Treg cells retain stable lineage commitment during pregnancy in mice after late gestation inflammatory challenge

TL;DR

This study shows that Treg cells in pregnant mice maintain their identity even when inflammation is induced, which could help understand preterm birth.

Contribution

The study reveals that Treg cells retain lineage stability during pregnancy despite inflammatory challenges.

Findings

Ex-Foxp3 cells were found in uterine-draining lymph nodes but did not increase with inflammation.

Treg cells retained their lineage program and did not adopt proinflammatory phenotypes during late gestation.

RNA sequencing showed loss of Treg lineage markers in ex-Foxp3 cells but no expansion due to inflammation.

Abstract

Inflammation is a major driver of preterm birth, a common pregnancy disorder and the leading cause of childhood death. T regulatory (Treg) cells are essential mediators of maternal fetal tolerance and are critical for constraining uterine inflammation. In some tissue settings, loss of Foxp3 expression can cause instability in Treg cell lineage commitment, elevated production of proinflammatory cytokines and compromised suppressive function. Whether preterm birth susceptibility is associated with loss of lineage fidelity and adoption of proinflammatory phenotypes in Treg cells is unknown. In this study, we investigated the lineage stability of Treg cells in vivo in pregnant mice using a Foxp3 fate‐mapping system and models of preterm birth induced by late‐gestation inflammatory challenge with lipopolysaccharide (LPS) or interleukin‐1β (IL‐1β). Ex‐Foxp3‐expressing (ex‐Foxp3) cells were observed in the uterus‐draining lymph nodes (udLNs) in non‐pregnant mice and in similar abundance across normal gestation, and a proportion expressed proinflammatory cytokines IFNγ and/or IL‐17A. Bulk RNA‐sequencing of sorted Treg and ex‐Foxp3 cells from late‐gestation udLNs revealed substantial loss of the Treg cell lineage program in ex‐Foxp3 cells, characterized by reversal in expression of canonical Treg cell genes and pathways. Late gestation LPS or IL‐1β administration to induce preterm birth did not expand the ex‐Foxp3 cell population in the uDLNs or uterine decidua. We conclude that uterine Treg cells exhibit a high level of lineage stability in pregnancy regardless of proinflammatory challenge. Whether there is any biological or pathophysiological significance of ex‐Foxp3 cells in gestational tissues remains to be defined.

In this study, we used Foxp3‐fatemapping mice to examine the cell lineage stability of Treg cells in pregnancy. Ex‐Foxp3 cells were identified in gestational tissues. However, Treg cells retained lineage stability with no increased ex‐Foxp3 generation, regardless of inflammatory challenges that induce preterm birth.

Figures

IL‐17A and CCR6 expression increases in udLN Foxp3+ Treg cells after IL‐1β administration to induce preterm birth. (a) Experimental design. Briefly, pregnant C57Bl/6 dams were administered 6 μg IL‐1β or vehicle control (PBS/0.1% BSA) on GD15.5 to induce preterm birth and flow cytometry of the udLNs was performed on GD16.5 to evaluate the T cell response. (b) Number of CD4+ T cells. (c) Proportion and (d) number of total Foxp3+ Treg cells. (e) Total number of tTreg cells and (f) pTreg cells. (g) Proportion and (h) number of IL‐17A+ Treg cells, and (i) their level of IL‐17A expression. (j) Proportion and (k) number of CCR6+ Treg cells, and (l) their level of CCR6 expression. (m) Proportion and (n) number of Tconv cells. (o) Proportion and (p) number of IFNγ+ Tconv cells (Th1 cells). (q) Proportion and (r) number of IL‐17A+ Tconv cells (Th17 cells). N = 11 per group. *P < 0.05, P < 0.01, *P < 0.005 by unpaired t test.

Ex‐Foxp3 cells are present in the udLNs throughout gestation. (a) Experimental design. The abundance of udLN Treg cells, ex‐Foxp3 cells and Tconv cells was tracked in non‐pregnant (NP) and pregnant Foxp3 fate‐mapping mice across gestation. (b) Gating of the 3 CD4+ T cell subsets based on fluorescent reporter protein expression. (c) Number and (d) proportion of Treg cells (GFP+RFP+). (e) Proportion of tTreg and pTreg cells. (f) Number and (g) proportion of ex‐Foxp3 cells (GFP−RFP+). (h) Proportion of Nrp1+ and Nrp1− cells among ex‐Foxp3 cells. (i) Ratio of ex‐Foxp3 cells:Treg cells. (j) Number and (k) proportion of Tconv cells. N = 4–11 per group. *P < 0.05, P < 0.01, all groups were compared by One‐way ANOVA and Tukey's post‐hoc test.

The ex‐Foxp3 population in udLNs is enriched for cytokine‐expressing cells. The experimental design is outlined in Figure 2. (a) Proportion of IFNγ+ and (b) IL‐17A+ Treg cells. (c) Proportion of IFNγ+ and (d) IL‐17A+ ex‐Foxp3 cells. (e) Proportion of IFNγ+ (Th1) and (f) IL‐17A+ (Th17) Tconv cells. N = 4–11 per group. All groups were compared by one‐way ANOVA and Tukey's post‐hoc test.

A distinct transcriptional program in ex‐Foxp3 cells compared to Treg cells. (a) Experimental design. Ex‐Foxp3 cells and Treg cells were sorted from udLNs of GD18.5 Foxp3 fate‐mapping mice by fluorescence‐activated cell sorting (FACS), then analyzed by RNAseq. (b) Principal‐component analysis showing clustering of samples by cell type. Ovals represent 95% confidence interval of the group mean. (c) Volcano plot showing significantly upregulated and downregulated genes in ex‐Foxp3 cells compared to Treg cells. A log fold‐change (lfc) cut‐off of 1 (equivalent to 2‐fold change in expression) and FDR‐adjusted P‐value <0.05 was used to determine significance. (d) Heatmap of relative expression levels of the top 25 most significantly DE genes in ex‐Foxp3 cells. (e) Comparison of the expression of significantly DE transcription factors, (f) cytokines, (g) granzymes and (h) chemokine receptors in the RNAseq. N = 5 mice pooled per sample, and 5 samples per group. CPM: counts per million.

Treg cells maintain stability after late‐gestation inflammatory challenge to induce preterm birth. (a) Experimental design. Pregnant Foxp3 fate‐mapping mice were administered 4 μg LPS on the morning (0900 h) of GD17.5 or 4 μg IL‐1β on the evening (1900 h) of GD17.5. Controls received vehicle PBS/0.1% BSA. Flow cytometry was performed on the udLNs and decidua on the following morning on GD18.5 to track Treg cell stability by fluorescent reporter protein expression. (b) Total number of udLN and (c) decidual CD4+ T cells. (d) Proportion of Treg cells in the udLNs and (e) decidua. (f) Number of Treg cells in the udLNs and (g) decidua. (h) Proportion of ex‐Foxp3 cells in the udLNs and (i) decidua. (j) Number of ex‐Foxp3 cells in the udLNs and (k) decidua. (l) Proportion of Tconv cells in the udLNs and (m) decidua. (n) Number of Tconv cells in the udLNs and (o) decidua. N = 6–8 per group. The LPS‐ and IL‐1β‐treated groups were compared to the control group by One‐way ANOVA and Dunnett's post‐hoc test.

Expression of cytokines by T cell subsets after late‐gestation inflammatory challenge. The experiment was carried out as shown in Figure 5. (a) Proportion of IFNγ+ Treg cells in the udLNs and (b) uterine decidua. (c) Proportion of IFNγ+ ex‐Foxp3 cells in the udLNs and (d) decidua. (e) Proportion of IFNγ+ Tconv cells in the udLNs and (f) decidua. (g) Proportion of IL‐17A+ Treg cells in the udLNs and (h) decidua. (i) Proportion of IL‐17A+ ex‐Foxp3 cells in the udLNs and (j) decidua. (k) Proportion of IL‐17A+ Tconv cells in the udLNs and (l) decidua. N = 7–8 per group. The LPS‐ and IL‐1β‐treated groups were compared to the control group by One‐way ANOVA and Dunnett's post‐hoc test.

Tables

Table 1. Differentially expressed genes in ex‐Foxp3 cells compared to Treg cells in the udLNs of pregnant Foxp3 fate‐mapper dams at GD18.5, with a logFC >1 and adj.P.Val. <0.05.

