Participants’ characteristics
A subset of plasma collected at National Centre for Infectious Diseases, Singapore, from a larger study was used in this study (5). The samples were categorised into non-severe (n=15) and severe COVID-19 (required oxygen supplementation, n=5, and ICU-admitted, n=5). The baseline characteristics of subjects at enrolment are presented in Table 1. Briefly, acute samples were collected on median day 6 (non-severe) and median day 11 (severe) post-onset of symptoms. Convalescent samples were collected between day 33-34 post-onset of symptoms. In the severe COVID-19 group, subjects were significantly older (median: 55.5 years) and a higher proportion (50.0%) were hypertensive compared to non-severe group (44.0 years; 13.3%). Additionally, neutrophil counts were significantly higher in the severe group (4.77 x 103/μL) compared to non-severe group (2.40 x 103/μL).
Plasma myeloperoxidase levels are increased in COVID-19
At acute and convalescent phases, COVID-19 positive subjects had significantly higher plasma MPO levels compared with controls [controls, median (IQR); 5.74 (5.51-11.61) ng/ml], both time points (p < 0.01). Additionally, plasma MPO levels were also significantly increased in severe [acute: 22.04 (17.54-42.37) ng/ml; convalescent: 32.50 (19.70-38.98) ng/ml] compared with non-severe cases [acute: 10.93 (9.32-14.80) ng/ml; convalescent: 12.17 (9.25-18.58) ng/ml], at both time points (p < 0.01). Despite clinical recovery, a trend of increasing MPO concentrations between acute and convalescent was observed in severe cases; however, no statistical significance was reached, Figure 1A. Lastly, we observed MPO levels to be significantly correlated to neutrophil count (R = 0.54, p < 0.01), Supplementary Table 1.
Plasma MPO activities are elevated in COVID-19
We then evaluated MPO activities. At acute and convalescent phases, plasma MPO activities were significantly higher in COVID-19 subjects compared with controls [controls, median (IQR); 36.99 (34.80 - 39.01) unit/ml, acute (p <0.01); convalescent (p<0.001)]. Despite clinical recovery, a trend of increasing MPO activities in both non-severe [acute: 51.01 (42.68 – 68.12) unit/ml; convalescent: 66.77 (63.33 – 75.82) unit/ml, p < 0.01] and severe [acute: 60.16 (46.03 – 84.89) unit/ml; convalescent: 60.16 (46.03 – 84.89) unit/ml, p < 0.01] groups were observed, Figure 1B. Additionally, quantitative MPO levels correlated positively with MPO activities, acute (R = 0.42, p < 0.05) convalescent (R = 0.45, p < 0.01).
Inhibiting myeloperoxidase activities with inhibitors
Because of higher MPO activities observed in convalescent samples, we then sought to explore on the effectiveness of MPO-IN-28 and AZD-5904 in supressing MPO activities. We randomly selected 13 samples consisting of five non-severe, five severe and three controls, and treated samples with 10μM of inhibitors (concentration selected based on previous study [(22)]. Samples treated with MPO-IN-28 [mean (SD): 35.92 (14.48) unit/ml] or AZD-5904 [29.82 (7.80) unit/ml) demonstrated an approximate 51-59% decrease in MPO activities compared to non-treated samples [68.75 (22.54) unit/ml], both p <0.0001, Figure 2. These samples were then used to explore on the association on MPO activities inhibition and EG degradation, described later.
Increased soluble endothelial glycocalyx breakdown in COVID-19
At acute and convalescent phases, COVID-19 positive subjects had significantly higher syndecan-1 [acute (p < 0.01); convalescent (p < 0.05)] and glypican-1 [acute and convalescent both p < 0.01)] breakdown compared with controls [both proteins below limit of detection]. At acute phase, comparing between severe and non-severe groups, syndecan-1 levels [median (IQR), severe: 5.22 (1.97-5.57) ng/ml; non-severe: 1.64 (0.00-2.66) ng/ml, p < 0.05] but not glypican-1 levels [severe:2.81 (2.36-5.34) ng/ml; non-severe: 2.05 (1.10-3.06) ng/ml, p = 0.13] were significantly raised in severe COVID-19 group. At convalescent, no significant differences in soluble EG proteins were observed between both groups. Additionally, syndecan-1 levels at convalescent were significantly lower compared to enrolment samples in the severe and non-severe groups, p < 0.05 and p < 0.01, respectively, Figure 1C and 1D.
