Amniotic fluid glycoproteins as potential ligands for macrophage galactose-type C-type lectin and their possible implications for immunoregulation during pregnancy

Abstract

Immunity during pregnancy must be precisely balanced, because the mother’s body needs to effectively fight pathogens while maintaining tolerance of the semi-allogeneic fetus. Factors that participate in achieving the proper immune balance during pregnancy include glycoproteins that expose sugar residues recognized by specific lectin receptors. Several types of lectins and their ligands have been detected at the maternal-fetal interface, and changes in their expression levels have been correlated with pregnancy complications. Although the presence of potential sugar ligands for human macrophage galactose-type lectin (MGL), including LacdiNAc and (sialo)Tn antigen, has been detected in amniotic fluid, placenta and fetal tissues, the interaction between them and its possible role have not been elucidated so far. The aims of the present study were to evaluate reactivity of MGL with amniotic fluid proteins and to identify and characterize the amniotic ligands of MGL. The analysis proved the ability of MGL to interact with amniotic glycoproteins and revealed the potential protein carriers of glycans recognized by MGL, including mucins, mucin-like proteins, and uromodulin. Bioinformatics analysis assigned the identified glycoproteins as being involved in immune processes such as regulation of signaling pathways activated in the response to Toll-like receptor (TLR) agonist stimulation or modulation of antimicrobial responses. Our research suggests that MGL-mediated interactions may be involved in maintaining the delicate immune balance during pregnancy. Our results may be helpful in future implementation of modern therapies based on the glycan ligands of MGL in the context of avoiding miscarriages and preterm birth.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-16909-2.

Introduction

One of the most remarkable phenomena in nature is the protection of the semiallogeneic fetus in utero from the maternal immune system’s responses, while preserving the ability to effectively fight pathogens. Understanding the immunological interactions underlying this phenomenon has been the subject of intensive research since the middle of the last century, but to this day there are gaps in the knowledge about the immunological mechanisms accompanying pregnancy. Numerous cases of recurrent miscarriages, pre-term birth or preeclampsia, the cause of which is at least partially related to the maternal immune response, leave no doubt about the need for further research in this field¹.

Important immunoregulatory factors that actively participate in achieving the proper immune balance during pregnancy include, among others, glycoproteins that expose sugar residues recognized by specific lectin receptors^2–9. To date, there have been described several types of lectins that support immune system functions, including I-type lectins (Siglecs), galectins, and C-type lectins (CTLs). The presence of some of these lectins has been confirmed in amniotic fluid, placenta, and trophoblast cells (e.g. Siglecs 5, 6, and 14 recognizing the sialyl Tn antigen)^10–12. Changes in the expression level of lectin receptors and alteration in their glycan ligands correlate with pregnancy complications such as preeclampsia, and may translate to premature birth or even fetal death^13–16.

Recently, there has been growing interest in the macrophage galactose-type C-type lectin (MGL, also known as CLEC10A/CD301), which belongs to the CTL family¹⁷. MGL is expressed by macrophages and dendritic cells (DCs), and is the only C-type lectin receptor within the human immune system that exclusively recognizes terminal N-acetylgalactosamine (GalNAc) residues, including Tn antigen (αGalNAc-Ser/Thr) and the LacdiNAc (GalNAcβ1-4GlcNAc) epitope. Moreover, MGL has also been found to bind sialylated forms of Tn antigen (Neu5Acα2-6GalNAcα-Ser/Thr)^17–20. In healthy human tissues, the occurrence of such glycan structures is not common and is generally associated with pathological conditions, e.g., cancers^21–24. To date, it has been demonstrated that under physiological conditions the occurrence of endogenous Tn antigen is mainly limited to CD45 on T effector cells²⁵. However, the (sialyl) Tn epitope, known as oncofetal antigen, has also been detected in fetal tissues and amniotic fluid^10,26–29. Additionally, the terminal GalNAc residue in humans has been found to be present in the form of the unusual LacdiNAc epitope, e.g. on glycodelin A—the immunoregulatory glycoprotein of amniotic fluid^2,30,31.

Interaction of MGL with its ligands on CD45 reduces T-cell receptor (TCR) signaling pathways, leading to inhibition of T-cell proliferation, reduced synthesis of pro-inflammatory cytokines and therefore accelerated T-cell apoptosis. On the other hand, in MGL-expressing DCs, ligand binding augments signal transduction pathways, which increases secretion of Toll-like receptor (TLR)-induced IL-10^20,32,33. Interleukin-10 shows immunosuppressive properties. Increased concentration of this interleukin leads to the differentiation of regulatory T-cells (Tregs), which are crucial in silencing excessive immune responses and maintaining homeostasis³⁴. Moreover, MGL is overexpressed by antigen-presenting cells (APCs) with a tolerogenic phenotype, indicating the involvement of MGL in the mechanisms of immunosuppression and the development of immune tolerance³⁵. However, a more recent study suggested that the context of MGL ligands may influence the immunological outcomes mediated by the receptor. The specific structure of MGL ligands—the density of the glycans, length, and steric arrangement—may be important for triggering particular molecular pathways associated with a distinct immunological effect: activation or immunosuppression^{17,20,22,36–38}. The role of MGL in the human body is still not thoroughly understood. Nevertheless, it is believed that its primary function is to protect the organism from excessive inflammation and autoimmune diseases.

Many of the MGL-mediated interactions described so far concern microbial pathogenesis. Some CTLs evolved to differentiate ligands as self and non-self structures³⁹. Since MGL may be expressed on APCs residing in or infiltrating different tissues⁴⁰ it is able to interact with pathogens exposing the GalNAc epitope (e.g. via bacterial lipopolysaccharides, viral envelopes, or helminth shells) within different sites of the human body and trigger appropriate immune responses. It has been reported that the protective function of MGL during infection may, in certain circumstances, be based on the downregulation of inflammation in response to invading pathogens^17,41,42. Interestingly, some pathogens appear to exploit the immunosuppressive properties of the MGL receptor to evade the host immune response^43–47.

The Tn antigen, recognized by MGL, is a structure closely related to cancer cells. Its overexpression promotes cancer cell proliferation and invasiveness and has been associated with cancer progression and poor prognosis^48–51. Thus, many studies on the MGL interactome have also focused on carcinogenesis^22,52. Indeed, several reports indicate that Tn antigens present in various types of cancer are recognized by MGL and this interaction promotes an immunosuppressive effect^17–20,53. These data indicate that MGL-mediated interactions may be part of a cancer survival strategy.

According to the human fetoembryonic defense system (Hu-FEDS) hypothesis, the mechanisms involved in achieving maternal immune tolerance are similar to those that pathogens and metastatic cancer cells use to evade host immune surveillance and are based on recognition of specific glycans by lectins. The hypothesis assumes that soluble and cell surface glycoproteins associated with fertilization and pregnancy expose unique carbohydrate structures that have not been found in other normal tissues outside of the human reproductive system, and such glycosylation patterns are acquired or mimicked by persistent pathogens and aggressive tumor cells to target specific counterreceptors^2,3,54–56. Although many pregnancy-related glycoproteins have been described as ligands of several lectins of the immune system, the MGL-mediated interactions in the context of pregnancy have not been investigated so far.

In the light of the assumptions of the Hu-FEDS hypothesis, the immunomodulatory function of MGL, and data on the presence of oncofetal (sialyl)Tn antigen and LacdiNAc epitope in amniotic fluid and fetal tissues^2,10,26–31 this study aimed to evaluate whether these glycoepitopes detected in amniotic fluid are ligands for MGL.