No.	Gene	Status	logfc	AveExpr	adj.P.Val	Description	Gene biotype
1	Foxp3	Down	−3.214	8.787	9.49E‐13	Forkhead box P3	protein_coding
2	Il2ra	Down	−3.388	6.811	2.22E‐12	Interleukin 2 receptor, alpha chain	protein_coding
3	Socs2	Down	−2.916	5.439	3.53E‐12	Suppressor of cytokine signaling 2	protein_coding
4	Lrrc32	Down	−3.806	5.199	1.46E‐11	Leucine rich repeat containing 32	protein_coding
5	Cish	Down	−2.246	7.311	2.53E‐11	Cytokine inducible SH2‐containing protein	protein_coding
6	Pde3b	Up	5.120	4.004	2.56E‐11	Phosphodiesterase 3B, cGMP‐inhibited	protein_coding
7	Ecm1	Down	−2.453	7.083	5.56E‐11	Extracellular matrix protein 1	protein_coding
8	Gpr83	Down	−3.012	7.267	1.28E‐10	G protein‐coupled receptor 83	protein_coding
9	Cd40lg	Up	2.038	7.162	3.17E‐10	CD40 ligand	protein_coding
10	Ntrk3	Up	4.037	3.233	3.93E‐10	Neurotrophic tyrosine kinase, receptor, type 3	protein_coding
11	Flicr	Down	−3.386	3.794	6.22E‐10	Foxp3 regulating long intergenic noncoding RNA	lncRNA
12	Gsto1	Down	−2.671	4.868	1.97E‐09	Glutathione S‐transferase omega 1	protein_coding
13	Dennd2d	Up	2.220	7.853	2.02E‐09	DENN domain containing 2D	protein_coding
14	Gata1	Down	−2.914	3.866	3.38E‐09	GATA binding protein 1	protein_coding
15	Gsta4	Down	−3.705	3.608	3.38E‐09	Glutathione S‐transferase, alpha 4	protein_coding
16	Igfbp4	Up	2.244	7.333	3.96E‐09	Insulin‐like growth factor binding protein 4	protein_coding
17	Themis	Up	2.480	5.981	5.12E‐09	Thymocyte selection associated	protein_coding
18	Il1rl2	Up	4.028	2.340	1.71E‐08	Interleukin 1 receptor‐like 2	protein_coding
19	Rflnb	Up	2.126	6.336	1.95E‐08	Refilin B	protein_coding
20	Lpar3	Up	7.441	−2.380	2.13E‐08	Lysophosphatidic acid receptor 3	protein_coding
21	Ehd3	Up	2.154	7.135	2.13E‐08	EH‐domain containing 3	protein_coding
22	Ikzf4	Down	−2.128	6.220	2.13E‐08	IKAROS family zinc finger 4	protein_coding
23	Appl2	Up	2.382	4.517	2.13E‐08	Adaptor protein, phosphotyrosine interaction, PH domain and leucine zipper containing 2	protein_coding
24	Vipr1	Up	4.579	3.500	2.17E‐08	Vasoactive intestinal peptide receptor 1	protein_coding
25	Ggt5	Up	6.024	1.090	5.12E‐08	Gamma‐glutamyltransferase 5	protein_coding
26	Cyp4f18	Down	−5.252	3.365	9.53E‐08	Cytochrome P450, family 4, subfamily f, polypeptide 18	protein_coding
27	Dapl1	Up	4.247	4.006	9.69E‐08	Death associated protein‐like 1	protein_coding
28	Tcf7	Up	1.559	10.203	1.02E‐07	Transcription factor 7, T cell specific	protein_coding
29	Tgfbr3	Up	2.767	4.352	1.02E‐07	Transforming growth factor, beta receptor III	protein_coding
30	Tnfsf8	Up	2.006	7.431	1.31E‐07	Tumor necrosis factor (ligand) superfamily, member 8	protein_coding
31	Tdrp	Up	2.422	5.091	1.42E‐07	Testis development related protein	protein_coding
32	Gm39760	Down	−3.921	0.906	1.43E‐07	NA	NA
33	App	Up	3.793	2.714	1.84E‐07	Amyloid beta (A4) precursor protein	protein_coding
34	Rgs1	Down	−2.059	7.885	2.54E‐07	Regulator of G‐protein signaling 1	protein_coding
35	Capn3	Down	−2.227	5.315	3.50E‐07	Calpain 3	protein_coding
36	Sema4a	Up	2.156	5.733	3.50E‐07	Sema domain, immunoglobulin domain (Ig), transmembrane domain (TM) and short cytoplasmic domain, (semaphorin) 4A	protein_coding
37	Il4	Up	4.731	1.649	3.71E‐07	Interleukin 4	protein_coding
38	Spp1	Up	6.053	2.660	5.86E‐07	Secreted phosphoprotein 1	protein_coding
39	Vdr	Up	4.695	2.496	5.87E‐07	Vitamin D (1,25‐dihydroxyvitamin D3) receptor	protein_coding
40	Ifngr1	Down	−1.605	8.523	6.41E‐07	Interferon gamma receptor 1	protein_coding
41	Rnf32	Up	2.960	2.097	6.67E‐07	Ring finger protein 32	protein_coding
42	Itgae	Down	−2.858	4.586	6.67E‐07	Integrin alpha E, epithelial‐associated	protein_coding
43	Hs3st3b1	Up	3.492	2.720	6.67E‐07	Heparan sulfate (glucosamine) 3‐O‐sulfotransferase 3B1	protein_coding
44	Padi4	Up	4.643	1.930	6.67E‐07	Peptidyl arginine deiminase, type IV	protein_coding
45	1810009J06Rik	Up	7.314	−1.974	6.81E‐07	RIKEN cDNA 1 810 009 J06 gene	protein_coding
46	Atp8b4	Up	4.941	1.517	6.91E‐07	ATPase, class I, type 8B, member 4	protein_coding
47	Gm3521	Down	−2.651	2.472	7.75E‐07	NA	NA
48	Tmem121b	Down	−3.094	2.023	7.81E‐07	Transmembrane protein 121B	protein_coding
49	Gm4956	Down	−2.894	2.984	8.61E‐07	Predicted gene 4956	lncRNA
50	Mboat1	Up	2.097	4.104	1.10E‐06	Membrane bound O‐acyltransferase domain containing 1	protein_coding
51	Angptl2	Up	2.507	6.496	1.10E‐06	Angiopoietin‐like 2	protein_coding
52	Smtn	Up	4.608	2.516	1.14E‐06	Smoothelin	protein_coding
53	Gm32474	Down	−3.630	1.072	1.22E‐06	Predicted gene, 32474	lncRNA
54	Apol7e	Up	1.694	5.395	1.22E‐06	Apolipoprotein L 7e	protein_coding
55	Rragd	Down	−2.351	3.605	1.22E‐06	Ras‐related GTP binding D	protein_coding
56	Gm15915	Down	−3.592	1.180	1.26E‐06	Predicted gene 15915	lncRNA
57	Galnt9	Up	2.107	3.843	1.26E‐06	Polypeptide N‐acetylgalactosaminyltransferase 9	protein_coding
58	Syne2	Up	1.913	4.610	1.26E‐06	Spectrin repeat containing, nuclear envelope 2	protein_coding
59	Cxcr6	Up	3.958	4.452	1.26E‐06	Chemokine (C‐X‐C motif) receptor 6	protein_coding
60	Bmyc	Down	−1.807	6.753	1.26E‐06	Brain expressed myelocytomatosis oncogene	protein_coding
61	Sgk1	Up	2.070	5.766	1.26E‐06	Serum/glucocorticoid regulated kinase 1	protein_coding
62	Slc22a2	Down	−2.686	2.892	1.64E‐06	Solute carrier family 22 (organic cation transporter), member 2	protein_coding
63	Pdlim4	Up	5.050	1.371	1.81E‐06	PDZ and LIM domain 4	protein_coding
64	Ly6c1	Down	−2.132	5.383	2.26E‐06	Lymphocyte antigen 6 complex, locus C1	protein_coding
65	Slc26a11	Up	1.831	5.690	2.48E‐06	Solute carrier family 26, member 11	protein_coding
66	Stra6	Down	−5.187	−0.055	2.52E‐06	Stimulated by retinoic acid gene 6	protein_coding
67	Mfsd6	Up	1.628	5.815	2.52E‐06	Major facilitator superfamily domain containing 6	protein_coding
68	Mctp2	Up	3.636	1.929	2.63E‐06	Multiple C2 domains, transmembrane 2	protein_coding
69	Tnfrsf18	Down	−1.719	8.391	3.29E‐06	Tumor necrosis factor receptor superfamily, member 18	protein_coding
70	Hbegf	Down	−4.681	−1.560	3.40E‐06	Heparin‐binding EGF‐like growth factor	protein_coding
71	Gpr34	Down	−1.853	4.833	3.44E‐06	G protein‐coupled receptor 34	protein_coding
72	Edn3	Up	6.320	0.331	3.70E‐06	Endothelin 3	protein_coding
73	Il2	Up	5.845	−0.579	3.77E‐06	Interleukin 2	protein_coding
74	Klhl4	Up	3.465	0.796	3.86E‐06	Kelch‐like 4	protein_coding
75	Sdc4	Down	−3.109	2.915	4.48E‐06	Syndecan 4	protein_coding
76	Hopx	Down	−1.611	6.681	5.72E‐06	HOP homeobox	protein_coding
77	Nme4	Up	2.768	1.957	6.27E‐06	NME/NM23 nucleoside diphosphate kinase 4	protein_coding
78	Il21	Up	7.862	1.878	6.27E‐06	Interleukin 21	protein_coding
79	Gm14718	Up	2.509	2.826	8.07E‐06	Predicted gene 14718	lncRNA
80	Gm33561	Up	7.544	−1.758	8.10E‐06	NA	NA
81	Gm32020	Up	3.975	1.737	8.10E‐06	NA	NA
82	Pde7a	Up	1.619	5.910	8.20E‐06	Phosphodiesterase 7A	protein_coding
83	Drc1	Down	−3.943	0.155	8.36E‐06	Dynein regulatory complex subunit 1	protein_coding
84	Tnfrsf4	Down	−2.024	9.073	8.53E‐06	Tumor necrosis factor receptor superfamily, member 4	protein_coding
85	1110032F04Rik	Down	−1.703	5.450	8.74E‐06	RIKEN cDNA 1110032F04 gene	protein_coding
86	Dock4	Up	3.029	1.737	9.63E‐06	Dedicator of cytokinesis 4	protein_coding
87	Rapgef4	Up	2.295	3.672	1.01E‐05	Rap guanine nucleotide exchange factor (GEF) 4	protein_coding
88	Chn2	Up	1.808	4.430	1.02E‐05	Chimerin 2	protein_coding
89	Lratd1	Up	5.505	−0.385	1.02E‐05	LRAT domain containing 1	protein_coding
90	Gm2663	Up	4.700	1.187	1.09E‐05	Predicted gene 2663	protein_coding
91	H2‐Q2	Up	2.491	3.369	1.09E‐05	Histocompatibility 2, Q region locus 2	protein_coding
92	Sbk2	Up	4.537	−1.544	1.11E‐05	SH3‐binding domain kinase family, member 2	protein_coding
93	Id2	Up	1.643	7.754	1.13E‐05	Inhibitor of DNA binding 2	protein_coding
94	Lrrc75b	Up	4.187	0.021	1.22E‐05	Leucine rich repeat containing 75B	protein_coding
95	Inafm2	Up	1.787	4.508	1.23E‐05	InaF motif containing 2	protein_coding
96	Abhd15	Up	3.884	1.005	1.23E‐05	Abhydrolase domain containing 15	protein_coding
97	Gbp2b	Down	−3.246	2.529	1.40E‐05	Guanylate binding protein 2b	protein_coding
98	Acp3	Up	2.264	2.455	1.40E‐05	NA	NA
99	Gpm6b	Up	2.703	6.115	1.40E‐05	Glycoprotein m6b	protein_coding
100	Tgfb3	Up	3.986	−0.244	1.40E‐05	Transforming growth factor, beta 3	protein_coding
101	Gm32569	Down	−3.371	1.560	1.49E‐05	Predicted gene, 32569	lncRNA
102	Klrg1	Down	−3.107	4.112	1.64E‐05	Killer cell lectin‐like receptor subfamily G, member 1	protein_coding
103	8030451A03Rik	Up	6.551	−1.496	1.70E‐05	RIKEN cDNA 8030451A03 gene	lncRNA
104	Ppm1l	Down	−2.209	3.036	1.74E‐05	Protein phosphatase 1 (formerly 2C)‐like	protein_coding
105	Epcam	Down	−2.191	3.971	2.02E‐05	Epithelial cell adhesion molecule	protein_coding
106	Rcn1	Down	−1.535	6.409	2.05E‐05	Reticulocalbin 1	protein_coding
107	Vat1l	Up	5.670	−2.235	2.11E‐05	Vesicle amine transport protein 1 like	protein_coding
108	Lad1	Down	−2.262	4.076	2.21E‐05	Ladinin	protein_coding
109	Arhgap29	Up	2.730	1.816	2.24E‐05	Rho GTPase activating protein 29	protein_coding
110	P2ry1	Up	5.134	−0.734	2.50E‐05	Purinergic receptor P2Y, G‐protein coupled 1	protein_coding
111	A530021J07Rik	Up	4.931	0.945	2.59E‐05	Riken cDNA A530021J07 gene	lncRNA
112	Dgat2	Down	−2.792	1.907	2.62E‐05	Diacylglycerol O‐acyltransferase 2	protein_coding
113	Atp6v0d2	Down	−3.159	2.027	2.86E‐05	ATPase, H+ transporting, lysosomal V0 subunit D2	protein_coding
114	Gucy2g	Up	3.316	0.881	2.86E‐05	Guanylate cyclase 2 g	protein_coding
115	Tox2	Up	2.324	6.155	3.16E‐05	TOX high mobility group box family member 2	protein_coding
116	Pou2af1	Up	2.831	4.603	3.27E‐05	POU domain, class 2, associating factor 1	protein_coding
117	1700019D03Rik	Up	2.532	3.866	3.30E‐05	RIKEN cDNA 1700019D03 gene	protein_coding
118	Rgcc	Up	1.648	5.164	3.32E‐05	Regulator of cell cycle	protein_coding
119	Ccr6	Down	−1.977	6.541	3.36E‐05	Chemokine (C‐C motif) receptor 6	protein_coding
120	Tnfrsf9	Down	−1.653	8.273	3.55E‐05	Tumor necrosis factor receptor superfamily, member 9	protein_coding
121	Arhgap20	Down	−2.305	1.982	4.01E‐05	Rho GTPase activating protein 20	protein_coding
122	Gm39612	Down	−5.840	−1.441	4.01E‐05	NA	NA
123	Il1r1	Up	3.283	0.698	4.06E‐05	Interleukin 1 receptor, type I	protein_coding
124	Cyp2s1	Up	2.164	2.954	4.69E‐05	Cytochrome P450, family 2, subfamily s, polypeptide 1	protein_coding
125	Gm53265	Up	4.438	0.647	5.01E‐05	NA	NA
126	Gm45968	Down	−5.188	−2.915	5.16E‐05	NA	NA
127	Lcn10	Down	−5.485	−2.345	5.30E‐05	Lipocalin 10	protein_coding
128	Gm53531	Up	2.927	1.275	5.40E‐05	NA	NA
129	Qrfp	Up	4.469	−0.393	5.56E‐05	Pyroglutamylated RFamide peptide	protein_coding
130	Gm30318	Up	4.172	−1.005	5.61E‐05	NA	NA
131	Gm45552	Up	1.753	5.216	5.67E‐05	Predicted gene 45552	lncRNA
132	Gm34828	Up	2.763	2.196	6.12E‐05	NA	NA
133	Peli2	Down	−3.539	−0.316	6.45E‐05	Pellino 2	protein_coding
134	Fasl	Up	1.597	4.691	6.46E‐05	Fas ligand (TNF superfamily, member 6)	protein_coding
135	Gm32803	Down	−1.953	3.246	6.46E‐05	Predicted gene, 32803	lncRNA
136	Wfdc2	Down	−5.363	−2.832	7.11E‐05	WAP four‐disulfide core domain 2	protein_coding
137	Patl2	Down	−3.027	0.253	7.62E‐05	Protein associated with topoisomerase II homolog 2 (yeast)	protein_coding
138	Ppic	Up	1.708	4.720	8.67E‐05	Peptidylprolyl isomerase C	protein_coding
139	Ccr8	Down	−1.814	6.503	9.52E‐05	Chemokine (C‐C motif) receptor 8	protein_coding
140	Spin2c	Down	−1.865	4.024	9.52E‐05	Spindlin family, member 2C	protein_coding
141	Fbn2	Up	4.565	−0.150	9.79E‐05	Fibrillin 2	protein_coding
142	Itga7	Up	2.165	3.083	0.00010	Integrin alpha 7	protein_coding
143	Plscr4	Up	3.234	1.209	0.00010	Phospholipid scramblase 4	protein_coding
144	Ccs	Down	−1.582	6.544	0.00010	Copper chaperone for superoxide dismutase	protein_coding
