A Non-Invasive Neonatal Signature Predicts Later Development of Atopic Diseases

Abstract Background: Preterm birth is a major cause of morbidity and mortality in infants and children. Non-invasive methods for screening the neonatal immune status are lacking. Archaea, a prokaryotic life domain, comprise methanogenic species that are part of the neonatal human microbiota and contribute to early immune imprinting. However, they have not yet been characterized in preterm neonates. Objective: To characterize the gut immunological and methanogenic Archaeal (MA) signature in preterm neonates, using the presence or absence of atopic conditions at the age of one year as a clinical endpoint. Methods: Meconium and stool were collected from preterm neonates and used to develop a standardized stool preparation method for the assessment of mediators and cytokines and characterize the qPCR kinetics of gut MA. Analysis addressed the relationship between immunological biomarkers, Archaea abundance, and atopic disease at age one. Results: Immunoglobulin E, tryptase, calprotectin, EDN, cytokines, and MA were detectable in the meconium and later samples. Atopic conditions at age of one year were positively associated with neonatal EDN, IL-1β, IL-10, IL-6, and MA abundance. The latter was negatively associated with neonatal EDN, IL-1β, and IL-6. Conclusions: We report a non-invasive method for establishing a gut immunological and Archaeal signature in preterm neonates, predictive of atopic diseases at the age of one year.


Introduction
Preterm birth, defined as delivery at fewer than 37 completed weeks of gestation, is the leading cause of neonatal mortality and morbidity and has long-term adverse health consequences [1]. The global incidence of preterm births was estimated at 10.6 in 2014, and 9.8 in 2000 [2]. The etiology of preterm birth is multifactorial and not yet fully understood [3]. However, factors related to preterm birth include maternal or fetal medical conditions, genetic and epigenetic [4] influences, environmental exposures, infertility treatments, behavioral and socioeconomic elements [3,5]. Preterm infants experience abnormal immune and metabolic programming, which might exert a lasting influence on the risk of future disease [6,7]. Preterm-born children have been shown to have immune mediator dysregulation [8], impaired innate immunity and adaptive responses characterized by reduced levels of immunoglobulin (Ig) G, opsonization and phagocytosis, and increased activation of Th1 cells compared to that of Th2 cells [9]. Cohort studies show that preterm-born children are at increased risk for preschool wheezing and school-age asthma [10], but not for food allergy [11,12] or atopic dermatitis (AD) [11,13].
During fetal life, maternal microbiota produces compounds that are transferred to the fetus and enhance the generation of innate immune cells [14]. This process is halted prematurely in preterm infants, leaving them vulnerable to disease [9]. Preterm infants have an inflammatory and hypoxic state, which has a negative impact on lung maturation, the risk of respiratory infections, and susceptibility to subsequent exposures [9,14,15]. As early as the neonatal period, the gut microbiota imprints a persistent effect on the immune system through multiple mechanisms, including the modulation of epithelial functions, the production of cytokines, and the recruitment and training of immune cells [16][17][18].
Archaea, considered a separate domain of life from Eukarya, giant bacteria and viruses, are part of the human microbiota [19][20][21]. Gut methanogenic Archaea consume hydrogen produced by bacterial fermentation, releasing methane and short chain fatty acids (SCFA) and thus taking part in the energy supply to the host [19]. They interact with the host immune system, triggering innate and adaptive immune responses, generation of specific T and B cells, and hypersensitivity responses in animals and humans [22][23][24]. We have shown that the neonatal gut is colonized by methanogenic Archaea from the first postnatal hours, possibly starting in utero [21,22]. Gut microbiome establishment is altered in preterm and low-birth-weight infants [25,26].
Clinical investigations and research studies in neonates, including those born before term, are usually performed with peripheral blood. However, the search for non-invasive alternatives has gained momentum in recent years [27,28].
We hypothesize the existence of an association between intestinal methanogenic Archaea and the intestinal immunological signature, understood as a pattern of immune biomarkers including cell-specific products, e.g., mast cell tryptase and eosinophil-derived neurotoxin (EDN) and major pro-and anti-inflammatory cytokines. The validation of this hypothesis would open the prospect of a predictive score for the later occurrence of immune disorders, including atopic diseases.
We addressed this question through the development of a non-invasive standardized method for the assessment of the neonatal gut immune and microbial status, implemented in a cohort of preterm infants. The aims of the present study were: (1) establish a noninvasive method adequate for the investigation of preterm neonates, (2) characterize the gut immune and Archaeal components longitudinally from birth to six weeks in the study cohort, and (3) correlate the results of gut immune and Archaeal investigations at birth and up to six weeks to the later occurrence of allergic or atopic conditions.