Methods

Clinical material

Amniotic fluid samples were obtained from women (n = 15) aged 24–37 who were hospitalized in the Second Department and Clinic of Gynecology and Obstetrics of the Wroclaw Medical University during the caesarean section procedure. The inclusion criteria included delivery of healthy newborns without malformations and chromosomal abnormalities, after 38 weeks of gestation, birth weight appropriate for gestational age, and no alcohol consumption or cigarette smoking during pregnancy. Before storage, amniotic fluid samples were centrifuged at 3 000 × g for 20 min to remove cells. The samples were pipetted into Eppendorf tubes and stored at -78 °C until assayed⁵⁷. Pooled amniotic fluid samples were prepared by mixing 100 µL of each individual sample. Protein concentration was then determined using the bicinchoninic acid (BCA) protein assay following the manufacturer’s protocol (Thermo Fisher Scientific, Waltham, MA, USA). All participants gave their informed written consent to be enrolled in the study. The study complied with guidelines of the Declaration of Helsinki. The project has been approved by Wroclaw Medical University Bioethics Council (no. of approval: KB 160/2023 N).

Western blot

Western blot was used for the preliminary assessment of the presence of MGL ligands in amniotic fluid. Pooled samples of amniotic fluid were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). SDS-PAGE was conducted according to the standard Laemmli procedure⁵⁸ in a 12.5% gel. A constant amount of protein (16 µg) was loaded onto the gel lanes. SERVA Triple Color Protein Standard III (SERVA Electrophoresis GmbH, Heidelberg, Germany) was used to determine the glycoprotein molar masses. For further testing of MGL interaction, each electrophoretic separation in the gel contained two control lines, namely 1 µg of GalNAc-BSA glycoconjugate (Abcam, Cambridge, UK) as a positive control and 1 µg of bovine serum albumin (BSA) (SERVA Electrophoresis GmbH, Heidelberg, Germany) as a negative control. After separation, the gels were divided and subjected to: Coomassie Brilliant Blue (CBB) staining (0.5% CBB in 40% methanol and 10% acetic acid)—for further analysis using mass spectrometry; silver staining⁵⁹; or transfer of separated amniotic proteins to a nitrocellulose membrane (NC) using semi-dry blotter (Hoefer Inc., Holliston, MA, USA; 120 mA, 90 min). The membranes were blocked overnight using 15 mM Tris–HCl buffer, pH 7.4, with 0.15 M NaCl (TBS) with 1% Tween-20 solution and then incubated with: (1) recombinant human MGL (CLEC10a) protein (Elabscience, Houston, Texas, USA) diluted 1:500 in 15 mM Tris–HCl buffer, pH 7.4, with 0.15 M NaCl; 0.1% Tween-20 (TBST) containing 2 mM Ca²⁺ or additionally 25 mM GalNAc; (2) biotinylated lectin from Vicia villosa (VVL, Vector Labs, Burlingame, CA, USA) diluted 1:200 in TBST; and (3) biotinylated lectin from Wisteria floribunda (WFL, Vector Labs, Burlingame, CA, USA) diluted 1:500 in TBST, for 1 h at room temperature (RT) with constant gentle shaking. To detect glycoprotein-MGL complexes, the membranes were incubated for 1 h at RT with biotinylated anti-human MGL antibody (anti-human CLEC10A/CD301; BAF4888, R&D Systems, Minneapolis, MN, USA), diluted 1:2000 in TBST. For all lectins (MGL, VVL, and WFL), the complexes were detected with ExtrAvidin-alkaline phosphatase (Sigma-Aldrich, St. Louis, MO, USA), diluted 1:10,000 in TBST, and incubated for 45 min at RT, followed by development with BCIP/NBT substrate (Sigma-Aldrich, St. Louis, MO, USA).

MGL pull-down assay

To identify amniotic fluid glycoproteins interacting with MGL, magnetic Protein G beads (Dynabeads Protein G Immunoprecipitation Kit, Thermo Fisher Scientific, Baltics UAB, Vilnius, Lithuania) were used following the manufacturer’s protocol with the modification described by Pirro et al.⁶⁰. Briefly, we coupled MGL-Fc chimera (extracellular domains of MGL fused to human immunoglobulin G1 Fc fragment) in the proportion of 12.5 µg of MGL-Fc per 50 µL of magnetic beads. The coupling reaction was carried out for 10 min at RT. After extensive washing, the lectin-conjugated magnetic beads were pre-incubated with 15 mM Tris–HCl buffer, pH 7.4, with 0.15 M NaCl (TBS) containing 10 mM Ca²⁺ or additionally 25 mM GalNAc (as a negative control), followed by adding a pooled amniotic sample containing 1 mg of protein. Reaction of binding of target ligands was conducted for 30 min at RT with rotation. The unbound proteins were washed out followed by elution of MGL-reactive proteins with 10 mM EDTA in TBS. Elution was performed for 2 min with rotation at RT, and eluate containing target ligands was collected for further analysis.

Affinity chromatography

To confirm that the proteins released in the pull-down assay indeed carried carbohydrate structures constituting a target for MGL, affinity chromatography was performed using agarose-immobilized plant lectins, namely VVL and WFL (Vector Labs, Vector Laboratories, Mowry Ave, Newark, CA, US), specific for single GalNAc or LacdiNAc epitopes respectively. A pooled sample of amniotic fluid, containing 0.4 mg of protein, was mixed with 400 µL of lectin-agarose in TBS and stirred gently overnight at 4 °C. The unbound proteins were washed out on the column and lectin-reactive glycoproteins were eluted with 0.2 M GalNAc in 0.2 M acetate buffer, pH = 4.0. The released proteins were immediately neutralized, buffer exchanged against PBS, washed three times and concentrated on 3 kDa MWCO protein concentrators (Thermo Fisher Scientific, Waltham, MA, USA). In the next step, 7.5 µL of each isolate was separated by SDS-PAGE in a 12.5% gel, and to detect proteins, a silver-staining procedure⁵⁹ was applied.

Liquid chromatography–mass spectrometry (LC–MS) identification of glycoproteins

In-gel digestion

Mass spectrometry experiments were performed at the Mass Spectrometry Laboratory at the Institute of Biochemistry and Biophysics PAS. Gel pieces were dried with acetonitrile (ACN) and subjected to reduction with 10 mM dithiothreitol (DTT) in 100 mM ammonium bicarbonate for 30 min at 57 °C. Cysteines were then alkylated with 0.5 M iodoacetamide in 100 mM ammonium bicarbonate (45 min in a darkroom at RT) and proteins were digested overnight with 10 ng/µL trypsin (Promega GmbH) in 25 mM ammonium bicarbonate at 37 °C.

Protein digestion

Precipitates were reconstituted in 100 mM ammonium bicarbonate buffer (ABC). Samples were incubated for 60 min on a vortex with 5 mM tris(2-carboxyethyl)phosphine (TCEP) at 60 °C, followed by the addition of methyl methanethiosulfonate (MMTS) to a final concentration of 20 mM. Protein digestion was performed overnight using 0.5 µg of trypsin (Promega GmbH) at 37 °C. Peptides were then acidified to 0.1% formic acid (FA).