145	Crtam	Up	2.593	2.251	0.00011	Cytotoxic and regulatory T cell molecule	protein_coding
146	Apol9b	Down	−1.774	5.070	0.00012	Apolipoprotein L 9b	protein_coding
147	Pdzrn3	Up	4.734	−1.920	0.00012	PDZ domain containing RING finger 3	protein_coding
148	Malt1	Up	1.500	6.764	0.00012	MALT1 paracaspase	protein_coding
149	Gm14009	Down	−4.828	−2.636	0.00012	Predicted gene 14009	lncRNA
150	Sema4f	Up	1.599	4.329	0.00012	Sema domain, immunoglobulin domain (Ig), TM domain and short cytoplasmic domain	protein_coding
151	Cd109	Up	6.049	−0.490	0.00012	CD109 antigen	protein_coding
152	Evi2a	Up	1.558	5.319	0.00012	Ecotropic viral integration site 2a	protein_coding
153	Gm40372	Up	4.883	−1.038	0.00013	Predicted gene, 40372	lncRNA
154	Flrt2	Up	5.531	−1.741	0.00015	Fibronectin leucine rich transmembrane protein 2	protein_coding
155	Rad51b	Down	−2.018	3.165	0.00017	RAD51 paralog B	protein_coding
156	Dse	Up	2.113	3.206	0.00017	Dermatan sulfate epimerase	protein_coding
157	St3gal6	Up	2.294	2.568	0.00017	ST3 beta‐galactoside alpha‐2,3‐sialyltransferase 6	protein_coding
158	Socs1	Down	−1.471	6.400	0.00017	Suppressor of cytokine signaling 1	protein_coding
159	Slc16a5	Up	2.287	2.835	0.00018	Solute carrier family 16 (monocarboxylic acid transporters), member 5	protein_coding
160	Cacna1d	Up	3.663	1.500	0.00019	Calcium channel, voltage‐dependent, L type, alpha 1D subunit	protein_coding
161	H2az1	Down	−1.413	9.058	0.00020	H2A.Z variant histone 1	protein_coding
162	S1pr3	Up	5.002	−2.302	0.00021	Sphingosine‐1‐phosphate receptor 3	protein_coding
163	Rnf144a	Up	1.865	3.600	0.00022	Ring finger protein 144A	protein_coding
164	Medag	Down	−2.685	0.861	0.00023	Mesenteric estrogen dependent adipogenesis	protein_coding
165	Slc30a3	Up	4.696	−1.947	0.00023	Solute carrier family 30 (zinc transporter), member 3	protein_coding
166	Src	Down	−2.396	1.476	0.00023	Rous sarcoma oncogene	protein_coding
167	Stfa3	Up	4.531	−0.414	0.00023	Stefin A3	protein_coding
168	Ramp3	Up	1.844	3.096	0.00023	Receptor (calcitonin) activity modifying protein 3	protein_coding
169	Itih5	Down	−1.787	4.712	0.00024	Inter‐alpha‐trypsin inhibitor, heavy chain 5	protein_coding
170	Ly6k	Down	−1.675	4.450	0.00024	Lymphocyte antigen 6 complex, locus K	protein_coding
171	Lypd6b	Up	3.435	1.998	0.00024	LY6/PLAUR domain containing 6B	protein_coding
172	Ptger3	Down	−4.387	−1.463	0.00025	Prostaglandin E receptor 3 (subtype EP3)	protein_coding
173	Rnf19a	Up	1.400	7.351	0.00026	Ring finger protein 19A	protein_coding
174	Plekho1	Up	2.001	3.156	0.00027	Pleckstrin homology domain containing, family O member 1	protein_coding
175	Tnfsf4	Up	2.541	1.263	0.00027	Tumor necrosis factor (ligand) superfamily, member 4	protein_coding
176	Gm10640	Up	2.840	−0.440	0.00028	Predicted gene 10640	lncRNA
177	Klrb1a	Up	3.667	1.715	0.00028	Killer cell lectin‐like receptor subfamily B member 1A	protein_coding
178	Gm6637	Down	−1.992	5.311	0.00029	Predicted gene 6637	processed_pseudogene
179	Trpm6	Up	3.045	0.298	0.00032	Transient receptor potential cation channel, subfamily M, member 6	protein_coding
180	Anxa3	Up	4.053	0.374	0.00036	Annexin A3	protein_coding
181	Adh1	Up	2.257	2.606	0.00036	Alcohol dehydrogenase 1 (class I)	protein_coding
182	Chchd10	Down	−1.671	5.732	0.00036	Coiled‐coil‐helix‐coiled‐coil‐helix domain containing 10	protein_coding
183	Gpr153	Up	3.454	−0.391	0.00037	G protein‐coupled receptor 153	protein_coding
184	Ryr1	Up	3.325	0.324	0.00037	Ryanodine receptor 1, skeletal muscle	protein_coding
185	Hid1	Up	1.838	4.055	0.00040	HID1 domain containing	protein_coding
186	Rassf10	Up	4.412	−2.148	0.00040	Ras association (RalGDS/AF‐6) domain family (N‐terminal) member 10	protein_coding
187	Lancl3	Up	2.669	0.759	0.00043	LanC lantibiotic synthetase component C‐like 3 (bacterial)	protein_coding
188	Cnga1	Up	2.231	1.514	0.00044	Cyclic nucleotide gated channel alpha 1	protein_coding
189	Ascl2	Up	3.660	2.433	0.00044	Achaete‐scute family bHLH transcription factor 2	protein_coding
190	Chn1	Up	5.108	0.360	0.00047	chimerin 1	protein_coding
191	Ankrd55	Down	−1.830	2.757	0.00048	Ankyrin repeat domain 55	protein_coding
192	Col24a1	Up	4.098	−1.323	0.00048	Collagen, type XXIV, alpha 1	protein_coding
193	Art2a	Up	3.206	4.649	0.00048	ADP‐ribosyltransferase 2a	protein_coding
194	Nacc2	Up	2.112	1.791	0.00051	Nucleus accumbens associated 2, BEN and BTB (POZ) domain containing	protein_coding
195	Nrgn	Up	1.984	2.949	0.00051	Neurogranin	protein_coding
196	Pip5k1b	Down	−1.976	2.128	0.00054	Phosphatidylinositol‐4‐phosphate 5‐kinase, type 1 beta	protein_coding
197	Scart2	Up	4.603	0.735	0.00060	Scavenger receptor family member expressed on T cells 2	protein_coding
198	Gal	Up	4.700	1.022	0.00063	Galanin and GMAP prepropeptide	protein_coding
199	Ifitm3	Down	−2.915	2.200	0.00069	Interferon induced transmembrane protein 3	protein_coding
200	Gm51661	Up	2.616	0.968	0.00069	NA	NA
201	Nanos1	Up	2.006	3.860	0.00076	Nanos C2HC‐type zinc finger 1	protein_coding
202	Sostdc1	Up	6.212	4.208	0.00079	Sclerostin domain containing 1	protein_coding
203	Tmem158	Down	−1.622	3.957	0.00086	Transmembrane protein 158	protein_coding
204	Kcnj15	Down	−4.832	−2.232	0.00086	Potassium inwardly rectifying channel, subfamily J, member 15	protein_coding
205	Sh3rf2	Up	4.343	−1.778	0.00086	SH3 domain containing ring finger 2	protein_coding
206	Adamts3	Up	2.978	−0.483	0.00087	A disintegrin‐like and metallopeptidase (reprolysin type) with thrombospondin type 1 motif, 3	protein_coding
207	Prex1	Up	1.302	6.538	0.00088	Phosphatidylinositol‐3,4,5‐trisphosphate‐dependent Rac exchange factor 1	protein_coding
208	Rgs11	Up	1.512	4.007	0.00089	Regulator of G‐protein signaling 11	protein_coding
209	Ifitm10	Up	2.130	2.329	0.00093	Interferon induced transmembrane protein 10	protein_coding
210	Islr	Down	−4.393	−1.034	0.00097	Immunoglobulin superfamily containing leucine‐rich repeat	protein_coding
211	Klhl5	Up	1.561	4.392	0.00097	Kelch‐like 5	protein_coding
212	Ano4	Up	4.310	−2.357	0.00100	Anoctamin 4	protein_coding
213	Ankrd35	Up	4.778	−2.034	0.00100	Ankyrin repeat domain 35	protein_coding
214	Fez1	Up	3.370	−1.707	0.00104	Fasciculation and elongation protein zeta 1 (zygin I)	protein_coding
215	Ntn4	Down	−1.683	3.924	0.00104	Netrin 4	protein_coding
216	Serpina9	Up	3.093	2.726	0.00105	Serine (or cysteine) peptidase inhibitor, clade A (alpha‐1 antiproteinase, antitrypsin), member 9	protein_coding
217	Mtus1	Up	3.096	0.610	0.00105	Mitochondrial tumor suppressor 1	protein_coding
218	Rln3	Down	−2.105	4.270	0.00109	Relaxin 3	protein_coding
219	Syp	Down	−1.425	4.559	0.00113	Synaptophysin	protein_coding
220	Neb	Down	−1.854	3.800	0.00117	Nebulin	protein_coding
221	Scn8a	Up	3.212	−0.113	0.00121	Sodium channel, voltage‐gated, type VIII, alpha	protein_coding
222	Gm30948	Down	−4.111	−2.347	0.00121	Predicted gene, 30948	lncRNA
223	Akap12	Up	1.475	4.760	0.00121	A kinase (PRKA) anchor protein (gravin) 12	protein_coding
224	Pcp4	Up	2.018	3.189	0.00126	Purkinje cell protein 4	protein_coding
225	Pde5a	Up	3.505	−0.442	0.00136	Phosphodiesterase 5A, cGMP‐specific	protein_coding
226	Itsn1	Down	−2.098	1.952	0.00145	Intersectin 1 (SH3 domain protein 1A)	protein_coding
227	Gm11730	Up	3.620	−2.023	0.00149	Predicted gene 11730	lncRNA
228	Gzmb	Down	−2.093	2.195	0.00157	Granzyme B	protein_coding
229	Pim1	Down	−1.480	7.223	0.00158	Proviral integration site 1	protein_coding
230	Capg	Down	−1.375	8.741	0.00164	Capping protein (Actin filament), gelsolin‐like	protein_coding
231	Atp2b2	Up	2.790	−0.163	0.00170	ATPase, Ca++ transporting, plasma membrane 2	protein_coding
232	Atp1a3	Up	2.881	3.441	0.00171	ATPase, Na+/K+ transporting, alpha 3 polypeptide	protein_coding
233	Cpeb1	Up	3.623	−1.485	0.00181	Cytoplasmic polyadenylation element binding protein 1	protein_coding
234	Aldh1a2	Down	−3.773	−1.611	0.00181	Aldehyde dehydrogenase family 1, subfamily A2	protein_coding
235	Cxcr5	Up	2.105	3.765	0.00181	CXXC finger 5	protein_coding
236	Actr3b	Up	2.035	1.995	0.00185	ARP3 Actin‐related protein 3B	protein_coding
237	Tnc	Up	4.189	−0.548	0.00186	Tenascin C	protein_coding
238	Elmod1	Up	3.810	−2.544	0.00186	ELMO/CED‐12 domain containing 1	protein_coding
239	Kcnc2	Up	3.232	−1.918	0.00186	Potassium voltage gated channel, Shaw‐related subfamily, member 2	protein_coding
240	Itk	Up	1.383	8.483	0.00201	IL2 inducible T cell kinase	protein_coding
241	Pgpep1l	Down	−1.793	1.828	0.00201	Pyroglutamyl‐peptidase I‐like	protein_coding
242	Cdkn2b	Up	3.083	0.417	0.00202	Cyclin dependent kinase inhibitor 2B	protein_coding
243	Gramd1a	Up	1.322	8.978	0.00209	GRAM domain containing 1A	protein_coding
244	Ica1l	Up	2.523	1.574	0.00211	Islet cell autoantigen 1‐like	protein_coding
245	Kirrel3	Up	3.442	1.309	0.00211	Kirre like nephrin family adhesion molecule 3	protein_coding
246	Sst	Up	2.274	1.777	0.00212	Somatostatin	protein_coding
247	Rgmb	Up	1.498	3.588	0.00216	Repulsive guidance molecule family member B	protein_coding
248	Shtn1	Up	2.497	1.773	0.00217	Shootin 1	protein_coding
249	Prss12	Up	1.631	3.137	0.00218	Protease, serine 12 neurotrypsin (motopsin)	protein_coding
250	Tnfsf14	Up	1.515	3.903	0.00219	Tumor necrosis factor (ligand) superfamily, member 14	protein_coding
251	Rasip1	Down	−2.913	−1.017	0.00219	Ras interacting protein 1	protein_coding
252	Acsbg1	Down	−1.397	5.577	0.00224	Acyl‐CoA synthetase bubblegum family member 1	protein_coding
253	Ifng	Up	2.423	3.999	0.00227	Interferon gamma	protein_coding
254	Snhg11	Up	2.489	0.504	0.00237	Small nucleolar RNA host gene 11	protein_coding
255	Efnb2	Up	3.236	−0.770	0.00276	Ephrin B2	protein_coding
256	C030034L19Rik	Up	2.415	2.468	0.00288	RIKEN cDNA C030034L19 gene	lncRNA
257	Vwde	Up	5.564	−1.055	0.00288	von Willebrand factor D and EGF domains	protein_coding
258	Fam20a	Up	3.050	2.086	0.00299	FAM20A, golgi associated secretory pathway pseudokinase	protein_coding
259	Thbs1	Up	2.906	−0.594	0.00300	Thrombospondin 1	protein_coding
260	Ptgds	Up	2.036	2.811	0.00307	Prostaglandin D2 synthase (brain)	protein_coding
261	Emp1	Up	1.964	2.782	0.00334	Epithelial membrane protein 1	protein_coding
262	Ldlrad3	Up	2.755	0.637	0.00351	Low density lipoprotein receptor class A domain containing 3	protein_coding
263	Slc12a7	Up	1.380	6.607	0.00373	Solute carrier family 12, member 7	protein_coding
264	Fgf2	Up	2.299	1.401	0.00407	Fibroblast growth factor 2	protein_coding
265	Lhfpl1	Down	−2.366	0.237	0.00407	Lipoma HMGIC fusion partner‐like 1	protein_coding
266	Stc2	Up	4.085	0.483	0.00407	Stanniocalcin 2	protein_coding
267	St6galnac2	Up	2.512	4.033	0.00420	ST6 (alpha‐N‐acetyl‐neuraminyl‐2,3‐beta‐galactosyl‐1,3)‐N‐acetylgalactosaminide alpha‐2,6‐sialyltransferase 2	protein_coding
268	St6galnac1	Up	2.545	1.652	0.00426	ST6 (alpha‐N‐acetyl‐neuraminyl‐2,3‐beta‐galactosyl‐1,3)‐N‐acetylgalactosaminide alpha‐2,6‐sialyltransferase 1	protein_coding
269	Eea1	Up	1.372	6.304	0.00434	Early endosome antigen 1	protein_coding
270	Kirrel3os	Up	4.560	1.926	0.00444	Kirre like nephrin family adhesion molecule 3, opposite strand	lncRNA
271	Qpct	Up	2.375	1.391	0.00445	Glutaminyl‐peptide cyclotransferase (glutaminyl cyclase)	protein_coding
272	Nrbp2	Up	2.141	3.594	0.00446	Nuclear receptor binding protein 2	protein_coding
273	Slc9a9	Up	1.361	6.526	0.00449	Solute carrier family 9 (sodium/hydrogen exchanger), member 9	protein_coding
274	Raver2	Up	1.951	1.555	0.00449	Ribonucleoprotein, PTB‐binding 2	protein_coding
275	Utf1	Down	−1.874	1.728	0.00452	Undifferentiated embryonic cell transcription factor 1	protein_coding
276	Ston1	Down	−1.813	1.906	0.00483	Stonin 1	protein_coding
277	Gm6213	Up	2.453	−0.312	0.00502	Predicted gene 6213	lncRNA