Correlates of glycocalyx shedding in COVID-19
The impact of neutrophil granular proteins on EG shedding in COVID-19 is unclear. We therefore investigated the relationships between MPO and soluble EG shedding. We observed positive correlations between MPO levels and syndencan-1 shedding at both acute (R = 0.42, p = 0.03) and convalescent (R =0.39, p = 0.05) phases. Additionally, MPO activities were positively associated with syndecan-1 shedding, acute (R = 0.48, p = 0.003) and convalescent (R = 0.51, p = 0.001). Similar observations were observed for MPO levels [acute: R =0.39, p = 0.02; convalescent: R = 0.32, p = 0.06; and MPO activities [acute: R = 0.39, p = 0.02; convalescent: R = 0.35, p = 0.05] with glypican-1 shedding.
We then extended the investigation to other previously measured inflammatory mediators on their impact on EG shedding (5). At the acute phase, CRP and IP-10 levels correlated positively to syndencan-1 levels, whereas only IP-10 correlated with glypican-1 shedding. In contrast, IL-10 levels had a negative relationship with syndecan-1 levels. At the convalescent phase, IL-6 levels were positively associated with syndecan-1 shedding and IL-2 levels were negatively associated with glypican-1 shedding, Supplementary Table 2.
Assessing myeloperoxidase activities inhibition on soluble EG degradation
The strong relationship between MPO and soluble EG shedding led us to explore whether inhibiting MPO activity reduces soluble EG shedding, in vitro. HAEC treated with convalescent plasma demonstrated increased syndecan-1 shedding [mean (SD), non-severe: 1.23 (0.28) ng/ml; severe: 3.66 (1.73) ng/ml; and controls: 0.80 (0.07) ng/ml, p < 0.01]. When convalescent plasma was incubated with HAEC in the presence of MPO-IN-28, we observed reduced sydencan-1 shedding (non-severe: 0.73 (0.51) ng/ml, p = 0.05; severe: 3.13 (1.82) ng/ml, p = 0.006; controls: 0.48 (0.17) ng/ml, p = 0.06) compared with untreated convalescent plasma. Similar but more profound reduction in syndecan-1 shedding was observed in convalescent plasma treated with AZD5904 (non-severe: 0.36 (0.26) ng/ml, p = 0.001; severe: 1.48 (0.64) ng/ml, p = 0.02; controls: 0.73 (0.11) ng/ml, p = 0.06) compared with untreated plasma. Note, only four samples from non-severe group were available for AZD5904 treatment [mean, n=4, syndecan-1: 1.23 (0.31) ng/ml]. Interestingly, MPO-H2O2 catalysation or inhibiting MPO-H2O2 catalysation did not demonstrate significant syndecan-1 shedding [MPO-H2O2 + IN-28; 0.15 (0.09) ng/ml or AZ5904; 0.18 (0.01) ng/ml; MPO-H2O2; 0.19 (0.004) ng/ml] compared with HAEC supernatant [0.19 (0.13) ng/ml], Figure 3A.
On the other hand, when HAECs were incubated with untreated plasma, higher glypican-1 shedding was observed in the non-severe group [5.93 (0.92) ng/ml] but not in the severe group [4.47 (0.74) ng/ml] compared to the control group [3.95 (0.05) ng/ml], p = 0.01 and p = 0.28, respectively. In contrast to observations with syndecan-1 shedding, inhibiting MPO-H2O2 with MPO-IN-28 [5.17 (0.26) ng/ml, p = 0.01] appeared to enhanced glypican-1 shedding in the control group. MPO-H2O2 catalysation may induce higher glypican-1 shedding [1.53 (0.47) ng/ml] compared to supernatant control [0.54 (0.18) ng/ml] p = 0.01, Figure 3B. Of note, because inhibiting MPO activity with MPO-IN-28 did not demonstrate reduce glypican-1 shedding, we did not compare MPO activity inhibition with AZD5904 on glypican-1 shedding.