Mass spectrometry

Samples were analyzed using an LC-MS system comprising an Evosep One (Evosep Biosystems) directly coupled to an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). One third of each sample was loaded onto disposable Evotips Pure C18 trap columns (Evosep Biosystems) as per the manufacturer’s instructions. Chromatographic separation was achieved at a flow rate of 220 nL/min or 500 nL/min with an 88-min or 44-min preformed gradient using a PepSep analytical column (C18, 1.5 μm beads, 150 μm ID, 15 cm long, Bruker). Data acquisition was conducted in positive ion mode using a data-dependent method. The MS1 resolution was set at 60,000 with a normalized automatic gain control (AGC) target of 300%, automatic maximum injection time, and a scan range of 300 to 1600 m/z. For MS2, the resolution was set at 15,000 with a standard normalized AGC target, automatic maximum injection time, and the top 40 precursors within a 1.6 m/z isolation window were selected for MS/MS analysis. Dynamic exclusion was applied for 20 s with a mass tolerance of ± 10 ppm, and the precursor intensity threshold was set at 5 × 10³. Precursors were fragmented in higher-energy collisional dissociation (HCD) mode with a normalized collision energy of 30%. The spray voltage was 2.1 kV, funnel RF level was set to 40, and the heated capillary temperature was maintained at 275 °C.

Data analysis

Obtained raw data were preprocessed with Mascot Distiller (version 2.8, Matrixscience) and protein identification was performed with Mascot Server (version 2.8.3, Matrixscience) using the Homo sapiens protein database derived from Swissprot (version 2024_04 or 2024_01), supplemented with contaminant sequences (CRAP database, 115 sequences). The rest of the parameters were as follows: enzyme—Trypsin, fixed modification—Methylthion (C), variable modification—Oxidation (M) and HexNAc (ST), instrument—HCD, missed cleavages-1. Each file was subjected to offline mass recalibration with typical resulting parent mass tolerance of 5 ppm and fragment mass tolerance of 0.01 Da. The false discovery rate (FDR) was computed with the target/decoy strategy, using a reversed database generated by Mascot, and kept below 1% or below 5% (for gel samples). For gel samples, relaxed FDRs were applied due to the relatively small number of fragmentation spectra, which affects FDR calculations and is typical for gel samples.

Bioinformatics analysis

The STRING 12.0 database (STRING Consortium, Swiss Institute of Bioinformatics, Lausanne, Switzerland, access 31 October 2024)⁶¹ was used to identify predicted functional partners and to construct an interaction network of the proteins (PPI) identified as MGL-reactive. Analysis was performed using the evidence mode involving eight types of evidence (neighborhood in the genome, gene fusions, co-occurrence across genomes, protein homology, co-expression, experimental/biochemical data associated with curated databases, and co-mentions in PubMed abstracts). For proteins defined as the most efficient MGL binders (analysis A) the PPI network included no more than 5 interactions of the first shell with the medium confidence (0.400) whereas in analysis B, considering a broader range of proteins reacting with MGL, the PPI network included interactions between query proteins only with a medium confidence level of 0.400. To visualize significant gene clusters, the k-means algorithm was used, with the number of clusters set to 3. Enrichment was performed in terms of Gene Ontology and Reactome pathways using the whole genome as the background and considering terms with maximum FDR ≤ 0.05; minimum signal ≥ 0.01; minimum strength ≥ 0.01 and minimum count in network = 2.

Results

Detection of MGL ligands

For the preliminary assessment of MGL reactivity with amniotic fluid proteins, lectin-blotting analysis was performed. The pooled samples of amniotic fluid were separated by SDS-PAGE and subjected to the silver staining procedure (Fig. 1a) or transferred onto nitrocellulose membrane. The membranes containing GalNAc-BSA glycoconjugate (positive control) (Fig. 1b), BSA (negative control) (Supplementary Fig. S1) and amniotic fluid proteins (Fig. 1c) were incubated with recombinant MGL. The lanes containing separated proteins of amniotic fluid samples were additionally probed with plant lectins specific for the terminal GalNAc residue—VVL (Fig. 1d) and WFL (Fig. 1e). The MGL and WFL ligands were detected in five fractions (Fig. 1 fractions 1–5) while the ligands of VVL were present in four electrophoretic bands (Fig. 1 fractions 2–5). MGL binding was abolished in the presence of 25 mM GalNAc or in the absence of Ca²⁺ ions (Supplementary Fig. S1), which confirms that the interaction is calcium-dependent and occurs with the participation of sugar residues. Lectin-reactive protein fractions were excised from the gel after CBB staining, then trypsin-digested and analyzed by means of LC-MS to identify the proteins. The top 15 records from each band are presented in Table 1.

Fig. 1.

Lectin-reactivity patterns of pooled amniotic fluid samples with specified Gene Ontology assignments for proteins contained in each lectin-reactive fraction: a silver stained amniotic fluid proteins after SDS-PAGE separation—representation of protein bands pattern of whole amniotic fluid sample; b GalNAc-BSA glycoconjugate probed with MGL—positive control confirming the carbohydrate specificity of MGL; c amniotic fluid proteins probed with MGL—representation of MGL-reactive fractions; d amniotic fluid proteins probed with VVL—representation of VVL-reactive fractions overlapping with MGL-reactive bands (fractions: 2–5); e amniotic fluid proteins probed with WFL—representation of WFL-reactive fractions overlapping with MGL-reactive bands (fractions: 1–5). Original blots/gels are presented in Supplementary Fig. S1.

Table 1.

Proteins contained in lectin-reactive bands identified by LC–MS.