278	Ust	Up	1.734	2.237	0.00507	Uronyl‐2‐sulfotransferase	protein_coding
279	Xkrx	Up	1.901	3.070	0.00507	X‐linked Kx blood group related, X‐linked	protein_coding
280	Tmem26	Up	2.562	0.471	0.00519	Transmembrane protein 26	protein_coding
281	Cpne7	Up	2.545	2.882	0.00521	Copine VII	protein_coding
282	Pld2	Up	1.664	2.905	0.00541	Phospholipase D2	protein_coding
283	Als2cl	Up	1.465	6.216	0.00623	ALS2 C‐terminal like	protein_coding
284	Abtb3	Up	1.409	4.689	0.00668	Ankyrin repeat and BTB domain containing 3	protein_coding
285	Trim36	Up	2.087	1.779	0.00772	Tripartite motif‐containing 36	protein_coding
286	Lrrn2	Up	3.180	−1.549	0.00788	Leucine rich repeat protein 2, neuronal	protein_coding
287	Tgfbr2	Up	1.297	7.811	0.00795	Transforming growth factor, beta receptor II	protein_coding
288	Ceacam‐ps1	Down	−2.162	0.638	0.00797	CEA cell adhesion molecule, pseudogene 1	transcribed_unprocessed_pseudogene
289	6430571L13Rik	Down	−2.640	0.158	0.00805	RIKEN cDNA 6 430 571 L13 gene	protein_coding
290	Gm51535	Down	−2.859	−0.406	0.00829	NA	NA
291	Cd79b	Down	−1.686	5.330	0.00838	CD79B antigen	protein_coding
292	Fhad1	Up	2.246	0.019	0.00838	Forkhead‐associated (FHA) phosphopeptide binding domain 1	protein_coding
293	Csrp2	Down	−2.007	1.025	0.00838	Cysteine and glycine‐rich protein 2	protein_coding
294	Klk15	Up	2.232	0.136	0.00866	Kallikrein related‐peptidase 15	protein_coding
295	Ncmap	Down	−1.576	5.409	0.00904	Noncompact myelin associated protein	protein_coding
296	Immp2l	Down	−1.469	3.257	0.00905	IMP2 inner mitochondrial membrane peptidase‐like (S. cerevisiae)	protein_coding
297	Ggt1	Up	2.634	0.569	0.00905	Gamma‐glutamyltransferase 1	protein_coding
298	Fgf16	Down	−1.958	1.307	0.00921	Fibroblast growth factor 16	protein_coding
299	Garin3	Up	3.013	−0.364	0.00930	Golgi associated RAB2 interactor 3	protein_coding
300	Cc2d2b	Down	−2.531	0.539	0.00968	Coiled‐coil and C2 domain containing 2B	protein_coding
301	Tmcc3	Up	1.634	2.023	0.00972	Transmembrane and coiled coil domains 3	protein_coding
302	Mt3	Up	1.996	1.531	0.00977	Metallothionein 3	protein_coding
303	Srxn1	Up	2.029	2.628	0.00977	Sulfiredoxin 1 homolog (S. cerevisiae)	protein_coding
304	Itgb5	Up	1.993	1.811	0.01002	Integrin beta 5	protein_coding
305	Prss50	Up	4.085	−0.753	0.01012	Protease, serine 50	protein_coding
306	Prodh	Down	−1.721	2.570	0.01012	Proline dehydrogenase	protein_coding
307	Ebi3	Down	−1.578	3.658	0.01012	Epstein–Barr virus induced gene 3	protein_coding
308	Myh3	Up	2.974	−0.012	0.01026	Myosin, heavy polypeptide 3, skeletal muscle, embryonic	protein_coding
309	Nap1l5	Up	2.815	−0.790	0.01035	Nucleosome assembly protein 1‐like 5	protein_coding
310	Il17re	Up	3.810	−0.941	0.01038	Interleukin 17 receptor E	protein_coding
311	Mmp25	Up	2.440	0.223	0.01100	Matrix metallopeptidase 25	protein_coding
312	Fsd1l	Down	−2.939	0.271	0.01125	Fibronectin type III and SPRY domain containing 1‐like	protein_coding
313	Meg3	Up	2.256	0.293	0.01125	Maternally expressed 3	lncRNA
314	Cemip2	Up	1.556	4.048	0.01149	Cell migration inducing hyaluronidase 2	protein_coding
315	Swap70	Down	−1.502	6.087	0.01160	SWA‐70 protein	protein_coding
316	Nav2	Up	1.774	3.706	0.01160	Neuron navigator 2	protein_coding
317	Noxa1	Down	−1.586	2.822	0.01231	NADPH oxidase activator 1	protein_coding
318	Gbp3	Down	−1.424	5.834	0.01235	Guanylate binding protein 3	protein_coding
319	Adgrg1	Down	−2.318	0.499	0.01260	Adhesion G protein‐coupled receptor G1	protein_coding
320	Acvr1c	Down	−3.317	−1.981	0.01263	Activin A receptor, type IC	protein_coding
321	Snap25	Up	2.069	0.805	0.01263	Synaptosomal‐associated protein 25	protein_coding
322	Col5a3	Up	2.830	−0.994	0.01266	Collagen, type V, alpha 3	protein_coding
323	Zeb2	Up	2.591	0.179	0.01267	Zinc finger E‐box binding homeobox 2	protein_coding
324	Plek	Up	1.809	3.245	0.01281	Pleckstrin	protein_coding
325	Ncf1	Down	−1.672	2.751	0.01282	Neutrophil cytosolic factor 1	protein_coding
326	C1ql3	Down	−2.654	−1.297	0.01282	C1q‐like 3	protein_coding
327	Rhod	Down	−1.611	2.506	0.01297	Ras homolog family member D	protein_coding
328	Fam240a	Up	2.424	−0.761	0.01297	Family with sequence similarity 240 member A	protein_coding
329	Sparc	Up	1.983	0.306	0.01298	Secreted acidic cysteine rich glycoprotein	protein_coding
330	Crppa	Up	2.777	2.600	0.01310	CDP‐L‐ribitol pyrophosphorylase A	protein_coding
331	Tcf7l2	Down	−2.456	−1.321	0.01322	Transcription factor 7 like 2, T cell specific, HMG box	protein_coding
332	Extl1	Down	−1.877	0.783	0.01328	Exostosin‐like glycosyltransferase 1	protein_coding
333	Rnf128	Up	1.743	4.273	0.01380	Ring finger protein 128	protein_coding
334	Cebpd	Up	3.233	−1.435	0.01421	CCAAT/enhancer binding protein (C/EBP), delta	protein_coding
335	Cdhr3	Down	−2.995	−1.282	0.01424	Cadherin‐related family member 3	protein_coding
336	Cbfa2t3	Up	2.401	−0.363	0.01424	CBFA2/RUNX1 translocation partner 3	protein_coding
337	Lrrc66	Down	−4.017	−1.591	0.01424	Leucine rich repeat containing 66	protein_coding
338	Map2	Up	2.333	1.404	0.01427	Microtubule‐associated protein 2	protein_coding
339	Gfra4	Up	3.572	−0.459	0.01445	Glial cell line derived neurotrophic factor family receptor alpha 4	protein_coding
340	Epb41	Up	1.281	6.365	0.01448	Erythrocyte membrane protein band 4.1	protein_coding
341	Gzmk	Up	4.155	1.213	0.01478	Granzyme K	protein_coding
342	Gm31135	Up	2.888	−1.002	0.01500	Predicted gene, 31135	lncRNA
343	C3	Up	3.231	0.532	0.01637	Complement component 3	protein_coding
344	Gpr15	Down	−1.344	4.792	0.01646	G protein‐coupled receptor 15	protein_coding
345	Gm39650	Down	−3.030	−1.017	0.01672	NA	NA
346	Pipox	Up	2.324	0.907	0.01672	Pipecolic acid oxidase	protein_coding
347	Lrrn3	Up	1.683	3.432	0.01692	Leucine rich repeat protein 3, neuronal	protein_coding
348	Pheta2	Down	−2.076	1.023	0.01709	PH domain containing endocytic trafficking adaptor 2	protein_coding
349	Tspan7	Up	2.546	−1.036	0.01744	Tetraspanin 7	protein_coding
350	Gm10392	Up	2.813	−1.920	0.01773	Predicted gene 10392	lncRNA
351	Sestd1	Up	1.962	1.316	0.01776	SEC14 and spectrin domains 1	protein_coding
352	Hectd2	Up	1.507	2.405	0.01776	HECT domain E3 ubiquitin protein ligase 2	protein_coding
353	Klrb1c	Up	3.073	−0.893	0.01781	Killer cell lectin‐like receptor subfamily B member 1C	protein_coding
354	Igf2bp2	Up	2.801	0.468	0.01854	Insulin‐like growth factor 2 mRNA binding protein 2	protein_coding
355	Sparcl1	Up	2.347	−0.120	0.01918	SPARC‐like 1	protein_coding
356	Slc1a4	Up	2.984	1.591	0.02044	Solute carrier family 1 (glutamate/neutral amino acid transporter), member 4	protein_coding
357	Asb2	Up	1.928	4.379	0.02044	Ankyrin repeat and SOCS box‐containing 2	protein_coding
358	Stmn2	Up	2.286	−0.145	0.02055	Stathmin‐like 2	protein_coding
359	Fgf18	Up	3.190	−0.214	0.02085	Fibroblast growth factor 18	protein_coding
360	Gm3752	Down	−1.823	1.433	0.02104	Predicted gene 3752	protein_coding
361	Sytl3	Up	2.227	0.411	0.02207	Synaptotagmin‐like 3	protein_coding
362	Resp18	Up	2.338	0.079	0.02277	Regulated endocrine‐specific protein 18	protein_coding
363	Amph	Up	2.381	−0.952	0.02307	Amphiphysin	protein_coding
364	Pcsk1n	Up	2.333	−0.611	0.02329	Proprotein convertase subtilisin/kexin type 1 inhibitor	protein_coding
365	Pmch	Up	2.285	0.604	0.02329	Pro‐melanin‐concentrating hormone	protein_coding
366	Ifi202b	Up	3.165	−0.846	0.02329	Interferon activated gene 202B	protein_coding
367	Unc5a	Up	2.380	1.550	0.02349	Unc‐5 netrin receptor A	protein_coding
368	Acod1	Up	3.634	−2.294	0.02376	Aconitate decarboxylase 1	protein_coding
369	Gm50287	Up	1.860	0.922	0.02492	Predicted gene, 50287	lncRNA
370	Rgs8	Up	3.412	−1.604	0.02492	Regulator of G‐protein signaling 8	protein_coding
371	Gtf2i	Up	1.250	7.236	0.02512	General transcription factor II I	protein_coding
372	Amz1	Up	2.423	−0.304	0.02561	Archaelysin family metallopeptidase 1	protein_coding
373	Mcoln2	Up	1.890	1.753	0.02678	Mucolipin 2	protein_coding
374	Gm15742	Down	−3.808	−1.798	0.02678	Predicted gene 15742	lncRNA
375	Wnt5b	Up	2.067	0.903	0.02678	Wingless‐type MMTV integration site family, member 5B	protein_coding
376	Fam171b	Up	2.433	−0.362	0.02684	Family with sequence similarity 171, member B	protein_coding
377	Epas1	Up	2.129	−0.168	0.02937	Endothelial PAS domain protein 1	protein_coding
378	Jak2	Up	1.300	7.485	0.02950	Janus kinase 2	protein_coding
379	Npnt	Up	3.175	1.098	0.03015	Nephronectin	protein_coding
380	Susd2	Up	2.140	5.066	0.03106	Sushi domain containing 2	protein_coding
381	Klrb1f	Up	2.436	−0.438	0.03222	Killer cell lectin‐like receptor subfamily B member 1F	protein_coding
382	Dennd3	Up	1.673	2.024	0.03291	DENN domain containing 3	protein_coding
383	Cyp1a1	Up	4.184	−0.785	0.03291	Cytochrome P450, family 1, subfamily a, polypeptide 1	protein_coding
384	Selp	Down	−3.511	−1.173	0.03391	Selectin, platelet	protein_coding
385	Gm13522	Down	−2.372	−1.256	0.03413	Predicted gene 13522	lncRNA
386	Dusp5	Down	−1.289	5.296	0.03433	Dual specificity phosphatase 5	protein_coding
387	Rassf6	Up	2.442	0.839	0.03453	Ras association (RalGDS/AF‐6) domain family member 6	protein_coding
388	Gzmc	Down	−2.104	−0.098	0.03612	Granzyme C	protein_coding
389	Ttr	Up	1.808	2.181	0.03618	Transthyretin	protein_coding
390	6030468B19Rik	Down	−2.527	−1.342	0.03618	RIKEN cDNA 6030468B19 gene	protein_coding
391	Sytl1	Down	−1.269	4.904	0.03673	Synaptotagmin‐like 1	protein_coding
392	Nid2	Down	−1.979	0.308	0.03709	Nidogen 2	protein_coding
393	Tubb3	Up	2.313	−0.377	0.03815	Tubulin, beta 3 class III	protein_coding
394	Gxylt2	Down	−2.595	−0.536	0.03822	Glucoside xylosyltransferase 2	protein_coding
395	Prkar2a	Up	1.300	4.366	0.03833	Protein kinase, camp dependent regulatory, type II alpha	protein_coding
396	Uchl1	Up	2.269	0.086	0.03917	Ubiquitin carboxy‐terminal hydrolase L1	protein_coding
397	Etfbkmt	Down	−1.289	5.527	0.03931	Electron transfer flavoprotein beta subunit lysine methyltransferase	protein_coding
398	Gm34150	Up	1.913	−0.004	0.03961	Predicted gene, 34150	lncRNA
399	Ndn	Up	2.253	−1.028	0.03961	Necdin, MAGE family member	protein_coding
400	Xcr1	Up	1.803	1.080	0.03967	Chemokine (C motif) receptor 1	protein_coding
401	Ctla4	Down	−1.246	8.645	0.04145	Cytotoxic T‐lymphocyte‐associated protein 4	protein_coding
402	Samd3	Up	1.856	1.562	0.04197	Sterile alpha motif domain containing 3	protein_coding
403	B9d1	Down	−1.706	0.995	0.04226	B9 protein domain 1	protein_coding
404	Enpp2	Up	2.331	1.870	0.04273	Ectonucleotide pyrophosphatase/phosphodiesterase 2	protein_coding
405	Rph3a	Up	2.456	−1.491	0.04297	Rabphilin 3A	protein_coding
406	Prag1	Up	1.299	4.431	0.04302	PEAK1 related kinase activating pseudokinase 1	protein_coding
407	Cyp39a1	Up	1.769	1.352	0.04302	Cytochrome P450, family 39, subfamily a, polypeptide 1	protein_coding
408	Col16a1	Up	2.184	0.014	0.04302	Collagen, type XVI, alpha 1	protein_coding
409	Pawr	Up	2.942	1.066	0.04312	PRKC, apoptosis, WT1, regulator	protein_coding
410	Fam43a	Up	1.807	4.221	0.04468	Family with sequence similarity 43, member A	protein_coding
411	Esm1	Up	1.886	1.569	0.04566	Endothelial cell‐specific molecule 1	protein_coding
412	Dnase1l3	Up	2.511	−0.511	0.04575	Deoxyribonuclease 1‐like 3	protein_coding
413	Ces2d‐ps	Up	1.625	1.471	0.04656	Carboxylesterase 2D, pseudogene	unprocessed_pseudogene
414	Ptprk	Up	2.409	−0.750	0.04738	Protein tyrosine phosphatase, receptor type, K	protein_coding
415	Dgki	Up	2.390	0.434	0.04796	Diacylglycerol kinase, iota	protein_coding
416	Ccdc80	Up	1.792	1.327	0.04796	Coiled‐coil domain containing 80	protein_coding
417	Ccr3	Down	−2.142	0.710	0.04944	Chemokine (C‐C motif) receptor 3	protein_coding
418	Gm8206	Down	−2.513	−0.773	0.04971	Predicted gene 8206	protein_coding