Patients and Sampling
Stool samples from 43 preterm neonates were collected without the use of preservatives at the Nimes and Montpellier University Hospitals and stored at −80 • C. Samples were collected as meconium (n = 33) and later stool samples at 2 (n = 33), 4 (n = 29), and 6 (n = 24) weeks.

A-Immunological Analysis Preparation of Fecal Samples
One gram of feces was solubilized in two milliliters of an in-house extraction buffer consisting of phosphate buffered saline supplemented with 4 mM 4-(2-aminoethyl)-benzensulphonyl fluoride, 0.26 mM bestatin, 28 µM E-64, 2 µM leupeptin and 0.6 µM aprotinin, pH 7.4, and a protease inhibitor (Sigma-Aldrich, St. Louis, MN, USA [29]. The stool-buffer mixture was incubated for 20 min at room temperature, prior to centrifugation at 2000 rpm for 15 min at 4 • C. The supernatant liquids were freeze-dried for 24 h, resolubilized in one milliliter of extraction buffer, and used for mediator and cytokine determination ( Figure 1). were collected as meconium (n = 33) and later stool samples at 2 (n = 33), 4 (n = 29), and 6 (n = 24) weeks.

A-Immunological Analysis
Preparation of Fecal Samples One gram of feces was solubilized in two milliliters of an in-house extraction buffer consisting of phosphate buffered saline supplemented with 4 mM 4-(2-aminoethyl)-benzensulphonyl fluoride, 0.26 mM bestatin, 28 µM E-64, 2 µM leupeptin and 0.6 µM aprotinin, pH 7.4, and a protease inhibitor (Sigma-Aldrich, St. Louis, MN, USA [29]. The stoolbuffer mixture was incubated for 20 min at room temperature, prior to centrifugation at 2000 rpm for 15 min at 4 °C. The supernatant liquids were freeze-dried for 24 h, resolubilized in one milliliter of extraction buffer, and used for mediator and cytokine determination ( Figure 1).

Calprotectin and Total Protein Determination
Fecal calprotectin was measured using the BIOFLASH (Werfen, Barcelona, Spain) chemo-luminescent analyzer platform according to ISO 15,189 standards. Assay sensitivity was greater than 20 µg/mL. The total protein concentration of the samples was measured by the colorimetric method (BCA Protein assay, Thermo Fisher Scientific).

Calprotectin and Total Protein Determination
Fecal calprotectin was measured using the BIOFLASH (Werfen, Barcelona, Spain) chemo-luminescent analyzer platform according to ISO 15,189 standards. Assay sensitivity was greater than 20 µg/mL. The total protein concentration of the samples was measured by the colorimetric method (BCA Protein assay, Thermo Fisher Scientific).

