Citrulline

Blood samples were collected from the controls and the patients prior to any medical treatment. Blood samples (5 ml) were collected from forearm veins after overnight fasting in tubes containing ethylenediaminetetraacetic acid (EDTA; Termo, Tokyo, Japan) and were immediately placed on ice. Plasma was prepared by centrifugation at 3,000 rpm at 4°C for 15 min and then stored at −80°C until analysis. After the plasma collection, all samples were stored and processed at the Institute for Innovation of the Ajinomoto Co., Inc. (Kawasaki, Japan). To reduce any bias introduced prior to analysis, samples were analyzed in random order. The plasma samples were deproteinized using acetonitrile at a final concentration of 80% before measurement. The amino-acid concentrations in the plasma were measured by HPLC–ESI–MS, followed by precolumn derivatization. The analytical methods used were as described previously [34] , [35] (link), [36] (link).
Among the 20 genetically-encoded amino acids, glutamate (Glu), aspartate (Asp), and cysteine (Cys) were excluded from the analysis because they are unstable in blood. Citrulline (Cit) and ornithine (Orn) were measured instead because they are relatively abundant in blood and are known to play important roles in metabolism. The following 19 amino acids and related molecules were therefore measured and analyzed: alanine (Ala), arginine (Arg), asparagine (Asn), Cit, glutamine (Gln), glycine (Gly), histidine (His), isoleucine (Ile), leucine (Leu), lysine (Lys), methionine (Met), Orn, phenylalanine (Phe), proline (Pro), serine (Ser), threonine (Thr), tryptophan (Trp), tyrosine (Tyr), and valine (Val).
Two metrics were made for each of the 19 amino acids including the absolute concentration of each amino acid, which directly reflected its availability and consumption, and the ratios associated with the specific metabolic status in each organ. The concentrations of the amino acids in the plasma were expressed in µM, and the ratios of the amino acid concentrations were expressed by the follow equation: where X2_i,j is ratio of the amino-acid concentration of the j-th amino acid of i-th subject, and X_i,j is the plasma concentration (µM) of the j-th amino acid of i-th subject.

Detailed methods for anthropometric measurements, assessment of blood, stool, and urine biomarkers and statistical methods are provided in S1 File. As shown in Table 1, 375 children were enrolled and had initial anthropometry and were invited to provide a fecal specimen, have a L/M absorption test and to have blood obtained for testing potential biomarkers of intestinal barrier function, intestinal and systemic inflammation and injury repair. Shown in Table 2 are the 13 plasma, 4 fecal and urinary tests with at least 274 valid results obtained from within 1 month of enrollment. Markers of barrier function included urinary L/M absorption, fecal A1AT, Reg-1, and plasma LPS (acute levels, by neutralization luminescence [LUM] assay), IgG and IgA anti-LPS anti-FliC, zonulin, I-FABP and, with limited amounts of plasma available, claudin-15. Fecal MPO and neopterin were tested to assess intestinal inflammation, and plasma SAA, sCD14, LBP, citrulline, tryptophan and kynurenine were measured to assess systemic inflammatory responses and as potential predictors of intestinal injury repair. These 18 markers of intestinal barrier disruption and inflammation had adequate samples for assay in at least 274 (to 321) children at enrollment for study of associations with enrollment stunting (or wasting). Of several fecal biomarkers that have been used to assess intestinal inflammation, we selected myeloperoxidase (MPO) as our main test fecal marker of acute neutrophilic intestinal inflammation because of its ready availability, relatively less influence by age or breast-feeding and its potential use in murine models of enteropathy. In a separate analysis comparing fecal MPO, lactoferrin, calprotectin and lipocalin-2, we found that they correlate well with each other [33 (link)].
In addition, followup anthropometry at 2–6 months after initial sampling enabled us to assess these biomarkers as predictors of subsequent growth. We also assessed their associations with each other as well as with separately studied comparisons of fecal lactoferrin, calprotectin and lipocalin, and plasma hsCRP, and claudin-15. Repeated Measures MANOVA analyses were used to model mean growth from Study Start to follow-up anthropometric measurement obtained within 2 to 6 months later by children with low and high levels of each biomarker while controlling for child age and gender. Interaction effects of gender with biomarker were also assessed for each model. Similarly interaction effects of study start growth status (stunting present or not) of biomarkers on growth were assessed, although these results were not significant.
Missing data analysis (Pearson correlations between known characteristics of participants retained or failed to return for followup or were missing biomarker results because of sample limitations) showed that children with caregivers having fewer years of education were missing more of three biomarkers (A1AT, Reg1, MPO, r ≤ -0.18) compared to their more educated counterparts. Boys were disproportionately missing A1AT results (r = 0.15 p = 0.003), older children were more likely to be missing Neopterin (r = 0.12 p = 0.017) while nonbreastfeeding (r = -0.10, p = 0.37) and more wasted children (r = -0.12, p = 0.026) are missing more L/Ms. To account for bias in followup, multiple regression was used to impute missing data for all biomarkers with at least 274 (73% of 375) complete data. Principle Components Analyses (PCA) were then used to combine related combinations of 11 barrier function, 2 intestinal inflammation and then 7 systemic biomarkers. These related sets of biomarkers were then correlated among each other as well as to children’s growth status and growth trajectory initially using Partial Pearson correlations and ultimately using Multiple Regression. All analyses predicting anthropometric status and growth trajectories included controlling for child age and gender.