Table 2. Enriched GO terms for the top 100 DE genes identified in ex‐Foxp3 cells compared to Treg cells.

ONT	ID	Description	GeneRatio	BgRatio	BH P.adj.	qvalue	Genes
BP	GO:0042110	T cell activation	16/88	461/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Themis/Il1rl2/Dapl1/Tcf7/Tnfsf8/Sema4a/Il4/Il2/Sdc4/Il21/Tnfrsf4/H2‐Q2
BP	GO:0046631	Alpha‐beta T cell activation	10/88	170/11437	0.000713	0.000556	Foxp3/Il2ra/Dapl1/Tcf7/Tnfsf8/Sema4a/Il4/Il2/Il21/H2‐Q2
BP	GO:0050865	Regulation of cell activation	16/88	499/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Gata1/Il1rl2/Dapl1/Tcf7/Capn3/Il4/Il2/Sdc4/Il21/Tnfrsf4/H2‐Q2/Id2
BP	GO:1903131	Mononuclear cell differentiation	14/88	391/11 437	0.000713	0.000556	Foxp3/Il2ra/Cd40lg/Gata1/Themis/Il1rl2/Tcf7/Tnfsf8/Sema4a/Il4/Bmyc/Il2/Il21/Id2
BP	GO:0042098	T cell proliferation	10/88	195/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Il4/Il2/Sdc4/Il21/Tnfrsf4/H2‐Q2
BP	GO:0051249	Regulation of lymphocyte activation	14/88	408/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Il1rl2/Dapl1/Tcf7/Il4/Il2/Sdc4/Il21/Tnfrsf4/H2‐Q2/Id2
BP	GO:0002694	Regulation of leukocyte activation	15/88	471/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Gata1/Il1rl2/Dapl1/Tcf7/Il4/Il2/Sdc4/Il21/Tnfrsf4/H2‐Q2/Id2
BP	GO:0002637	Regulation of immunoglobulin production	7/88	81/11 437	0.000713	0.000556	Foxp3/Cd40lg/Il4/Il2/Il21/Tnfrsf4/H2‐Q2
BP	GO:0050863	Regulation of T cell activation	12/88	299/11 437	0.000713	0.000556	Foxp3/Il2ra/Lrrc32/Cd40lg/Il1rl2/Dapl1/Tcf7/Il4/Il2/Sdc4/Il21/H2‐Q2
BP	GO:1903037	Regulation of leukocyte cell–cell adhesion	11/88	291/11 437	0.001401	0.001092	Foxp3/Il2ra/Lrrc32/Cd40lg/Il1rl2/Dapl1/Il4/Il2/Sdc4/Il21/H2‐Q2
BP	GO:0010721	Negative regulation of cell development	9/88	188/11 437	0.001401	0.001092	Foxp3/Ntrk3/Rflnb/Appl2/Sema4a/Il4/Bmyc/Il2/Id2
BP	GO:0030282	Bone mineralization	6/88	71/11 437	0.001644	0.001281	Ecm1/Gata1/Rflnb/Spp1/Vdr/Gpm6b
BP	GO:0050731	Positive regulation of peptidyl‐tyrosine phosphorylation	7/88	109/11 437	0.001769	0.001378	Gata1/App/Il4/Tnfrsf18/Hbegf/Il2/Il21
BP	GO:0006954	Inflammatory response	14/88	495/11 437	0.001769	0.001378	Foxp3/Il2ra/Ecm1/Cd40lg/Il1rl2/Appl2/Ggt5/App/Il4/Ifngr1/Cxcr6/Il2/Tnfrsf4/H2‐Q2
BP	GO:1903708	Positive regulation of hemopoiesis	8/88	163/11 437	0.002017	0.001572	Foxp3/Il2ra/Cd40lg/Il1rl2/Il4/Il2/Il21/Id2
BP	GO:0002683	Negative regulation of immune system process	12/88	385/11 437	0.002017	0.001572	Foxp3/Il2ra/Lrrc32/Dapl1/Il4/Bmyc/Il2/Sdc4/Tnfrsf4/H2‐Q2/Id2/Tgfb3
BP	GO:0030279	Negative regulation of ossification	4/88	25/11 437	0.002017	0.001572	Ecm1/Gata1/Rflnb/Vdr
BP	GO:1903522	Regulation of blood circulation	7/88	120/11 437	0.002017	0.001572	Ehd3/App/Smtn/Hbegf/Edn3/Hopx/Dock4
BP	GO:0002377	Immunoglobulin production	7/88	126/11 437	0.002283	0.001779	Foxp3/Cd40lg/Il4/Il2/Il21/Tnfrsf4/H2‐Q2
BP	GO:0048566	Embryonic digestive tract development	3/88	10/11 437	0.002283	0.001779	Tcf7/Stra6/Id2
BP	GO:0071352	Cellular response to interleukin‐2	3/88	10/11 437	0.002283	0.001779	Il2ra/Flicr/Il2
BP	GO:2000317	Negative regulation of T‐helper 17 type immune response	3/88	11/11 437	0.002798	0.002181	Foxp3/Il4/Il2
BP	GO:0006935	Chemotaxis	9/88	234/11 437	0.002889	0.002251	Ntrk3/App/Sema4a/Il4/Spp1/Cxcr6/Hbegf/Edn3/Dock4
BP	GO:0042330	Taxis	9/88	234/11 437	0.002889	0.002251	Ntrk3/App/Sema4a/Il4/Spp1/Cxcr6/Hbegf/Edn3/Dock4
BP	GO:0048589	Developmental growth	13/88	496/11 437	0.003333	0.002597	Socs2/Ecm1/Ntrk3/Lpar3/Tgfbr3/App/Capn3/Sema4a/Stra6/Hbegf/Edn3/Hopx/H2‐Q2
BP	GO:0062009	Secondary palate development	3/88	14/11 437	0.004047	0.003154	Lrrc32/Tgfbr3/Tgfb3
BP	GO:1900221	Regulation of amyloid‐beta clearance	3/88	14/11 437	0.004047	0.003154	App/Il4/Ifngr1
BP	GO:0007186	G protein‐coupled receptor signaling pathway	10/88	320/11 437	0.004324	0.003369	Cish/Gpr83/Lpar3/Vipr1/App/Rgs1/Cxcr6/Gpr34/Il2/Rapgef4
BP	GO:0002252	Immune effector process	12/88	455/11 437	0.004536	0.003534	Foxp3/Cd40lg/Gata1/Appl2/App/Sema4a/Il4/Il2/Il21/Tnfrsf4/H2‐Q2/Tgfb3
BP	GO:0016444	Somatic cell DNA recombination	5/88	71/11 437	0.005041	0.003928	Foxp3/Cd40lg/Tcf7/Il4/Il2
BP	GO:0048302	Regulation of isotype switching to IgG isotypes	3/88	16/11 437	0.005278	0.004113	Foxp3/Il4/Il2
BP	GO:1903532	Positive regulation of secretion by cell	8/88	222/11 437	0.006344	0.004943	Gata1/Il4/Spp1/Vdr/Edn3/Sdc4/Rapgef4/Tgfb3
BP	GO:0001503	Ossification	8/88	229/11 437	0.007273	0.005667	Ecm1/Gata1/Rflnb/Tgfbr3/Spp1/Vdr/Id2/Gpm6b
BP	GO:0032653	Regulation of interleukin‐10 production	4/88	46/11 437	0.00778	0.006063	Foxp3/Cd40lg/Il4/Il21
BP	GO:0030335	Positive regulation of cell migration	10/88	361/11 437	0.008095	0.006308	Ntrk3/App/Sema4a/Il4/Syne2/Tnfrsf18/Hbegf/Edn3/Dock4/Rapgef4
BP	GO:1901888	Regulation of cell junction assembly	6/88	129/11 437	0.008416	0.006558	Ntrk3/App/Sema4a/Sdc4/Hopx/Gpm6b
BP	GO:0021782	Glial cell development	5/88	88/11 437	0.009677	0.00754	Ntrk3/App/Ifngr1/Id2/Gpm6b
BP	GO:0040017	Positive regulation of locomotion	10/88	379/11 437	0.010389	0.008095	Ntrk3/App/Sema4a/Il4/Syne2/Tnfrsf18/Hbegf/Edn3/Dock4/Rapgef4
BP	GO:0097242	Amyloid‐beta clearance	3/88	27/11 437	0.015884	0.012377	App/Il4/Ifngr1
BP	GO:0048286	Lung alveolus development	3/88	34/11 437	0.026462	0.020619	Stra6/Hopx/Tgfb3
BP	GO:0050830	Defense response to Gram‐positive bacterium	4/88	73/11 437	0.027792	0.021656	Tnfsf8/App/H2‐Q2/Gbp2b
BP	GO:0045165	Cell fate commitment	5/88	123/11 437	0.029356	0.022875	Foxp3/Ntrk3/Gata1/Il4/Id2
BP	GO:0001774	Microglial cell activation	3/88	36/11 437	0.029657	0.023109	App/Il4/Ifngr1
BP	GO:0050727	Regulation of inflammatory response	7/88	245/11 437	0.031174	0.024291	Foxp3/Il2ra/Il1rl2/App/Il4/Il2/H2‐Q2
BP	GO:0051090	Regulation of DNA‐binding transcription factor activity	8/88	317/11 437	0.032951	0.025676	Foxp3/Cd40lg/Tgfbr3/App/Capn3/Il4/Tnfrsf4/Id2
BP	GO:0045348	Positive regulation of MHC class II biosynthetic process	2/88	11/11 437	0.033218	0.025883	Cd40lg/Il4
BP	GO:2000678	Negative regulation of transcription regulatory region DNA binding	2/88	11/11 437	0.033218	0.025883	Gata1/Id2
BP	GO:0006874	Intracellular calcium ion homeostasis	6/88	187/11 437	0.033695	0.026255	Gsto1/App/Capn3/Spp1/Vdr/Edn3
BP	GO:0030879	Mammary gland development	4/88	80/11 437	0.034969	0.027248	Socs2/Vdr/Id2/Tgfb3
BP	GO:0002260	Lymphocyte homeostasis	4/88	82/11 437	0.036671	0.028575	Foxp3/Il2ra/Il2/Tnfrsf4
BP	GO:0070227	Lymphocyte apoptotic process	4/88	82/11 437	0.036671	0.028575	Il2ra/Il2/Il21/Tnfrsf4
BP	GO:0033008	Positive regulation of mast cell activation involved in immune response	2/88	12/11437	0.036671	0.028575	Gata1/Il4
BP	GO:0043306	Positive regulation of mast cell degranulation	2/88	12/11 437	0.036671	0.028575	Gata1/Il4
BP	GO:0060343	Trabecula formation	2/88	12/11 437	0.036671	0.028575	Tgfbr3/Vdr
BP	GO:0048762	Mesenchymal cell differentiation	5/88	134/11 437	0.036671	0.028575	Rflnb/Tgfbr3/Sema4a/Edn3/Tgfb3
BP	GO:0051496	Positive regulation of stress fiber assembly	3/88	41/11 437	0.037266	0.029038	Pdlim4/Sdc4/Tgfb3
BP	GO:1901654	Response to ketone	4/88	84/11437	0.038556	0.030044	Tgfbr3/Spp1/Sgk1/Tgfb3
BP	GO:0006837	Serotonin transport	2/88	13/11 437	0.039893	0.031085	Slc22a2/Gpm6b
BP	GO:0033280	Response to vitamin D	2/88	13/11 437	0.039893	0.031085	Spp1/Vdr
BP	GO:0098810	Neurotransmitter reuptake	2/88	13/11 437	0.039893	0.031085	Slc22a2/Gpm6b
BP	GO:0006749	Glutathione metabolic process	3/88	43/11 437	0.039893	0.031085	Gsto1/Gsta4/Ggt5
BP	GO:0006691	Leukotriene metabolic process	2/88	14/11 437	0.044726	0.034851	Ggt5/Cyp4f18
BP	GO:0035461	Vitamin transmembrane transport	2/88	14/11 437	0.044726	0.034851	Slc22a2/Stra6
BP	GO:0001667	Ameboidal‐type cell migration	7/88	278/11 437	0.047415	0.036946	Appl2/Tgfbr3/Sema4a/Il4/Hbegf/Edn3/Sdc4
BP	GO:0016049	Cell growth	8/88	352/11 437	0.047415	0.036946	Ntrk3/Igfbp4/Lpar3/App/Sema4a/Sgk1/Hbegf/Edn3
BP	GO:0031099	Regeneration	4/88	93/11 437	0.047415	0.036946	Ntrk3/Gata1/Capn3/Hopx
BP	GO:0014829	Vascular associated smooth muscle contraction	2/88	15/11 437	0.047415	0.036946	Edn3/Dock4
BP	GO:0003281	Ventricular septum development	3/88	48/11 437	0.048477	0.037774	Tgfbr3/Stra6/Id2
BP	GO:0007631	Feeding behavior	3/88	48/11 437	0.048477	0.037774	Gpr83/App/Stra6
CC	GO:0009986	Cell surface	15/90	498/11 506	0.001227	0.001175	Il2ra/Lrrc32/Cd40lg/Tgfbr3/App/Il4/Itgae/Cxcr6/Ly6c1/Tnfrsf18/Hbegf/Sdc4/Tnfrsf4/H2‐Q2/Tgfb3
CC	GO:0009897	External side of plasma membrane	10/90	251/11 506	0.002384	0.002283	Il2ra/Cd40lg/Tgfbr3/Il4/Itgae/Cxcr6/Ly6c1/Tnfrsf18/Tnfrsf4/H2‐Q2
MF	GO:0030545	Signaling receptor regulator activity	12/90	170/11 350	2.32E‐06	2.08E‐06	Cd40lg/Tnfsf8/App/Sema4a/Il4/Spp1/Ly6c1/Hbegf/Edn3/Il2/Il21/Tgfb3
MF	GO:0038023	Signaling receptor activity	16/90	402/11 350	4.78E‐06	4.29E‐06	Il2ra/Gpr83/Ntrk3/Il1rl2/Lpar3/Vipr1/Tgfbr3/Vdr/Ifngr1/Cxcr6/Stra6/Mfsd6/Tnfrsf18/Gpr34/Sdc4/Tnfrsf4
MF	GO:0060089	Molecular transducer activity	16/90	402/11 350	4.78E‐06	4.29E‐06	Il2ra/Gpr83/Ntrk3/Il1rl2/Lpar3/Vipr1/Tgfbr3/Vdr/Ifngr1/Cxcr6/Stra6/Mfsd6/Tnfrsf18/Gpr34/Sdc4/Tnfrsf4
MF	GO:0005126	Cytokine receptor binding	9/90	144/11 350	6.60E‐05	5.92E‐05	Socs2/Ecm1/Cd40lg/Tgfbr3/Tnfsf8/Il4/Il2/Il21/Tgfb3
MF	GO:0019955	Cytokine binding	7/90	92/11 350	0.000237	0.000212	Il2ra/Lrrc32/Tgfbr3/Ifngr1/Cxcr6/Gbp2b/Tgfb3