A-Microbiological Analysis: Methanogenic Archaea by qPCR DNA Extraction and PCR Assays
For DNA extraction, 0.2 g of each stool sample were mixed in 1.5 mL tubes with 500 µL of G2 lysis buffer from an EZ1 ® DNA Tissue Kit (QIAGEN, Hilden, Germany). Then, 0.3 g of acid-washed beads ≤ 106 µm (Sigma-Aldrich, Saint Quentin Fallavier, France) were added in each tube and shaken in a FastPrep BIO 101 device (MP Biomedicals, Illkirch, France) for 45 s for mechanical lysis before 10 min incubation at 100 • C. A 180 µL volume of the mixture was then incubated with 20 µL of proteinase K (QIAGEN, Hilden, Germany) at 56 • C overnight before a second mechanical lysis was performed. Total DNA was finally extracted with an EZ1 Advanced XL extraction kit (QIAGEN) and 50 µL eluted volume. Sterile phosphate buffered saline (PBS) (Fisher Scientific, Illkirch, France) was used as a negative control in each DNA extraction run. Extracted DNA was incorporated into real-time PCR performed using Metha_16S_2_MBF: 5 -CGAACCGGATTAGATACCCG-3 and Metha_16S _2_MBR: 5 -CCCGCCAATTCCTTTAAGTT-3 primers and the FAM_Metha_16S_2_MBP 6FAM-CCTGGGAAGTACGGTCGCAAG probe targeting the 16S DNA gene of methanogens, designed in our laboratory (Eurogentec, Angers, France) as previously described [31]. PCR amplification was done in 20 µL volume including 15 µL of mix and 5 µL of extracted DNA. Five microliters of ultra-pure water (Fisher Scientific, Illkirch, France) were used instead of DNA in the negative controls. The amplification reaction was performed in a CFX96 thermocycler (BioRad, Marnes-la-Coquette, France) incorporating a protocol with a cycle of 50 • C for 2 min, followed by 39 cycles of 95 • C for 45 s, 95 • C for 5 s and finally 60 • C for 30 s. Samples with a CT < 40 were considered positive. Gene amplification and PCR sequencing were performed as previously described [25,26,[32][33][34].

Statistical Analysis
The responses for each quantitative parameter were described using median and 25-75 percentile (interquartile range, IQR) unless otherwise stated. Analyses were performed using the Wilcoxon test when two groups were compared, and the Kruskal-Wallis test when more than two groups were compared. The association between the different biomarkers of interest were analyzed, at each sampling time, using Spearman's correlation coefficient. The association profiles between different biomarkers were also analyzed using a principal component analysis method. Statistical analyses were performed at the conventional two-tailed α level of 0.05, using R 2.13.2 statistical software (R Foundation for Statistical Computing, https://www.r-project.org (accessed on 18 March 2022), Vienna, Austria).

Demographic and Clinical Characteristics of Preterm Infants
The 43 preterm neonates included in our study had at birth an average weight of 1160.41 g (range 440-1750 g), an average gestational age of 29 weeks (range 24-32 weeks) and an average height of 37.11 cm (range 32-47 cm). Thirty-five (81%) were born by cesarean section and 8 (19%) by vaginal delivery. Only five mothers (11.6%) had received antibiotic therapy during the peripartum period. As part of the cohort follow-up, clinical evaluation (AF) was conducted at 1 year and assessed the presence or absence of health conditions, including atopic diseases. When necessary, allergy diagnosis was carried out according to current recommendations [35,36]. A total of nine children developed an atopic condition during the first year, manifested as asthma or cow's milk allergy (CMA) in eight and AD in three, with two patients presenting an association of AD, asthma, and CMA (Table 1).

Total Protein Determination
First, we measured the total protein content in all samples. The median concentration of fecal proteins was stable from birth to six weeks, ranging from 4.53 to 9.18 g/L (p = 0.10; Kruskal-Wallis) ( Table 2).

Immune Cell Markers and Cytokines
Total IgE was detectable in over 90% of the samples at all ages, in increasing amounts between birth (meconium) and six weeks (p < 0.0001; Kruskal-Wallis).
Conversely, tryptase detection increased with sampling age, reaching 58% in samples at six weeks, up from less than 15% at earlier times (p < 0.0001; Chi-square). As most values were lower than the quantification limit, quantitative comparison was not significant (p = 0.61, Kruskal-Wallis) ( Figure 2 and Table 2).