Fraction	Protein	Gene name	Molar mass [Da]	Score	Matches	Sequences
1	Apolipoprotein B-100	APOB	516,648	36,336	672	221
	Fibronectin	FN1	275,742	19,134	294	79
	Complement C3	C3	188,569	8905	157	70
	Alpha-2-macroglobulin	A2M	164,613	6367	116	46
	Deleted in malignant brain tumors 1 protein	DMBT1	268,039	5477	79	19
	Pappalysin-2	PAPPA2	203,486	3084	56	32
	Pregnancy zone protein	PZP	165,242	2988	61	31
	Complement C4-B	C4B	194,170	2826	49	25
	Complement C4-A	C4A	194,261	2598	46	25
	Inter-alpha-trypsin inhibitor heavy chain H1	ITIH1	101,782	2564	43	15
	Inter-alpha-trypsin inhibitor heavy chain H2	ITIH2	106,853	2377	44	21
	Mucin-5B	MUC5B	611,584	2162	37	31
	von Willebrand factor	VWF	322,401	1856	32	26
	Complement factor H	CFH	143,680	1791	33	23
	Vascular endothelial growth factor receptor 1	FLT1	152,554	1562	32	20
2	Serotransferrin	TF	79,280	25,185	419	68
	Albumin	ALB	71,317	13,796	236	59
	Complement C3	C3	188,569	12,480	183	67
	Complement C4-B	C4B	194,170	5472	93	44
	Complement C4-A	C4A	194,261	5404	93	44
	Alpha-1B-glycoprotein	A1BG	54,790	4981	88	17
	Hemopexin	HPX	52,385	4944	87	20
	Ceruloplasmin	CP	122,997	2912	52	21
	Immunoglobulin heavy constant mu	IGHM	52,518	1984	39	15
	C4b-binding protein alpha chain	C4BPA	69,042	1809	31	18
	Insulin-like growth factor-binding protein complex acid labile subunit	IGFALS	66,735	1605	25	17
	Bone marrow proteoglycan	PRG2	25,874	1269	23	10
	Inter-alpha-trypsin inhibitor heavy chain H1	ITIH1	101,782	1238	16	8
	Heparin cofactor 2	SERPIND1	57,205	1218	25	14
	Lumican	LUM	38,747	1165	19	9
3	Albumin	ALB	71,317	56,308	890	70
	Complement C3	C3	188,569	6055	101	48
	Serotransferrin	TF	79,280	3674	59	31
	Alpha-1-antitrypsin	SERPINA1	46,878	3537	58	25
	Hemopexin	HPX	52,385	2646	39	19
	Immunoglobulin heavy constant alpha 1	IGHA1	43,620	1969	29	11
	Alpha-2-HS-glycoprotein	AHSG	40,114	1766	24	9
	Ceruloplasmin	CP	122,997	1627	27	13
	Immunoglobulin heavy constant gamma 3	IGHG3	50,202	1319	19	11
	Kininogen-1	KNG1	72,996	1280	17	13
	Angiotensinogen	AGT	52,322	1246	14	7
	Antithrombin-III	SERPINC1	53,025	1200	26	13
	Immunoglobulin alpha-2 heavy chain	IGHA2	49,816	1055	14	7
	Complement C4-A	C4A	194,261	1044	16	12
	Alpha-1-antichymotrypsin	SERPINA3	47,792	1022	19	9
4	Alpha-1-antitrypsin	SERPINA1	46,878	18,937	344	37
	Albumin	ALB	71,317	13,347	227	58
	Immunoglobulin gamma-1 heavy chain	IGHG1	49,925	10,929	190	21
	Vitamin D-binding protein	GC	54,480	7,090	108	33
	Immunoglobulin heavy constant gamma 4	IGHG4	44,431	6,271	97	15
	Immunoglobulin heavy constant gamma 2	IGHG2	44,519	5,596	113	17
	Immunoglobulin heavy constant gamma 3	IGHG3	50,202	3,924	76	17
	Serotransferrin	TF	79,280	3,717	65	35
	Angiotensinogen	AGT	52,322	2,476	36	10
	Alpha-2-HS-glycoprotein	AHSG	40,114	2,475	36	10
	Lumican	LUM	38,747	2,460	42	12
	Antithrombin-III	SERPINC1	53,025	2,349	42	22
	Ceruloplasmin	CP	122,997	2,141	36	15
	BPI fold-containing family B member 1	BPIFB1	52,580	1,787	31	14
	Hemopexin	HPX	52,385	1,738	28	17
5	Immunoglobulin kappa constant	IGKC	11,929	11,534	164	10
	Apolipoprotein A-I	APOA1	30,759	6,707	111	11
	Albumin	ALB	71,317	5,830	113	26
	Immunoglobulin lambda constant 2	IGLC2	11,458	2,867	49	42
	Immunoglobulin lambda constant 1	IGLC1	11,512	2,304	45	7
	Complement factor D	CFD	27,529	1,989	29	7
	Pancreatic adenocarcinoma up-regulated factor	ZG16B	18,867	1,722	24	6
	Triosephosphate isomerase	TPI1	26,938	1,584	24	9
	Serotransferrin	TF	79,280	1,497	28	6
	Fibronectin	FN1	275,742	1,408	26	11
	Pulmonary surfactant-associated protein B	SFTPB	43,514	1,401	22	18
	Immunoglobulin kappa variable 3–20	IGKV3-20	12,663	1,391	20	17
	Metalloproteinase inhibitor 1	TIMP1	23,840	1,386	19	9
	Prostaglandin-H2 D-isomerase	PTGDS	21,243	1,333	22	4
	Immunoglobulin kappa variable 3D-20	IGKV3D-20	12,621	1,284	17	7

The analysis revealed a wide range of proteins within each identified band. Therefore, to simplify, the protein sets of each fraction have been categorized according to their function based on Gene Ontology enrichment in term of biological processes. Hits with scores above 1000 were included in the analysis. The most significant terms of Gene Ontology are presented in Fig. 1. Functional enrichment analysis indicated that most of the terms contained in all five fractions are involved in immunological processes such as regulation of the immune effector process (fraction 1), inflammatory response (fractions 3 and 5), complement activation (fraction 1), or defense response (fractions 2, 3 and 5). Moreover, many proteins of fractions 1–3 have been identified as being involved in regulation of metabolic processes, while the proteins predominating in fractions 3 and 4 show regulatory properties in other biological processes. Numerous proteins of fractions 3–5 are engaged in the response to stress and those in fractions 2 and 4 in the response to other stimuli including external stimuli. Fraction 5 is rich in proteins involved in maintaining homeostasis in the human body.

Isolation and identification of MGL ligands

To assess which proteins are the actual ligands for MGL, a pull-down assay was carried out to isolate those that are MGL-reactive. Additionally, affinity chromatography using agarose-immobilized plant lectins, namely VVL and WFL, specific for terminal GalNAc/LacdiNAc sugar epitopes, was performed. Lectin-reactive proteins identified by LC-MS analysis of the obtained isolates are shown in Table 2 (MGL ligands), Table 3 (WFL ligands), and Table 4 (VVL ligands).

Table 2.

LC–MS protein identification of MGL-pull down assay isolate.

Protein	Gene name	Molar mass [Da]	Score	Matches	Sequences
Mucin-5B	MUC5B	608,561	2613	82	30
Mucin-5AC	MUC5AC	598,719	1420	75	32
Deleted in malignant brain tumors 1 protein	DMBT1	266,593	1010	28	13
BPI fold-containing family B member 1	BPIFB1	52,546	720	24	10
Retinoic acid-induced protein 3	GPRC5A	40,547	690	20	5
Pulmonary surfactant-associated protein A2	SFTPA2	26,521	544	9	3
Fibrinogen beta chain	FGB	56,444	424	15	8
Uromodulin	UMOD	71,922	406	18	8
Apolipoprotein B-100	APOB	516,373	357	17	16
Epidermal growth factor receptor kinase substrate 8-like protein 2	EPS8L2	81,076	348	12	7
Fibrinogen gamma chain	FGG	51,985	293	13	7
G-protein coupled receptor family C group 5 member B	GPRC5B	45,180	280	6	3
Putative annexin A2-like protein	ANXA2P2	38,773	260	12	6
Ezrin	EZR	69,462	255	17	10
Moesin	MSN	67,870	245	16	9
Mucin-1	MUC1	122,167	244	10	5
Fibrinogen alpha chain	FGA	95,512	242	17	6
Transcriptional adapter 2-beta	TADA2B	48,900	238	14	1
Programmed cell death 6-interacting protein	PDCD6IP	96,469	159	6	5
IST1 homolog	IST1	39,863	156	8	3
Lysozyme C	LYZ	16,894	156	8	3
Apolipoprotein A-I	APOA1	30,759	148	7	5
Syntenin-1	SDCBP	32,562	145	6	2
Annexin A1	ANXA1	38,874	136	4	2
G-protein coupled receptor family C group 5 member C	GPRC5C	48,622	136	3	2
Dermcidin	DCD	11,369	136	4	3
CD9 antigen	CD9	25,859	134	2	1
Pancreatic adenocarcinoma up-regulated factor	ZG16B	18,867	133	3	2
Filaggrin-2	FLG2	249,032	129	2	2
Serotransferrin	TF	78,839	125	7	5
Hemoglobin subunit beta	HBB	15,949	117	6	3
Stomatin	STOM	31,849	108	4	3
Protein S100-A8	S100A8	10,874	102	6	4

Table 3.

LC-MS protein identification of WFL ligands.