Table 3. Enriched Kegg terms for the top 100 DE genes identified in ex‐Foxp3 cells compared to Treg cells.

ID	Category	Subcategory	Description	Gene Ratio	BgRatio	BH P.adj.	qvalue	Genes
mmu04060	Environmental information processing	Signaling molecules and interaction	Cytokine‐cytokine receptor interaction	12/52	130/5240	2.90E‐09	3.78E‐07	Il2ra/Cd40lg/Il1rl2/Tnfsf8/Il4/Ifngr1/Cxcr6/Tnfrsf18/Il2/Il21/Tnfrsf4/Tgfb3
mmu05321	Human diseases	Immune disease	Inflammatory bowel disease	6/52	46/5240	4.92E‐06	0.00032	Foxp3/Il4/Ifngr1/Il2/Il21/Tgfb3
mmu04630	Environmental information processing	Signal transduction	JAK–STAT signaling pathway	7/52	94/5240	3.31E‐05	0.00144	Il2ra/Socs2/Cish/Il4/Ifngr1/Il2/Il21
mmu04080	Environmental information processing	Signaling molecules and interaction	Neuroactive ligand‐receptor interaction	6/52	85/5240	0.00017	0.00557	Gpr83/Lpar3/Vipr1/1810009J06Rik/Edn3/Gm2663
mmu04659	Organismal systems	Immune system	Th17 cell differentiation	6/52	91/5240	0.00025	0.00650	Foxp3/Il2ra/Il4/Ifngr1/Il2/Il21
mmu05320	Human diseases	Immune disease	Autoimmune thyroid disease	4/52	36/5240	0.00040	0.00749	Cd40lg/Il4/Il2/H2‐Q2
mmu05330	Human diseases	Immune disease	Allograft rejection	4/52	36/5240	0.00040	0.00749	Cd40lg/Il4/Il2/H2‐Q2
mmu04672	Organismal systems	Immune system	Intestinal immune network for IgA production	3/52	33/5240	0.00408	0.06652	Cd40lg/Il4/Il2
mmu04658	Organismal systems	Immune system	Th1 and Th2 cell differentiation	4/52	73/5240	0.00567	0.08037	Il2ra/Il4/Ifngr1/Il2
mmu00983	Metabolism	Xenobiotics biodegradation and metabolism	Drug metabolism ‐ other enzymes	3/52	39/5240	0.00655	0.08037	Gsto1/Gsta4/Nme4
mmu05200	Human diseases	Cancer: overview	Pathways in cancer	9/52	350/5240	0.00677	0.08037	Il2ra/Gsto1/Gsta4/Lpar3/Tcf7/Il4/Ifngr1/Il2/Tgfb3
mmu05310	Human diseases	Immune disease	Asthma	2/52	14/5240	0.00814	0.08637	Cd40lg/Il4
mmu00480	Metabolism	Metabolism of other amino acids	Glutathione metabolism	3/52	43/5240	0.00860	0.08637	Gsto1/Gsta4/Ggt5
mmu00982	Metabolism	Xenobiotics biodegradation and metabolism	Drug metabolism ‐ cytochrome P450	2/52	19/5240	0.01483	0.13166	Gsto1/Gsta4
mmu04151	Environmental information processing	Signal transduction	PI3K‐Akt signaling pathway	6/52	205/5240	0.01513	0.13166	Il2ra/Lpar3/Il4/Spp1/Sgk1/Il2
mmu05204	Human Diseases	Cancer: overview	Chemical carcinogenesis ‐ DNA adducts	2/52	20/5240	0.01637	0.13357	Gsto1/Gsta4
mmu05140	Human diseases	Infectious disease: parasitic	Leishmaniasis	3/52	56/5240	0.01766	0.13557	Il4/Ifngr1/Tgfb3
mmu04820	NA	NA	Cytoskeleton in muscle cells	4/52	110/5240	0.02299	0.16672	Capn3/Syne2/Pdlim4/Sdc4
mmu00980	Metabolism	Xenobiotics biodegradation and metabolism	Metabolism of xenobiotics by cytochrome P450	2/52	25/5240	0.02505	0.17065	Gsto1/Gsta4
mmu04350	Environmental information processing	Signal transduction	TGF‐beta signaling pathway	3/52	65/5240	0.02614	0.17065	Lrrc32/Id2/Tgfb3
mmu05225	Human diseases	Cancer: specific types	Hepatocellular carcinoma	4/52	119/5240	0.02964	0.18422	Gsto1/Gsta4/Tcf7/Tgfb3