Immune Cell Markers and Cytokines
Total IgE was detectable in over 90% of the samples at all ages, in increasing amounts between birth (meconium) and six weeks (p < 0.0001; Kruskal-Wallis).
Conversely, tryptase detection increased with sampling age, reaching 58% in samples at six weeks, up from less than 15% at earlier times (p < 0.0001; Chi-square). As most values were lower than the quantification limit, quantitative comparison was not significant (p = 0.61, Kruskal-Wallis) ( Figure 2 and Table 2). Calprotectin and EDN were detected in all samples at comparable levels irrespective of age (p = 0.13 and 0.21, Kruskal-Wallis) ( Figure 2 and Table 2).
All cytokines except TNF-α were detectable in meconium and fecal samples. TGF-β and IL-6 were the most prevalent, detected in up to 90% of samples, while IL-10 was the less prevalent, found in 20% or less of the fecal samples. The frequency of detection of IL-6 and IL-1β increased with age (p < 0.0001; chi-square), although there was a sharp drop in IL-6 frequency of detection and measured levels between meconium (75%, median 11.6 pg/L) and samples at two weeks (21%, median 0.2 pg/L). Calprotectin and EDN were detected in all samples at comparable levels irrespective of age (p = 0.13 and 0.21, Kruskal-Wallis) ( Figure 2 and Table 2).
All cytokines except TNF-α were detectable in meconium and fecal samples. TGF-β and IL-6 were the most prevalent, detected in up to 90% of samples, while IL-10 was the less prevalent, found in 20% or less of the fecal samples. The frequency of detection of IL-6 and IL-1β increased with age (p < 0.0001; chi-square), although there was a sharp drop in IL-6 frequency of detection and measured levels between meconium (75%, median 11.6 pg/L) and samples at two weeks (21%, median 0.2 pg/L).
TGF-β and IL-1β median concentrations increased with age (p = 0.014 and 0.001, respectively; Kruskal-Wallis). The median level of IL-6 was the highest in meconium samples and increased again at six weeks (p < 0.0001; Kruskal-Wallis). IL-10 median concentrations did not vary with age ( Figure 2 and Table 2).
Maternal antibiotic therapy and route of delivery did not significantly affect the meconium levels of cytokines, total IgE, tryptase, calprotectin, and EDN (Supplementary  Table S1). However, analysis according to the development of atopic disease during the first year showed that meconium calprotectin levels were lower in neonates who subsequently developed asthma or CMA compared to those who did not (p = 0.02; Wilcoxon test) (Figure 2). Levels of other mediators and cytokines were not associated with the occurrence of an atopic disease (Table 3). Comparing the different biomarkers according to the presence or absence of an atopic condition, we observed a significant difference at week six for IgE between the presence and absence of cow's milk allergy and asthma. No significant difference was observed for AD (Supplementary Table S2)

Correlation between Biomarkers
As an expected control, significant correlations were found between weight and height (R = 0.90; p < 0.0001), between gestational age and height (R = 0.75; p < 0.0001), and between gestational age and weight (R = 0.75; p < 0.0001).
Total IgE and tryptase levels were strongly correlated in samples taken at any age. IL-10 and IL-6 were correlated at all ages except at two weeks (Table 4).

Meconium Samples
Tryptase levels were correlated to levels of IL-10 (R = 0.48, p = 0.001) and IL-6 (R = 0.46; p = 0.001). Tryptase levels were correlated to levels of IgE (R = 0.91, p = 0.0001), and strong correlations were observed between levels of calprotectin and IL-1β (R = 0.90; p < 0.0001). A negative correlation between total protein concentration and TGF-β (R = −0.36; p = 0.01) was observed (Table 3). CT methanogenic Archaea are expressed as median and interquartile ranges (IQR). n (%): number of samples in which methanogenic Archaea were detected (relative frequency of detection). The stool concentration factor and median and RDI were not included in our calculations, and the median and RDI were calculated by restricting the results above the lower LOQ (limit of quantitation) for each analyte. Statistical test: Kruskal-Wallis.