For each infectious disease, a two-stage CITS analysis was applied to quantify the impacts of the NPIs on infectious diseases during the COVID-19 pandemic in 2020 [15 (link)]. This design is developed from the basic interrupted time-series (ITS) design that involves a before-after comparison in a single population exposed to the intervention between the observed change in the outcome of interest and the best approximation of the true counterfactual [16 (link), 17 (link)]. To provide stronger evidence to support a causal effect of the intervention on the outcome of interest, CITS design includes a control series to exclude problems due to co-interventions or other events occurring around the time of intervention [16 (link)]. In this study, we compared the infectious disease incidence before the NPIs with the post-NPIs period. To build a self-controlling design, we selected the first month (January) of each year as the control series as NPIs were not introduced to the study population until the end of January, 2020 (Fig. 1).
Fig. 1

Timelines of COVID-19, non-pharmaceutical interventions during the study period, and the selection of control series

In the first stage, a quasi-Poisson time-series regression was applied in each PLAD [18 (link), 19 (link)]. The PLAD-specific equation was as follows: $$Y_{\mathit{it}}\sim Poisson\left(\mu ,;,\theta \right)$$ $$\text{E}\left(Y_{\mathit{it}}\right)={\text{exp}(\alpha }_i+{\beta }_{i}NPIs+{{ϵ}_{i}Group+\gamma }_i{\mathit{Trend}}_{\mathit{secular}}{+\delta }_i{\mathit{Trend}}_{\mathit{season}}+ns\left({\mathit{Temp}}_{\mathit{it}}\right)+ns\left({\mathit{RH}}_{\mathit{it}}\right)+ns\left({\mathit{Prep}}_{\mathit{it}}\right)+\text{log}\left(P,O,P\right))$$ $$VAR\left(Y_{\mathit{it}}\right)=\theta \mu$$ where $Y_{\mathit{it}}$ denotes cases in PLAD $i$ on month $t$ ; ${\alpha }_i$ is the intercept in PLAD $i$ ; ${\beta }_i$ , ${\gamma }_i$ , and ${\delta }_i$ represent the coefficients in PLAD $i$ . $\mathit{NPIs}$ is a binary variable to indicate the introduction of NPIs (coded 1) and without the NPIs (coded 0) in the model [20 ]. $\mathit{Group}$ is a binary variable representing the intervention group (Group = 1, February to December) or control group (Group = 0, January). ${\mathit{Trend}}_{\mathit{secular}}$ denotes a continuous term for time (months since the start of the study) to model the secular trend. ${\mathit{Trend}}_{\mathit{season}}$ stands for an indicator for the month of the year to control for seasonality. $ns\left({\mathit{Temp}}_{\mathit{it}}\right)$ , $ns\left({\mathit{RH}}_{\mathit{it}}\right)$ , and $ns\left({\mathit{Prep}}_{\mathit{it}}\right)$ are mean temperature, relative humidity, and precipitation controlled in the model using natural spline functions with 3 degrees of freedom (df s). Population was also included in the model as an offset term, $\text{log}\left(P,O,P\right)$ . $VAR\left(Y_{\mathit{it}}\right)$ and $\mu$ denote the variance and expectation of $Y_{\mathit{it}}$ , and $\theta$ is an over-dispersion parameter. As we only had the 1-year (year 2020) data after the intervention which couldn’t generate the slope of the post-intervention trend, we assumed a constant slope before and after the intervention. The effect of NPIs was expressed as the intercept change ( ${\beta }_i$ ). As the incidence of infectious diseases had seasonal trends that varied by year, we kept the month constant in different years (using the sixth month for each year) when estimating the secular trend and slope change before and after the intervention. The counterfactual incidence in 2020 was predicted using the same model while we change the intervention term in the model (NPIs) from 1 to 0, assuming that there is no intervention in 2020.
In the second stage, we pooled the PLAD-specific estimates using a random-effect meta-analysis with restricted maximum likelihood estimation [21 (link), 22 (link)]. The impacts of NPIs on infectious diseases’ incidence were expressed as the incidence rate ratio (IRR, calculated as the exponents of coefficients) and 95% confidence intervals (95% CI).

For identification of putative deiminated/citrullinated proteins in crab haemolymph, enrichment was carried out using the F95 pan-citrulline antibody (MABN328, Merck) in conjunction with the Catch-and-Release Immunoprecipitation Kit (Merck). Immunoprecipitation was carried out on mini-agarose columns together with the F95 antibody and the affinity ligand, overnight at 4°C on a rotating platform and proteins thereafter eluted according to the manufacturer’s instructions (Merck, UK). F95-enriched proteins from the parasitized and control haemolymph were then subjected either to SDS-PAGE and silver staining, or to liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis for identification of protein hits.

Thermal shift assays were performed using Bio-Rad CFX Connect Detection System. For these assays 10 μM of RESC5 WT and mutant proteins were analyzed in the absence and presence of varying concentrations of L-citruline (AdooQ Bioscience) or dimethylamino arginine (AdooQ Bioscience). The RESC5 proteins and all other reagents used in this experiment were diluted in filter sterilized low salt buffer (25 mM Tris pH 7.5, 150 mM NaCl, 5% (v/v) glycerol, and 1mM β-ME). Melting curves were collected for 10 μM of RESC5 (WT or mutant), in the presence of 0 μM, 100 μM, 200 μM, 500 μM and 1 mM of L-citrulline or dimethylamino arginine. The appropriate volumes of RESC5 protein, L-citrulline, or dimethylamino arginine for each condition described above were added to wells of a 96 well Bio-Rad Hard-Shell Plate (thin walls) containing a final volume of 20 μL and 1X of GloMelt (GloMelt^TM Thermal Shift Protein Stability Kit from Biotium-Cat No. 33021–1). The plates were briefly spun before to transfer to the Bio-Rad CFX Connect Detection System. Fluorescence was detected over a temperature range of 25–100°C with 0.5°C steps and a time hold of 1 min for each temperature step.