Table 4. Antibodies used in this study.

Antigen	Fluorochrome	Clone	Supplier	Application
CD45	BUV395	30‐F11	BD Biosciences	Analysis & Sorting
CD4	BUV496	GK1.5	BD Biosciences	Analysis
CD4	BV421	GK1.5	BD Biosciences	Sorting
CD8	APC‐H7	53–6.7	BD Biosciences	Analysis
CD25	PE‐Cy7	PC61	BD Biosciences	Analysis
CD304 (Nrp1)	BV421	3E12	BioLegend	Analysis
CCR6	BV650	140 706	BD Biosciences	Analysis
Foxp3	AlexaFluor 488	MF23	BD Biosciences	Analysis
IFNγ	BV510	XMG1.2	BioLegend	Analysis
IL‐17A	PerCP‐Cy5.5	TC11‐18H10	BD Biosciences	Analysis
Tbet	AlexaFluor 647	4B10	BD Biosciences	Analysis
RORγt	BV786	Q31‐378	BD Biosciences	Analysis

Full text

INTRODUCTION

Maternal immune tolerance is critical for embryo implantation and pregnancy success, and Treg cells are essential for mediating this tolerance as evidenced by embryo implantation failure and pregnancy loss after Treg cell depletion in mice.1, 2, 3 Deficiency in Treg cell abundance and/or function is found in inflammatory disorders of human pregnancy including recurrent miscarriage,4, 5, 6, 7, 8, 9 preeclampsia10, 11, 12 and preterm birth,13, 14 and is accompanied by skewing of the Treg/T effector balance in favor of inflammatory T effector cells.15, 16 Mouse models show a causal role for Treg cell deficiency in excessive T effector cell activity underpinning pregnancy loss,17, 18 demonstrating that an appropriate balance between Treg cells and T effector cells is pivotal for pregnancy success. As well as suppressing uterine inflammation, Treg cells interact with macrophages, dendritic cells (DCs) and uterine natural killer (uNK) cells residing in the uterine decidua (the layer of the uterus where maternal and placental cells interact) to actively promote key gestational processes including remodeling of the uterine vasculature, trophoblast invasion, placental morphogenesis and fetal growth.19, 20, 21, 22

Establishing and maintaining commitment to the Treg cell lineage is achieved through expression of the transcription factor Foxp3. Foxp3 synergizes with other transcription factors and cofactors, namely Eos, interferon regulatory factor 1 (IRF4), special AT‐binding protein 1 (Satb1), lymphoid enhancer‐binding factor 1 (Lef1) and GATA binding protein 1 (GATA1), to induce the Treg cell transcriptional program and repress effector T cell differentiation pathways.23, 24, 25 A Treg cell‐specific epigenetic signature is induced following T‐cell receptor (TCR) stimulation and is necessary for Treg cell lineage establishment and commitment. This pattern of CpG hypomethylation includes sites in the highly conserved Treg‐specific demethylated region (TSDR) of the Foxp3 promoter.26 These epigenetic changes also activate the Treg cell gene signature required for immune suppressive function, including expression of Ctla4, Il2ra (CD25), Tnfrsf18 (GITR) and Ikzf2 (Eos).26, 27

Foxp3^+^ Treg cells can exhibit phenotypic plasticity under certain environmental conditions, featuring expression of transcription factors, chemokine receptors and/or cytokines typically associated with other T effector cell lineages, to enable functional heterogeneity.28, 29 For example, some Treg cells express the Th1 transcription factor Tbet, and Tbet^+^ Treg cells possess immunosuppressive function and are critical for constraining Th1 and CD8^+^ cytotoxic T‐cell responses.29, 30, 31 While Foxp3^+^ Treg cells are reported to be stable in homeostatic and inflammatory conditions,32 other evidence demonstrates that some Treg cells display lineage instability, manifesting as loss of Foxp3 expression and conversion to other cell types.33, 34, 35, 36 For example, Treg cell fate‐mapping experiments using transgenic mouse models revealed that Foxp3^+^ Treg cells can convert to Th1 cells in Type I diabetes33 or in the presence of inflammatory interleukin‐6 (IL‐6) in vitro.37 Ex‐Foxp3 cells may arise from loss of Foxp3 expression in Treg cells33 or from T conventional (Tconv) cells that transiently express Foxp3 without fully committing to the Treg cell lineage.34 Of concern, ex‐Foxp3 cells can exhibit pathogenicity and are shown to contribute to disease progression in a range of inflammatory or lymphopenic environments including models of Type I diabetes, inflammatory arthritis and inflammation of the lung and liver.33, 34, 35, 36, 38, 39 Whether ex‐Foxp3 cells exist in gestational tissues and contribute to pregnancy pathologies is so far unexplored.

Parturition, the process by which the uterus and gestational tissues transition to birth, exhibits features of an inflammatory process whereby changes in gene expression in the myometrium and cervix cause production of proinflammatory cytokines including IL‐1, IL‐6 and IL‐8 and elicit an influx of activated immune cells into gestational tissues.40, 41 Ultimately, this process culminates in chorioamniotic membrane rupture, cervical ripening, and contractions of the myometrium to expel the fetus. Though the mechanisms triggering this proinflammatory transition are not fully understood, it is thought that in physiological parturition, sterile inflammation is initiated by endogenous damage associated molecular patterns (DAMPs)42 and release of surfactant protein‐A by the maturing fetal lung into the amniotic fluid, resulting in increased expression of IL‐1β by amniotic fluid macrophages. Stimulation of uterine nuclear factor kappa B (NF‐κB) expression following migration of amniotic fluid macrophages into the uterus drives an inflammatory signaling cascade at the maternal‐fetal interface.43 In pathological parturition, including preterm labor (birth <37 weeks gestation), this inflammatory signaling cascade is activated excessively and prematurely by sterile inflammation and/or pathogen‐associated molecular patterns (PAMPs) from extra‐ or intra‐uterine infection.41

During parturition, T cells are preferentially recruited to the fetal membrane rupture zone, and decidual effector T cells express activation markers in both term and preterm labour.44, 45 A role for proinflammatory adaptive immune cells in labor is also suggested by the accumulation of Th17 cells in the decidua at term in healthy pregnancy46 and the production of IL‐1β, tumor necrosis factor alpha (TNFα), and matrix metalloproteinase‐9 (MMP9) by decidual T cells in spontaneous term labour.47 A growing body of evidence suggests that Treg cells in the decidua and systemically undergo demise or exhibit phenotypic or functional changes during parturition in term pregnancy and in preterm labor, thus facilitating inflammatory activation. Women in both term and preterm labor show a decline in peripheral blood Treg cell abundance compared to gestational age‐matched non‐laboring women.48, 49 Whether Treg cells also decline in the decidua in preterm labor is contested. One study reports that, compared to preterm non‐laboring women, women who labored preterm and had chronic or acute placental inflammation had fewer Treg cells in the decidua parietalis, while preterm laboring women with no placental inflammation had fewer Treg cells in the decidua basalis.14 Another study found no changes in decidual Treg cell abundance in women who labored preterm compared with gestational age‐matched control women who did not labour.50 However, Treg cells from women with preterm labor, and to a lesser extent term labor, demonstrate reduced immunosuppressive activity.13, 51 Furthermore, mouse studies suggest that Treg cells may be protective against preterm birth. Late gestation Treg cell depletion in Foxp3^DTR^ mice increases the spontaneous preterm birth rate and heightens susceptibility to LPS‐induced preterm birth.14 Rag1 ^−/−^ mice, which are deficient in both T and B cells, are particularly susceptible to LPS‐induced preterm birth, while the transfer of T cells into Rag1 ^−/−^ mice rescues this effect, likely due to proliferation of Treg cells in the udLNs post‐transfer.52 Taken together, these studies and others imply that Treg cells are important for suppressing inflammation across gestation that might otherwise lead to premature labor.