Frequency of Detection of Methanogenic Archaea and Relationship with the Subsequent Development of Atopic Diseases
Using real-time PCR with 16S rRNA archaeal gene PCR primers, we detected methanogenic Archaea DNA in 30/33 (90%) meconium samples, 27/33 (81%) two-weekold samples, 23/29 (79%) four-week-old samples, and 19/24 (73%) six-week-old samples, respectively. We found no significant difference in the frequency of detection nor in CTs according to age at sampling (Table 4).

Unsupervised Analysis of Immunological Markers and Methanogenic Archaea Atthe Neonatal Period, and the Subsequent Occurrence of AD, Asthma and CMA during the First Year
We performed unsupervised analysis of the immunological data, CT of Archaea, and the clinical information of the occurrence of allergic events during the first year of life. Data were analyzed for each of the four sampling times.
For meconium, calprotectin, EDN, and IL-1β levels were negatively and significantly (p < 0.001) correlated (r = −0.64) with subsequent development of AD. Calprotectin, EDN and IL-1β had the largest and most significantly (p < 0.01) correlated positive correlation coefficients, which were 0.79, 0.53, and 0.51, respectively. No correlation was observed for the Archaea CT with the other parameters (Figure 4b(A)).

Discussion
In this study, we describe a non-invasive screening method for profiling neonatal immunity and its validation in a preterm neonate cohort as a predictive tool for subsequent development of atopic diseases. The method was also applied to meconium samples, which reflect intrauterine processes and contain almost 1000 identified proteins with important functions [37]. The clinical endpoints of this study were evaluated at the age of one year, while the total duration of cohort follow-up will be three years.
Total IgE was detected in over 90% of the samples, at increasing concentrations with age. Transplacental delivery of allergens and preterm sensitization have long been recognized, possibly inducing sensitization and detectable meconial IgE [38,39]. Transplacental transport of maternal IgE able to sensitize fetal mast cells has been recently demonstrated [40], but its role in neonatal immune defenses or subsequent immune disorders is only speculative.
Addressing mast cell tryptase in meconium and later samples, we found that it was detectable only in a minority of meconium samples and during the first month of life, however, it became a common finding at the end of the neonatal period, represented by samples collected at six weeks. Fecal tryptase and IgE levels were strongly associated at each of the studied time points. These results suggest that gastrointestinal mast cells, as opposed to skin mast cells [40], are mostly recruited postnatally, and mature after birth with IgE levels exerting a positive effect. Conversely, maturity of gastrointestinal mast cell populations might be attained during late pregnancy. Tryptase is a serine protease able of autocrine activation of mast cells and induction of proinflammatory effects such as proteolytic cleavage and activation of PAR2 receptors and inactivation of VIP (Vasoactive Intestinal Peptide), associated with smooth muscle relaxation [41]. Through PAR-2 activation, luminal tryptase can contribute to the dysfunction of the gut epithelial barrier [42]. The presence of tryptase in stool samples has been associated with food allergic diseases, dietary exposure and/or mast cell stimulation or increased intestinal mast cell count [43,44]. In addition, fecal tryptase has also been shown to be associated with inflammatory bowel disease and irritable bowel syndrome [45,46].
Focusing on two secreted biomarkers of innate immune cells, neutrophil-derived calprotectin, and eosinophil-derived EDN, we found that fecal samples at all studied ages contained detectable and stable levels of both biomarkers. The levels measured in our preterm cohort were much lower than the reference values [47], but similar to those reported during the first postnatal month in another preterm cohort [48]. These results suggest that the preterm gut contains small numbers of granulocytes, or that such granulocytes are not activated. Indeed, lower neutrophils were reported in preterm infant cord blood [49]. An association between low levels of fecal calprotectin and adverse health conditions, including obesity and sepsis, by age two has been suggested [48].
