incubation with donkey anti-rabbit Alexa-Fluor 546 (Thermo Fisher #A10040; 1:1000) in 1% BSA/PBS for 1h at room temperature. Nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole), and coverslips were mounted onto glass slides using ProLong Gold Antifade reagent (Thermo Fisher). Images were acquired using an Olympus Fluoview FV-1000 confocal coupled to an Olympus 1X81 inverted microscope, 20× objective, and FV-10 ASW 2.1 software. A multi-channel scanning configuration with sequential line scanning was setup for acquisition of DAPI, FITC (P.stomatis staining), and AF546 (lactoferrin immunostaining). Optimal brightness setting for each channel was configured to yield maximal intensity without saturation and exclusion of non-specific emission. Images from all planes of cells were acquired and assembled z-stacks of the planes are presented.

For double-stranded DNA (dsDNA) quantification, neutrophils (4×10⁶) were resuspended in 1 ml of colorless RPMI-1640 media. The neutrophils were unstimulated or stimulated with P. stomatis at multiplicity of infection (MOI) of 50. Aliquots of 250 µl of cells (1×10⁶) were plated in triplicate in 96 well plate followed by centrifugation at 600xg at 14°C for 4 mins. The plate was then incubated for 3 h at 37°C, 5% CO₂. After incubation the plate was centrifuged at 800 g at 4°C for 2 min and supernatants were transferred to tubes and stored at 4°C until use. The supernatants were quantified using Quant-iT PicoGreen dsDNA assay kit (Invitrogen, USA). The samples were incubated with equal volume of 1X PicoGreen 5 min at room temperature and protected from light. The plates were excited at 485 nm, and the fluorescence emission intensity was measured at 520 nm using a spectrofluorometer (SpectraMax M2; Molecular Devices).

T cells were isolated from healthy human PBMCs, Fc blocked, and surface stained with viability dye eFluor 780 (eBioscience) and anti-human CD3-PerCp-Cy5.5mAb (Clone HIT3a, Biolegend) as mentioned earlier. The cells were then washed with 5 ml of sterile MACs running buffer (Miltenyi Biotech), filtered, and resuspended in appropriate volume of MACs buffer for acquisition by BD FACsAria III sorter. The T cells were gated as viable and CD3+ cells. The isolated T cells were washed (500 g for 5 minutes using a room temperature centrifuge) with pre-warmed 0.1% Bovine Serum Albumin (BSA, Equitech-Bio, Inc). The pellet was then suspended in appropriate volume of pre-warmed 0.1% BSA (1-10 x 10⁶ cells/ml) and CFSE (1µM/ml) (Cell Trace™ CFSE Cell Proliferation Kit, eBioscience) was added to the cell suspension, gently mixed and incubated at 37°C for 10 minutes with gentle mixing 2-3 times during the incubation. Upon the completion of incubation, the CFSE labeling was stopped by addition of 5 ml of ice-cold cRPMI and incubation for 5 minutes on ice and centrifugation at 500 g for 5 minutes at 4°C. The CFSE-labelled T cells were then suspended in appropriate volume of cRPMI and added to U-bottomed 96 well plates (100 µl) pre-coated with anti-human CD3 antibody (Clone OKT3,1µg/ml in PBS at 4°C overnight followed by gentle washing with sterile PBS) to which 100 µl of neutrophils (a ratio of 1:1) was added, gently mixed and incubated at 37°C with 5% CO₂ for 4 days. After 4 days, the culture supernatants were collected and stored at -20°C until analyzed. The cells were harvested, washed, Fc-blocked and stained for surface markers. The cells were stained for viability dye eFluor 780 (eBioscience), anti-human CD4-PE-Dazzle™ 594 (Clone A161A1, Biolegend) and anti-human CD8-APC (Clone SK1, Biolegend)mAb and acquired using FACsCanto for analysis of the T cell suppression. The supernatants were used to perform ELISA using a human IFN-γ ELISA Kit (Biolegend). ELISA was performed as per the instructions provided in the Kit protocol.

Results are shown as mean and standard error of the mean (SEM), unless otherwise stated. Analysis of groups was by analysis of variance using SPSS V27 (IBM, Chicago). The dependent variables were phagocytosis, H₂O₂, CD35, CD11b, CD66b, or CD63 and the independent factors were basal and stimulated. In a separate analysis we performed a repeated measures analysis on basal and P. stomatis stimulated NET formation for the groups HD NDN, COVID LDN, and COVID NDN. A post hoc analysis was performed using a one way analysis of variance comparing NET formation between groups. Difference was determined at p=0.05 using Student-Newman-Keuls. Comparison of 2 data sets was performed by Student t-test using GraphPad Prism v7.0 (La Jolla, California) with significance defined as p < 0.05.