Protein	Gene name	Molar mass [Da]	Score	Matches	Sequences
Fibronectin	FN1	275,047	3002	98	50
Alpha-2-macroglobulin	A2M	164,338	2897	98	43
Mucin-5B	MUC5B	608,561	2875	91	43
Uromodulin	UMOD	71,922	2266	76	16
Lactotransferrin	LTF	79,650	2260	62	28
Serotransferrin	TF	78,839	2137	71	29
Immunoglobulin heavy constant mu	IGHM	52,397	1962	61	14
Ceruloplasmin	CP	122,831	1633	41	25
Plasma protease C1 inhibitor	SERPING1	55,303	1631	50	12
Mucin-5AC	MUC5AC	598,719	1551	58	28
Galectin-3-binding protein	LGALS3BP	66,025	1517	38	15
Apolipoprotein A-I	APOA1	30,759	1405	42	14
Vascular endothelial growth factor receptor 1	FLT1	152,190	1164	43	20
Complement C3	C3	188,272	939	34	27
Diamine oxidase [copper-containing]	AOC1	85,646	880	37	18
Immunoglobulin gamma-1 heavy chain	IGHG1	49,804	847	28	8
Immunoglobulin kappa constant	IGKC	11,896	763	20	5
Alpha-1-antitrypsin	SERPINA1	46,845	753	26	12
Immunoglobulin heavy constant gamma 2	IGHG2	44,376	728	25	4
Deleted in malignant brain tumors 1 protein	DMBT1	266,593	721	23	13
Polymeric immunoglobulin receptor	PIGR	84,197	638	24	13
Zinc-alpha-2-glycoprotein	AZGP1	34,421	576	23	12
Attractin	ATRN	162,479	534	14	10
Immunoglobulin kappa light chain	IGK	23,594	523	13	6
Desmoplakin	DSP	333,546	514	22	22
Desmoglein-1	DSG1	114,504	501	18	11
Filaggrin-2	FLG2	249,032	426	10	7
Annexin A2	ANXA2	38,764	412	12	8
Apolipoprotein B-100	APOB	516,373	409	15	15
Haptoglobin	HP	45,728	409	12	8
Pregnancy zone protein	PZP	164,955	407	19	9
Pappalysin-1	PAPPA	184,719	403	17	14

Table 4.

LC-MS protein identification of VVL ligands.

Protein	Gene name	Molar mass [Da]	Score	Matches	Sequences
Serotransferrin	TF	78,839	2371	76	30
Lactotransferrin	LTF	79,650	1097	37	20
Immunoglobulin gamma-1 heavy chain	IGHG1	49,804	805	29	9
Immunoglobulin heavy constant gamma 2	IGHG2	44,376	657	25	6
Ceruloplasmin	CP	122,831	572	18	12
Desmoglein-1	DSG1	114,504	398	12	9
Vitamin D-binding protein	GC	54,171	383	17	11
Prolactin-inducible protein	PIP	16,792	361	9	4
Immunoglobulin kappa constant	IGKC	11,896	354	10	3
Alpha-1-acid glycoprotein 1	ORM1	23,709	354	10	5
Immunoglobulin heavy constant alpha 1	IGHA1	43,465	352	10	7
Alpha-1-acid glycoprotein 2	ORM2	23,818	249	6	4
Alpha-1-antitrypsin	SERPINA1	46,845	238	10	7
Hornerin	HRNR	282,963	231	6	5
Haptoglobin	HP	45,728	226	6	4
Hemopexin	HPX	52,241	215	7	4
Zinc-alpha-2-glycoprotein	AZGP1	34,421	211	11	9
Desmocollin-1	DSC1	101,119	199	4	2
Dermcidin	DCD	11,369	190	4	3
Complement decay-accelerating factor	CD55	42,201	189	12	5
Desmoplakin	DSP	333,546	188	8	7
Immunoglobulin heavy constant alpha 2	IGHA2	42,997	171	5	4
Actin, cytoplasmic 1	ACTB	41,986	162	6	5
Deleted in malignant brain tumors 1	DMBT1	266,593	160	3	2
Plasma protease C1 inhibitor	SERPING1	55,303	159	7	5
Uromodulin	UMOD	71,922	158	9	7
Transthyretin	TTR	15,969	157	4	3
Galectin-3-binding protein	LGALS3BP	66,025	149	3	3
Immunoglobulin kappa variable 3D-20	IGKV3D-20	12,599	145	1	1
Prostaglandin-H2 D-isomerase	PTGDS	21,199	142	4	4
Transcriptional adapter 2-beta	TADA2B	48,900	140	9	1
Caspase-14	CASP14	27,892	137	4	4
Apolipoprotein A-I	APOA1	30,759	130	3	3

The data presented in Tables 2, 3 and 4 show the 33 best scoring proteins. Considering that Protein G used for our pull-down assay strongly interacts with immunoglobulins and that the C1q molecule of complement 1 complex (C1) is able to bind antigen-antibody complex via Fc fragments, our analysis excluded all fragments of identified immunoglobulins as well as subcomponents of C1 complex—C1s, C1r, C1q (subunits A, B, C)—considering them as potentially co-isolated. Keratins, as common contaminant and proteins additionally isolated in the presence of inhibiting sugar—25 mM GalNAc, scoring above 100 (in negative control)—were also excluded from the potential set of MGL ligands. Venn diagrams showing the number of shared proteins isolated in MGL pull-down assays—in the presence and absence of inhibiting sugar—are presented in Supplementary Fig. S2. Based on the data obtained, the most effective MGL-binding proteins (scoring above 400) appear to be: mucin-5B (MUC5B); mucin-5AC (MUC5AC); deleted in malignant brain tumors 1 protein (DMBT1); BPI fold-containing family B member 1 (BPIFB1); retinoic acid-induced protein 3 (GPRC5A); pulmonary surfactant-associated protein A2 (SFTPA2); fibrinogen beta chain (FGB); and uromodulin (UMOD). The set of WFL-reactive, high scoring proteins is much wider, and top hits include: fibronectin (FN1), alpha-2-macroglobulin (A2M), mucin-5B (MUC5B), uromodulin (UMOD), lactotransferrin (LTF), serotransferrin (TF), immunoglobulin heavy constant mu (IGHM), ceruloplasmin (CP), plasma protease C1 inhibitor (SERPING1), mucin-5AC (MUC5AC), galectin-3-binding protein (LGALS3BP), apolipoprotein A-I (APOA1), and vascular endothelial growth factor receptor 1 (FLT1) (scores above 1000). Conversely, in the case of the VVL ligands, fewer proteins were identified in the isolates, and their panel had less overlap with the LC-MS identification of MGL-pull down assay. For identification of VVL-binding glycoproteins, serotransferrin (TF), lactotransferrin (LTF), immunoglobulin gamma-1 heavy chain (IGHG1), immunoglobulin heavy constant gamma 2 (IGHG2), and ceruloplasmin (CP) scored over 400.

The comparison of scores of proteins isolated using the MGL-pull down assay to those isolated with agarose-immobilized VVL and WFL is demonstrated in Fig. 2. The set of WFL-binding glycoproteins shows a greater degree of similarity in terms of the profile of proteins and their scores compared to VVL-reactive proteins. Three of the most significant isolated MGL ligands—MUC5B, MUC5AC, DMBT1—also show high scores in the range of WFL ligands. Interestingly, UMOD, indicated in the set of proteins scoring above 400 for MGL-reactive proteins, shows one of the highest scores in eluates of WFL-affinity chromatography. On the other hand, for VVL-affinity chromatography, the isolated proteins do not overlap significantly with the range of identified MGL ligands. However, DMBT1, GPRC5A, UMOD, identified among the top scoring hits from MGL-pull down assay, are also ligands for VVL. Considering the lectin preferences, it may suggest the presence of single GalNAc residues/Tn antigens in these proteins.

Fig. 2.