Proinflammatory and anti-inflammatory cytokines were detected in the meconium and later samples of preterm infants, with the notable exception of TNF-α which was not demonstrated in any sample. Different temporal patterns were demonstrated: IL-6 levels were higher in the meconium than in later samples, while IL-10 was seldom detected and TGF-β and Il-1β displayed a progressive increase between birth and six weeks. Mostly undetectable IL-10 levels were also reported in a pilot study of fecal biomarkers in preterm infants [8]. Although we did not determine the cellular source of fecal cytokines, a shift in immune cells lining the intestine has been demonstrated for macrophages, with resident fetal macrophages being replaced after birth by bone-marrow-derived macrophages [50]. Macrophage cytokine production, most notably of proinflammatory IL-6 and IL-1β, can be persistently altered by metabolic conditions [51]. The increase in TGF-β levels from birth to six weeks might provide a counter-acting mechanism in a proinflammatory environment. Allergic events (asthma or cow's milk allergy) and atopic dermatitis were also positively correlated with EDN, IL-1β, IL-10, and IL-6 at four and six weeks. The high production of fecal EDN, IL-1β, and IL-10 during the first weeks of life may therefore be an indicator for later risk of allergic diseases.
The neonatal period is paramount for the establishment of the intestinal microbiota. Intrauterine life is associated with low levels of maternal microbial translocation [52]. However, we have recently demonstrated the presence of the viable methanogenic Archaea Methanobrevibacter smithii in the meconium, suggesting intrauterine colonization of the fetus by this microorganism [26]. Here, we provided evidence for postnatal persistence of methanogenic Archaea in fecal samples and suggest a possible role in the orientation of intestinal immunity, supported by the negative association between Archaea abundance (inversely proportional to CT values) and the concentrations of EDN, IL-1β, and IL-6. We also found that Archaea abundance at four and six weeks was positively associated with later occurrence of allergic events. Archaea have been shown to produce SCFA which induce regulatory T cell differentiation, downregulate proinflammatory cytokines, and may protect against the occurrence of atopic conditions [19,[53][54][55][56][57] However, in a cohort study, the protective effect of methanogenic Archaea was restricted to the species Methanobrevibacter stadtmanae [58]. A decrease in the load of beneficial methanogenic Archaea during the first years of life could therefore favor the occurrence of allergic events during the first years of life.
A third line of contribution to protection or increased risk of developing atopic conditions is the genetic background. As an example, a del/del genotype (−2549 −2567 del18) of Vascular Endothelial Growth Factor (VEGF) has been associated with asthma occurrence and irreversible bronchoconstriction [59].
The strengths of our study are methodological and medical: (1) miniaturization and standardization, using small quantities of stool (1 g) and small volumes of extraction buffer (2 mL). The dilution of the samples was corrected by the freeze-drying process, as the lyophilizates were contained in 1 mL of buffer. (2) prevention, thanks to the use of protease inhibitors, of the risk of potential contamination of the handler. The main weakness of this study is the lack of microbiological data outside Archaea. Further studies are warranted for longitudinal immuno-microbiological profiling of meconium and neonatal samples, in preterm and at-term infants. Its validation as a non-invasive diagnostic method will be in line with the currently unmet needs in terms of non-invasive diagnosis of allergy.

Conclusions
This study allowed us to highlight the presence of mediators in the meconium and feces of preterm infants. We provide proof of concept of the feasibility and value of a standardized fecal mediator assay for non-invasive profiling of neonatal immunity. Such assays can be used for early characterization of the immune status of a newborn. Technical optimization for a multiplex assay could facilitate the implementation of fecal immune profiling in clinical and research laboratories. We report evidence of a correlation between the meconial and neonatal load of methanogenic Archaea and selected fecal biomarkers, and later occurrence of atopic conditions in preterm children.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/jcm11102749/s1, Table S1: Comparison of biomarkers and CT of Archaea according to the presence or absence of maternal antibiotic therapy and according to the mode of delivery, Table S2: Comparison of median CT of Archaea in preterm infants according to the presence or absence of allergic events during the first year of life, Table S3