Assay development and validation.
Antibody screening. We first screened various antibodies against oligodendrocytes, and found that anti-myelin oligodendrocyte glycoprotein (MOG) as the best capture agent. MOG is a transmembrane protein with a large extracellular domain at N-terminal. Our antibody recognizes the extracellular domain of MOG, so that antibody can bind to ODE under physiological condition without permeabilization or lysis procedure.
Capture antibody specificity (Fig. 2AB). Control human plasma or buffer alone was applied to enzyme-linked immunosorbent assay (ELISA) wells, where various concentrations of both control IgG (rabbit IgG) or anti-MOG IgG (rabbit IgG) were immobilized, followed by the probe reaction with anti- clusters of differentiation 9 (CD9). ELISA readings of relative light units (RLU) of anti-CD9 probes in the 3 controls (buffer and plasma in control IgG wells (Fig. 2A), and buffer alone in anti-MOG wells (Fig. 2B, open triangle), showed very low values compared with large increases in RLU in plasma in MOG wells (Fig. 2B, closed circle).
Plasma volume dilution (Fig. 2C-D). Three different plasma samples (0, 2.5, 5, and 10 mL) were suspended in a final volume of 40 mL, and applied to MOG wells, followed by the probe reaction with anti-CD9 (Fig. 2C) or tropomyosin receptor kinase B (TrkB) (Fig. 2D), respectively. TrkB is a receptor of brain-derived neurotropic factor and is known to be expressed on the surface of ODC. Both signals showed linear dose dependent curves.
Confirmation of EV (Fig. 2E). The signals of anti-CD9 on anti-MOG wells [CD9 on MOG] was reversible by exposing to pH 2 solution (data not shown). Since the amounts of captured ODE in ELISA well were very small, we immobilized anti-MOG to magnetic beads, which showed the same surface characteristics as ELISA wells. After elution of ODE, pH was neutralized then applied to nanoparticle tracking analysis to analyze size distribution. As shown in Fig. 2E, we confirmed the presence of 100-200 nm EV sized particles, as well as much larger large 300-500 nm particles, which indicate fused EV or EV aggregates.
ODC specificity (Fig.2FG). TrkB is an excellent tool to demonstrate the ODC specificity on anti-MOG wells. Anti-CD81 (mouse IgG) and control mouse IgG (mIgG) were immobilized to ELISA wells to capture whole EV. In the separate wells, anti-MOG (rabbit IgG) and control rabbit IgG (rIgG) were also immobilized to capture ODE, respectively. After plasma samples and buffer alone were applied, ELISA wells were exposed to anti-CD9 probes for the quantification of captured whole EV (Fig. 2F) or anti-TrkB probes for the quantification of captured ODE (Fig. 2G). Both anti-CD81 and anti-MOG captured EV (Fig. 2F), but the amounts of captured ODE on anti-MOG wells were 42% of those of anti-CD81. However, the amounts of TrkB on anti-MOG wells were 1,162% of those of anti-CD81 (Fig. 2G), indicating over a 2,000% (20 folds) enrichment of ODE.
Probe specificity (Fig. 2H-J). We then screened various cytokine biomarkers on anti-MOG wells and found that interleukin 1B (IL1B) was the best marker for this study (data not shown). In our probe solution, a huge excess volume of non-biotinylated control IgG was included. Thus, the detection of IL1B was anti-IL1B-specific. In order to further validate IL1B specificity, recombinant IL1B (rIL1B) was added to the probe solution to block anti-IL1B binding. As shown in Fig. 2H-J, anti-IL1B reaction was dose dependent in 3 different plasma (open triangle). Similarly, rIL1B decreased signals to half the amounts of the probes (Fig. 2H-J, closed circle). Thus, IL1B signals were IL1B-specific.
Test of human samples.