Heat-map represents the comparative overview of MGL and plant lectins ligands isolated in MGL-pull down assay and affinity chromatography using agarose-immobilized VVL and WFL. Among the plant lectins ligands, only those that overlap with MGL-binding proteins are included. Scale represents LC-MS scores.

The presence of such a modification in DMBT1 was also suggested by searching the LC-MS/MS data for glycopeptides carrying a single HexNAc (N-acetylhexosamine) linked to serine and threonine residue in the polypeptide backbone—potentially corresponding to O-linked single GalNAc/Tn antigen. Representative spectral data containing the glycopeptides with specific HexNAc ions at m/z 203.0794 have been extracted and presented in Fig. 3.

Fig. 3.

MS/MS fragmentation of A SAPGNAQFGQGSGPIVLDDVR and B SAPGNAWFGQGSGPIALDDVR sequences carrying HexNAc epitopes, found in DMBT1. Yellow square: HexNAc. Yellow y ions represent fragments with the HexNAc. Red y ions represent fragments without the HexNAc.

Bioinformatics analysis

Protein–protein interaction network

The STRING database was used to construct a protein-protein interaction (PPI) network complex. Two analyses were performed, the first for the most efficiently binding MGL ligands (with the highest scores)—with a cut-off score of 400—and their functional partners (A) and the second for a broader set of proteins with scores above 100 (B). The PPI network contained 13 nodes and 25 edges for analysis A (Fig. 4A) or 32 nodes and 57 edges for analysis B (Fig. 5A). In both cases, A and B, the analysis grouped the genes/proteins in 3 clusters (Figs. 4B–D, 5B–D). The first cluster of analysis A consists of 5 genes: DMBT1, MUC5AC, MUC5B, MUC7, TFF2. Reactome pathway enrichment analysis revealed that this cluster is mainly associated with O-glycan processing including termination of O-glycan biosynthesis as well as signaling of C-type lectin receptors (CLRs), including the dectin-2 family. The second cluster, containing four genes—BPIFA1, BPIFB1, BPIFB2, FGB—is mainly related to innate immune system and antimicrobial signaling pathways. Finally, the third cluster consists of 2 genes, SFTPA1 and SFTPA2, whose functions could be associated with surfactant metabolism. The first cluster of analysis B contains 10 genes—APOA1, APOB, FGA, FGB, FGG, HBB, LYZ, S100A8, TF, ZG16B—involved in Toll-like receptor cascades including regulation of TLR by endogenous ligands. The second cluster of the PPI network of proteins included in analysis B consists of 9 genes—ANXA1, CD9, EPS8L2, EZR, FLG2, IST1, MSN, PDCD6IP, SDCBP—participating in positive regulation of exocytosis. The third cluster includes 8 genes—BPIFB1, DCD, DMBT1, MUC1, MUC5AC, MUC5B, SFTPA2, UMOD—involved in pathways of O-glycan processing.

Fig. 4.

STRING analysis of glycoproteins identified as most efficiently binding MGL ligands and their functional partners. A PPI network; B–D significant gene clusters; E functional enrichment visualization—terms are grouped by similarity ≥ 0.8 and sorted by signal score. STRING v12.0 was used to generate the PPI network and functional enrichment visualization (https://string-db.org).

Fig. 5.

STRING analysis of glycoproteins identified as MGL ligands (LC-MS score above 100). A PPI network; B–D significant gene clusters; E functional enrichment visualization—terms are grouped by similarity ≥ 0.8 and sorted by signal score. STRING v12.0 was used to generate the PPI network and functional enrichment visualization (https://string-db.org).

Gene ontology and reactome pathways enrichment analysis

Functional enrichment analysis for sets of isolated proteins were performed using online databases. The enrichment was considered in the category of biological processes—in terms of Gene Ontology (GO) as well as pathways—using Reactome pathways. The first category of enrichment (biological processes) for analysis A assigned analyzed proteins/genes as involved in the antimicrobial humoral response and the latter (pathways) identified most of the genes as significantly enriched in innate immune system processes (Fig. 4E). Moreover, among the most significant enrichments, considering the signal score, are pathways related to termination of O-glycan biosynthesis, signaling of C-type lectin receptors (CLRs) including the dectin-2 family, and pathways modulated by antimicrobial peptides. Another relevant enrichment is regulation of TLR signaling by endogenous ligands. In addition, the analysis indicated annotation to pathways related to metabolism of proteins including surfactant metabolism and pathways involved in development of diseases of metabolism. Similar to analysis A, in analysis B a significant number of genes from the network were assigned to pathways of the innate immune system (Fig. 5E). The analysis identified the most significant enrichment pathways as those involved in regulation of TLR by endogenous ligands and events associated with defective TLR signaling cascades, such as MyD88 deficiency (TLR2/4) and IRAK4 deficiency (TLR2/4). In the category of biological processes (Gene Ontology), the following terms were identified as the most relevant: regulation of heterotypic cell-cell adhesion, induction of bacteria agglutination, antibacterial humoral response, plasminogen activation, and positive regulation of exocytosis.

Discussion

In humans, sugars attached to a protein or lipid core can act as specific markers on the surfaces of cells or soluble macromolecules. The large diversity in glycan structures and composition allows for the storage of biological information in these structures⁶². Such a glyco-code, referred to as ‘the third alphabet of life’ can be read by various types of proteins, including C-type lectin receptors^63,64. The importance of lectin–glycan interactions is most often studied in the context of pathogen and host immune cell interactions or the mechanisms of cancer metastasis⁶⁵. C-type lectins have evolved to differentiate ligands as self, non-self or altered-self structures and trigger the appropriate immune responses^66–68. Depending on the local microenvironment and the cell subpopulation by which lectins are expressed, CLRs may contribute to fine-tuning immunity at different levels—from induction of pro-inflammatory pathways to their resolution^69,70. However, these interactions do not always remain supportive of the host immune strategy. Invading cancer cells or some pathogens use physiological mechanisms for their purposes and have learned to trick the host immune system by interacting with CLRs, to suppress the host immune response and thus evade immune surveillance⁴⁰.

Protein-sugar interactions are also important at various stages of the fertilization process and pregnancy⁷¹. One aspect that is the subject of intensive research is the involvement of glycan structures in the development of proper immunological balance during pregnancy and maternal tolerance of the fetus^6,72. Clark et al. proposed the human fetoembryonic defense system hypothesis (hu-FEDS), which postulates that glycan-mediated interactions in the female reproductive tract during fertilization and gestation are similar to the mechanisms that pathogens and metastatic cancer cells use to escape host immune surveillance. One of the first assumptions of the hu-FEDS hypothesis is the involvement of glycans, exposed by glycodelin-A (GdA) and mucins present in the placenta, amniotic fluid, and pregnant uterus, in suppression of the maternal immune response. Moreover, although not discussed in the original Hu-FEDS concept, uromodulin was described later as another pregnancy-related glycoprotein with immune-deviating activities, which strongly supports the Hu-FEDS postulates^2,3.

Over time, researchers began to analyze the role of lectins—recognizing particular glycan structures—in interactions regulating the immune response in pregnancy, such as galectins or Siglecs^{9–11,73−76}. Interestingly, the presence of potential sugar ligands for MGL, including LacdiNAc and (sialo) Tn antigen, has been detected in amniotic fluid, on cells of placental or endometrial origin, and in fetal tissues. However, the interaction between them and its possible role have not been determined so far^2,10,26–31.

Our preliminary analysis revealed calcium-dependent binding of MGL to glycoproteins contained in electrophoretically separated amniotic fluid fractions. The analysis identified five MGL-reactive fractions with molecular mass from 245 to 23 kDa. Proteomic profiling of the fractions revealed a broad panel of proteins engaged mainly in immunological processes as well as having regulatory properties in metabolic/biological processes and in maintaining homeostasis in the human body. Numerous proteins are engaged in the response to stress and other stimuli, including external stimuli.

To indicate which proteins actually exhibit MGL binding ability, a pull-down assay was carried out to isolate those that are MGL-reactive. Additionally, lectin affinity chromatography with agarose-immobilized plant lectins—VVL and WFL (specific for terminal GalNAc/LacdiNAc) was used in parallel. The analysis showed a greater similarity between the protein composition of the WFL eluate and the set of proteins identified in the MGL-pull down assay than between the proteins contained in isolates of the VVL and MGL-pull down assay. WFL recognizes LacdiNAc structure with greater affinity than GalNAc, whereas VVL shows preference for a single GalNAc residue attached directly to a serine or threonine in the polypeptide backbone via an O-glycosidic bond (Tn antigen)^77–79. Based on detailed comparison of obtained data and lectin preferences, we can hypothesize that the interaction of MGL with amniotic fluid proteins occurs mainly via more elongated structures than the Tn antigen. However, using LC-MS/MS data, the presence of the Tn antigen—a single GalNAc residue—has been suggested in DMBT1, which is one of the most efficiently binding ligands of MGL.

The most relevant ligands binding MGL, as well as WFL, seem to be: mucins—MUC5AC and MUC5B; BPIFB1; DMBT1; and UMOD. As mentioned above, some of the proteins identified in our study, i.e. mucins and uromodulin, but also glycodelin A, form the basis of the original Hu-FEDS assumptions.

According to the available data, glycodelin A shows the presence of LacdiNAc—a glycan acting as a sugar ligand for MGL^2,31. However, it was not detected in our study as a protein binding MGL or plant lectins. The lack of identification may be caused by a low amniotic glycodelin A concentration at the normal delivery date, since the highest values are observed between the 12th and 20th weeks of gestation⁸⁰.

An interesting result is the ability of MGL to bind uromodulin. UMOD is a protein exclusively produced by the kidney, and it has been detected in the urine of pregnant women and amniotic fluid—which it reaches along with the fetal urine^81–85. It is also present in the urine of men and non-pregnant females, designated as Tamm–Horsfall protein (THP)⁸³. Research studies have revealed the multifunctional nature of UMOD, and it has been reported that interactions of UMOD with the immune system are one of the key properties of this protein^86,87. Both urinary isoforms of UMOD exhibit immunoregulatory properties, although the pregnancy-related one has been shown to be more effective⁸⁸. It is suggested that the immune-deviating activity is associated with glycosylation patterns, but their structural profile and biological activities are still controversial⁸⁷. Both urinary isoforms express the specific Sda antigen, exposing the GalNAc residue⁸⁸. Such a structure may constitute a sugar ligand for MGL, and the interaction may be involved in the immunoregulatory activity of UMOD. However, our study focuses on UMOD of amniotic fluid; hence it would be valuable to confirm such glycan structures in the amniotic isoform or to determine other glycan structures relevant for MGL binding. Nevertheless, the presence of UMOD in amniotic fluid—and its possible interaction with MGL—may have an important role in fetal protection against rejection and may be relevant to the assumptions of the hu-FEDS hypothesis.

Our study identified MUC5AC and MUC5B as the main amniotic ligands for MGL. Mucin-5B has already been identified as an amniotic fluid protein⁸⁹ and, together with mucin-5AC, as a component of the cervical mucus plug during pregnancy⁹⁰. Cervical MUC5B and MUC5AC are involved in the maintenance of pregnancy and prevention of preterm birth. It is believed that the protective role of these mucins during pregnancy is based on their antibacterial properties. Due to the ability to trap bacteria, both mucins prevent their invasion into the tissues, thus constituting a physical barrier to ascending pathogens^91,92. Furthermore, it has been suggested that cervical mucins express immune-active glycans, pivotal for the local anti-microbial immune defense strategy^91,93. However, according to the established paradigm, immunity during pregnancy must be precisely balanced, because the mother’s body needs to effectively fight pathogens while maintaining tolerance of the semi-allogeneic fetus until late gestation and parturition. Labor is a pro-inflammatory state, and responding to the infection may contribute to untimely activation of the inflammatory pathways related to onset parturition, resulting in preterm birth⁹³. Therefore, maintaining immune homeostasis during pregnancy is such a challenging act, important for the development of the fetus, healthy gestation and termly delivery. We assume that the association of MUC5B/AC with the maintenance of pregnancy may not only be related to the trapping of invasive bacteria but, potentially through its ability to bind MGL, may also fine-tune the immune response both in the cervix and the intrauterine space, thereby preventing the activation of pathways that induce preterm labor. The fact that the presence of LacdiNAc epitopes on cervical plug glycoproteins was negatively correlated with pro-inflammatory factors such as matrix metalloproteinases (MMPs), complement components, and inflammatory cytokines further supports this supposition⁹³.

The amniotic fluid shows various defense mechanisms to prevent intra-amniotic infections and inflammation. Amniotic antimicrobial proteins and peptides provide the front-line defense in the innate immune response⁹⁴. BPI fold-containing family B member 1 is a highly glycosylated glycoprotein and has been identified as part of the human embryonic secretome^95–97. BPIFB1 is known to participate in sensing and responding to bacteria due to its structural similarity with lipopolysaccharide (LPS)-binding protein (LPB). The LBP domain of BPIFB1 causes binding to LPS and contributes to anti-bacterial immunity. However, BPIFB1 has also been reported to stimulate anti-inflammatory pathways during microbial infections and TLR agonist stimulation⁹⁸. A similar property has been suggested for DMBT1, another MGL ligand identified by us in term amniotic fluid, potentially originating from the fetal gastrointestinal tract⁹⁹. DMBT1 is a member of the secreted cysteine-rich scavenger receptor protein family and has been shown to bind and aggregate a broad spectrum of bacteria. It has also been suggested that DMBT is able to recognize self-structures, such as surfactant protein D or surfactant protein A. In response to LPS, DMBT1 has been shown to contribute to the inhibition of TLR4-dependent cellular responses, such as NF-kB activation and cytokine secretion¹⁰⁰. In accordance with the above, both amniotic glycoproteins, namely BPIFB1 and DMBT1, appear to be important for immune homeostasis. We hypothesized that in amniotic fluid MGL-reactive BPIFB1 and DMBT1 may play a protective role in supporting tolerance for the semi-allogeneic fetus. Moreover, BPIFB1 and DMBT1, similarly as we assume for MUC-5B/AC, may make it possible to achieve a critical immune balance—necessary to maintain the ability to defend against pathogens, but also to prevent untimely activation of pro-inflammatory pathways related to spontaneous premature birth. These immunomodulatory properties of the mentioned proteins may be supported by the ability to interact with the MGL receptor. However, future studies are essential to confirm the involvement of MGL-mediated interactions in immunomodulation during pregnancy.

Bioinformatics analysis of the protein-protein interaction network and functional enrichment of the panel of isolated MGL ligands may support the above assumptions, indicating the participation of most of the identified proteins and their partners in the immune response, the response against pathogens, and the regulation of TLR signaling. Interestingly, the impact of MGL engagement on TLR pathways has been extensively analyzed. It has been found that MGL-mediated induction of immunosuppressive IL-10 requires concomitant TLR stimulation, e.g. TLR4 induced by bacterial LPS. Research has demonstrated that MGL interaction with its specific ligand modulates TLR signal transduction for enhanced IL-10 and TNF-α synthesis^37,101.