Preliminary studies. The target biomarker in this study was anti-IL1B signals on anti-MOG-immobilized ELISA wells [IL1B on MOG], a marker of neuroinflammation on ODE. We first tested intra-assay reproducibility, and found that the coefficient of variation (CV) was <20% (Data not shown). This is a huge benefit of this test, because we can save precious clinical samples for various analysis by running singlicate. Also, we found that serum from capillary blood is acceptable (data not shown). Then, we tested control and athlete samples using only 5 mL samples in singlicate.
Control values (Supplemental Fig. 1). ELISA readings of RLU were converted to units/ml by using the dilution curve of our standard plasma, arbitrarily assigned as 100 units/mL. When we tested 63 control plasma samples, values were widely distributed over 2 logs (Supplemental Fig. 1), similar to our previous studies on neuron-derived EV (10). Thus, as shown in the next section, we focused on the longitudinal studies.
Controls (Fig. 3A-D). Control plasma samples (venous blood) of 6 adults were collected every week for 3 times. As shown in Fig. 3A, difference in [IL1B on MOG] from the first blood collection (%Control) was less than + 50%. The second control was cross-country (non-contact sport) athletes from 3 different high schools (Fig. 3B-D) (n=8, 32, and 31, respectively), where no head insult was reported. Plasma samples (venous blood) were collected once every month from July to November (Fig. 3B), as well as 2 time points (July and November) of serum collection in the other 2 schools (Fig. 3C-D) from capillary blood. As shown in Fig. 3A-D, %Control of [IL1B on MOG] was all less than 200%.
A case study of post-concussion syndrome (PCS) (Fig. 3E). This was a case of severe concussion of professional ice hockey player (adult male), reported previously (11). [IL1B on MOG] did not increase over the first 9 days after concussion, but substantially increased, then moved to undetectable levels after 36 days. He sustained a mild, glancing blow to the right temporal area of the head during a slow pace practice drill. He had no loss of consciousness and mild symptoms initially. Over the next few days, he began to develop a significant headache, balance issues during activities of daily living, and an aversion to loud, sudden noises. His concussion symptoms (headache and sensitivity to noise) persisted for more than a month and resolved at day 36 after concussion. MRI at day 36 showed no abnormality. As shown in Fig. 3E, [IL1B on MOG] changes were well correlated with clinical condition.
High school football (Fig. 3F-K) (all male). In the 1st high school football team, (Fig. 3F), plasma samples were collected by venipuncture, at each game as well as before (July) and after the season (November). In the subsequent studies, serum was collected from 5 different high schools by capillary blood once every month from July to November (Fig. 3G-K). Not all athletes showed any signs of concussion or traumatic brain injuries (TBI). As shown in Fig. 3F-K, [IL1B on MOG] stayed <200% in a majority of athletes, however, 14% showed >200% increase. All schools had at least one >200% subjects, although the incidence was varied from 8.3% (Fig. 3K) to 23.8% (Fig. 3G). More interestingly, 4 subjects (Fig. 3*) showed the increase in the 1st month and gradually returned to the baseline, whereas 12 subjects showed gradual increase toward the end of the season (Fig. 3·).
College rugby (Fig. 4AB) (all male). Plasma samples were collected by venipuncture, 3-4 times/year from 2014 to 2017 (n=18). These were the same subjects as described in our previous report (11). Since one subject showed a >16 fold (1600%) increase from the baseline (Fig. 4A, *2), the Y axis was changed from 0 to 600% in Fig. 4B. In Fig. 4A, subject #1 (*1) did not increase in 2014, but gradually increased in 2015 season. Subject #2 (*2) was striking, and gradually increased during 2015 season, then showed a much higher increase in the 2016 season. Since this person graduated in 2017, we could not follow up on his current health condition. In Fig. 4B, one person showed periodical increase in both 2016 and 2017. These 3 subjects did not show any concussion or TBI. One person showed concussion 4 times during this study period, and this person also showed >200% increase (Fig. 4A, B).