Moreover, our analysis showed that some isolated proteins participate in CLR signaling, which may support their ability to be recognized by MGL and possibility to act in a network of interactions with other CLRs. In addition to immunological processes, the analysis also assigned a group of identified proteins as engaged in O-glycan processing, including termination or defective O-glycan biosynthesis. It suggests that MGL binding of proteins included in this group (mucins) may occur through O-linked glycans. Another interesting annotation is surfactant metabolism. Considering this enrichment, it is worth mentioning one of the surfactant proteins: SFTPA2. SFTPA2 was isolated using the MGL-pull down assay, although it was not detected among VVL and WFL ligands. As mentioned, SFTPA2 is a binding partner for DMBT1, so its co-isolation is possible, especially since the protein has a carbohydrate recognition domain (CRD) that allows it to bind other glycoproteins¹⁰². However, its ability to interact directly with MGL should not be completely excluded, and the possible role of this binding is worth considering. In addition to surfactant-related functions, surfactant proteins, including SFTPA2, have been found to play an important role in the innate host defense of the lung¹⁰². Moreover, SFTPA2 is crucial for maintaining the immune homeostasis in the lung, due to its ability to control inflammatory processes. Amniotic SFTPA is derived from the fetal lung, since during fetal development, surfactant is synthesized and released into the surrounding environment. The levels of surfactant proteins in amniotic fluid reflect fetal lung maturity, which is crucial for labor initiation¹⁰³. SFTPA in amniotic fluid has been reported to mediate a unique immunoregulatory mechanism during gestation¹⁰⁴. As in the lung environment, the immunomodulatory features of SFTPA2 within the uterus are associated with the control of cytokine synthesis and modulation of TLR functions (including in response to infection). Moreover, it has been suggested that SFTPA2-mediated inhibition of pro-inflammatory cytokine leads to regulation of the onset of labor^103,104. SFTPA2 has been shown to interact with multiple soluble receptors as well as membrane receptors on the surface of macrophages¹⁰². Thus, taking into account all aspects, MGL seems to be a novel receptor capable of binding SFTPA2 and thus modulating the immune responses during pregnancy in an SFTPA-dependent manner, and on the other hand, SFTPA2 may be a new potential ligand for MGL, beside the mucins, DMBT1, BPIFB1, and others considered above. The possible interactions between MGL and SFTPA2 may support the maintenance of delicate immune balance during gestation and also potentially participate in pathways related to control of parturition.

Although MGL has recently been the subject of intensive research, especially in the context of carcinogenesis and infectious disease^17,20 its role still remains somewhat enigmatic. It is believed, however, that the primary function of MGL is to maintain proper immune balance, preventing excessive inflammation and tissue damage⁵³. This function may be crucial during pregnancy, as the mother’s body must tolerate the semiallogeneic fetus in the uterus for many weeks while maintaining the ability to fight pathogens. It is worth noting that the most efficient MGL-binding partners identified in our analysis constitute the basis for the assumptions of the hu-FEDS hypothesis. Considering the reports of the participation of MGL in carcinogenesis and pathogenesis of some infectious diseases, the results of our research may support the assumptions of this hypothesis. The role of MGL binding in the context of pregnancy may be related to the development of tolerance towards the fetus, its maintenance at later stages, but especially to the regulation of the response to ascending pathogens. However, the limitation of our research is analysis of amniotic fluid collected at the time of delivery—after 38 weeks of gestation. The amniotic fluid proteome profile changes with gestational age. Thus, it seems worthwhile to investigate whether the interaction between MGL and amniotic fluid glycoproteins also occurs at earlier stages of pregnancy. Such analysis could reveal the broader context of the MGL-amniotic fluid interactome. However, obtaining amniotic fluid samples at earlier stages of pregnancy may be difficult due to the invasive nature of the procedure. Another issue that remains to be investigated is the nature of MGL binding to amniotic fluid proteins and the involvement of specific glycan structures in such interactions. Although the aim of our study was to identify proteins that are potential MGL ligands in amniotic fluid, we attempted to initially assess the presence of HexNAc modifications in peptides identified in eluates. We found sequences with HexNAc modifications at serine and threonine residues—which may potentially correspond to the Tn antigen/single GalNAc. Therefore, it would be valuable to perform a study focusing on glycan profile, employing a more sensitive (glycomic) approach to verify the presence of such modifications and to characterize extended N- and O-glycan structures bearing terminal GalNAc residues, which seem to be more important in term amniotic fluid.

Nevertheless, MGL has the potential to be an important factor in maintaining homeostasis during gestation, orchestrating signaling pathways to resolve inflammation and preventing the activation of pathways related to premature birth during the fight against microorganisms. Thus, the MGL-mediated interactions may be an important part of the signaling network that ensures the critical immune balance necessary to maintain the fetus in the uterus. However, future studies are essential for elucidating the exact nature and role of the interactions between MGL and amniotic fluid ligands.

Conclusions

We have demonstrated for the first time the potential ligands of MGL in amniotic fluid which may participate, via MGL binding, in development of immune homeostasis—essential for the maintenance of pregnancy and the prevention of miscarriage and preterm birth. However, full understanding of this issue remains elusive, warranting further exploration.

We suggest that the identified amniotic MGL ligands may serve in future as predictors in assessing the risk of preterm birth. Their exact glycosylation profile and the exact nature of binding to MGL are also worth investigating, as they may be helpful in the development and implementation of modern therapies in the context of avoiding preterm birth.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MGL

Macrophage galactose-type C-type lectin

CTL

C-type lectin

DCs

Dendritic cells

TCR

T-cell receptor

TLR

Toll-like receptor

Tregs

Regulatory T cells

APCs

Antigen-presenting cells

LPS

Lipopolysaccharide

VVL

Lectin from Vicia villosa

WFL

Lectin from Wisteria floribunda

MUC5B

Mucin-5B

MUC5AC

Mucin-5AC

DMBT1

Deleted in malignant brain tumors 1 protein

BPIFB1

BPI fold-containing family B member 1

GPRC5A

Retinoic acid-induced protein 3

SFTPA2

Pulmonary surfactant-associated protein A2

FGB

Fibrinogen beta chain

UMOD

Uromodulin

FN1

Fibronectin

A2M

Alpha-2-macroglobulin

LTF

Lactotransferrin

TF

Serotransferrin

IGHM

Immunoglobulin heavy constant mu

CP

Ceruloplasmin

SERPING1

Plasma protease C1 inhibitor

LGALS3BP

Galectin-3-binding protein

APOA1

Apolipoprotein A-I

FLT1

Vascular endothelial growth factor receptor 1

IGHG1

Immunoglobulin gamma-1 heavy chain

IGHG2

Immunoglobulin heavy constant gamma 2

Hu-FEDS

Human fetoembryonic defense system

THP

Tamm–Horsfall protein

Author contributions

JS: conceptualization, methodology, experiments—planning and performing, data curation, formal analysis, writing—original draft preparation, writing—review and editing, visualization, funding acquisition. MOP: supervision, writing—reviewing and editing, funding acquisition. MZ: patients’recruitment for study, sample collection and clinical characteristics. All authors approved the manuscript.

Funding

This research was funded by the Wroclaw Medical University, Poland, grant number SUBK.A411.23.063.

Data availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE^105,106 partner repository with the data set identifier PXD063586 and 10.6019/PXD063586.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study complied with guidelines of the Declaration of Helsinki. Ethical approval was obtained from the Wroclaw Medical University Bioethics Council (no. of approval: KB 160/2023 N). All participants gave their informed written consent to be enrolled in the study.