Robust Foxp3 expression by Treg cells is important for pregnancy success,18, 53 implying that Treg cell lineage stability is critical. Furthermore, increased methylation of the Foxp3 TSDR and reduced Foxp3 protein expression have been observed in Treg cells in pregnancy pathologies including recurrent miscarriage,54 which has raised the question of whether Treg cell lineage instability predisposes to these conditions. Treg cells may be most vulnerable to loss of lineage commitment when they are exposed to the inflammatory cytokine milieu of parturition, given that proinflammatory cytokines including IL‐1 and IL‐6 that are major drivers of preterm labor55, 56, 57, 58 are linked with Treg cell instability in other tissue settings and in vitro.37 However, Treg cell stability in pregnancy has not been definitively investigated. Therefore, we utilized a genetic mouse model enabling tracking of Treg cell fate to test the hypothesis that Treg cells are stable throughout pregnancy but lose lineage stability during the late gestation parturition transition, and particularly in preterm birth when parturition is accelerated and inflammation is heightened.

RESULTS

Late gestation systemic IL‐1β administration increases Treg cell expression of Th17‐related molecules

Initially, we investigated whether exposure to an inflammatory stimulus in late gestation can increase inflammatory T cell cytokine expression by Treg cells serving the uterus. To achieve this, we used an established mouse model,56 in which pregnant C57Bl/6 females mated to Balb/c stud males were administered IL‐1β intraperitoneally on gestation day (GD) 15.5 to induce systemic inflammation in a protocol that leads to preterm birth (Supplementary figures 1 and 2). A control group was administered PBS/0.1% BSA (vehicle) only. In a pilot dose–response experiment, we found that a dose of 6 μg IL‐1β caused 88.9% of dams to deliver prematurely, with almost all pups being non‐viable (Supplementary figure S1).

To examine the impact of IL‐1β on T cells serving the uterus, the uterus‐draining para‐aortic lymph nodes (udLNs) were examined, as this is the primary site where uterine T cells are generated over the course of pregnancy. After proliferation in the udLN, T cells circulate via the peripheral blood and are recruited into the uterus.59, 60 Under baseline conditions, the decidual lining of the uterus abutting the placenta has a much lower density and diversity of T cells due to active epigenetic mechanisms that exclude T effector cells while permitting Treg cells, although inflammation can modulate this selective constraint.61 The response in the udLN was examined by flow cytometry on GD16.5, 24 h after administration of 6 μg IL‐1β (Figure 1a). At this time point, adverse effects on fetal development were evident, with IL‐1β administration causing a 5.9% reduction in mean fetal weight and a reduced fetal:placental weight ratio, a measure of placental functional competence (Supplementary figure 2).

*IL‐17A and CCR6 expression increases in udLN Foxp3+ Treg cells after IL‐1β administration to induce preterm birth. (a) Experimental design. Briefly, pregnant C57Bl/6 dams were administered 6 μg IL‐1β or vehicle control (PBS/0.1% BSA) on GD15.5 to induce preterm birth and flow cytometry of the udLNs was performed on GD16.5 to evaluate the T cell response. (b) Number of CD4+ T cells. (c) Proportion and (d) number of total Foxp3+ Treg cells. (e) Total number of tTreg cells and (f) pTreg cells. (g) Proportion and (h) number of IL‐17A+ Treg cells, and (i) their level of IL‐17A expression. (j) Proportion and (k) number of CCR6+ Treg cells, and (l) their level of CCR6 expression. (m) Proportion and (n) number of Tconv cells. (o) Proportion and (p) number of IFNγ+ Tconv cells (Th1 cells). (q) Proportion and (r) number of IL‐17A+ Tconv cells (Th17 cells). N = 11 per group. *P < 0.05, **P < 0.01, **P < 0.005 by unpaired t test.

Dams administered IL‐1β showed an overall increase in udLN leukocytes (Supplementary figure 3a, b), including an increased number of CD4^+^ T cells (Figure 1b), although CD4^+^ T cells were not specifically enriched as a proportion of CD45^+^ leukocytes (Supplementary figure 3c). Similarly, while Foxp3^+^ Treg cells did not increase as a proportion of CD4^+^ T cells (Figure 1c), there was an increased number of Foxp3^+^ cells (Figure 1d). An increase in the total number of both thymic Treg (tTreg) and peripheral Treg (pTreg) cells was seen, but tTreg cells comprised a smaller proportion of the Treg cell pool following inflammatory challenge (Figure 1e, f, Supplementary figure 3d).

Treg cells exhibiting phenotypic plasticity in other studies often adopt expression of Th17 cell‐associated molecules. We therefore next examined expression of the Th17‐related molecules RORγt, IL‐17A and CCR6 by Foxp3^+^ cells. The proportions of Foxp3^+^ cells expressing RORγt or IL‐17A did not change after IL‐1β challenge, although due to the increase in total Foxp3^+^ cells, overall there were 1.84‐fold more IL‐17A^+^ Treg cells (Figure 1g, h, Supplementary figure 3e, f). Also, the level of IL‐17A expression by IL‐17A^+^ Treg cells was increased, as indicated by a 1.23‐fold increased mean fluorescence index (MFI; Figure 1i). A similar pattern was observed for CCR6 expression, whereby there was no significant increase in the proportion of CCR6^+^ Treg cells in IL‐1β‐treated mice, although they increased 1.78‐fold in number and exhibited a 1.3‐fold elevated level of CCR6 expression (Figure 1j‐l). We also investigated the expression of Th1‐associated molecules Tbet and IFNγ and found non‐significant trends for increased Tbet^+^ and IFNγ^+^ Treg cell numbers, although their proportions among Treg cells were unchanged (Supplementary figure 3g–j). The level of IFNγ expression by Treg cells indicated by MFI was also unchanged (data not shown).

Tconv cells were increased 1.5‐fold in number after IL‐1β challenge, but were not enriched as a proportion of CD4^+^ T cells (Figure 1m, n). IFNγ^+^ Th1 cells comprised a very small population (<3% of CD4^+^ T cells, Figure 1o), and while their proportion among CD4^+^ T cells decreased, their total number was unchanged (Figure 1p). IL‐1β challenge did not change the proportion or number of IL‐17A^+^ Th17 cells (Figure 1q, r), or the MFI of IL‐17A or IFNγ expressed by Th1 or Th17 cells (data not shown).

Ex‐Foxp3 cells are present throughout gestation and are predominantly Nrp1− cells

The increased level of expression of Th17‐related molecules IL‐17A and CCR6 by Treg cells after IL‐1β administration in late gestation raised the question of whether these Treg cells exhibit lineage instability. To examine whether Treg cells undergo loss of Foxp3 in pregnancy, we employed an in vivo Foxp3 fate‐mapping system.33 Foxp3‐eGFP^Cre^ mice62 were crossed with Rosa26^RFP^ mice to generate Foxp3^eGFP‐Cre^Rosa26^RFP^ Treg cell fate‐mapping mice (Supplementary figure 4). In these mice, expression of Cre recombinase under the control of the Foxp3 promoter leads to recombination of two loxP sites flanking a STOP codon that precedes RFP transcription, resulting in excision of the STOP codon. Therefore, all cells that initiate Foxp3 expression become permanently marked by RFP, including cells that later switch off Foxp3 expression, while current Foxp3 expression is simultaneously marked by GFP. Therefore, Treg cells are identified as RFP^+^GFP^+^ cells, while RFP^+^GFP^−^ cells represent a population of ex‐Foxp3 cells converted from Treg cells, or alternatively a population of CD4^+^ T cells that previously expressed Foxp3 at a low level without committing to the Treg cell transcriptional program of differentiation.34

Using the Foxp3 fate‐mapping model, we looked for ex‐Foxp3 cells in the udLNs of non‐pregnant mice and pregnant mice across gestation including around embryo implantation (GD4.5), mid‐late gestation (GD16.5), and late gestation (GD18.5) (Figure 2a, b). As well as udLN, we examined spleen to inform on systemic effects. The total cellularity of the udLNs and spleen and the number of CD4^+^ and CD8^+^ cells in these tissues did not change across gestation (Supplementary figure 5). In the udLNs, the number of RFP^+^GFP^+^ Treg cells did not change across timepoints (Figure 2c). However, as a proportion of CD4^+^ T cells, their frequency at GD18.5 was increased 1.8 and 2.0‐fold compared to non‐pregnant and early pregnancy (GD4.5), respectively (Figure 2d). tTreg cells (Nrp1^+^) comprised the majority (76–80%) of Treg cells in the udLNs at all time‐points (Figure 2e). In the spleen, Treg cells were also not significantly altered in proportion or number, and splenic tTreg cells were similarly more abundant than pTreg cells across gestation (Supplementary figure 6a–d).

*Ex‐Foxp3 cells are present in the udLNs throughout gestation. (a) Experimental design. The abundance of udLN Treg cells, ex‐Foxp3 cells and Tconv cells was tracked in non‐pregnant (NP) and pregnant Foxp3 fate‐mapping mice across gestation. (b) Gating of the 3 CD4+ T cell subsets based on fluorescent reporter protein expression. (c) Number and (d) proportion of Treg cells (GFP+RFP+). (e) Proportion of tTreg and pTreg cells. (f) Number and (g) proportion of ex‐Foxp3 cells (GFP−RFP+). (h) Proportion of Nrp1+ and Nrp1− cells among ex‐Foxp3 cells. (i) Ratio of ex‐Foxp3 cells:Treg cells. (j) Number and (k) proportion of Tconv cells. N = 4–11 per group. *P < 0.05, *P < 0.01, all groups were compared by One‐way ANOVA and Tukey's post‐hoc test.

A small population of RFP^+^GFP^−^ ex‐Foxp3 cells was present in the udLNs at all time‐points. In non‐pregnant mice, ex‐Foxp3 cells were 3.6% of udLN CD4^+^ T cells, and they did not change in proportion or number in non‐pregnant versus pregnant mice, or over the course of pregnancy (Figure 2f, g). The majority of udLN ex‐Foxp3 cells, ranging from 52% to 62% across gestation, were Nrp1^−^. In contrast, Nrp1^+^ cells comprised only 39–48% of the total udLN ex‐Foxp3 pool, which was 1.6 to 2.1‐fold less than for RFP^+^GFP^+^ Treg cells (Figure 2h). Ex‐Foxp3 cells were also present in the spleen, representing 6.2% of CD4^+^ T cells in non‐pregnant mice, but their abundance did not change over gestation (Supplementary figure 6e, f). As in the udLNs, splenic Nrp1^−^ cells comprised a greater proportion of ex‐Foxp3 cells compared with Treg cells (Supplementary figure 6g). The fact that Nrp1^−^ cells contributed to a larger fraction of the ex‐Foxp3 pool in udLNs and spleen is consistent with ex‐Foxp3 cells arising from either unstable pTreg cells or briefly activated T conventional (Tconv) cells.

To further determine whether increased conversion to ex‐Foxp3 occurred during pregnancy, we measured the ex‐Foxp3:Treg cell ratio, expecting it would be increased if more Treg cells converted to ex‐Foxp3 cells. In contrast, there was a significant decrease in the ratio of ex‐Foxp3:Treg cells in the udLN at GD18.5 compared to GD4.5, indicating that overall, Treg cells do not exhibit increased instability and conversion to ex‐Foxp3 cells in late gestation (Figure 2i). Instead, they appear to remain stable or may even increase in stability as gestation progresses. In the spleen, the ratio of ex‐Foxp3:Treg cells was unchanged across gestation, implying that Treg cells are stable in pregnancy at the systemic level (Supplementary figure 6h).

Like Treg cells and ex‐Foxp3 cells, Tconv cells were unchanged in number in the udLN across gestation. However, there was a decrease in the proportion of Tconv cells in late gestation (GD18.5) compared to early pregnancy (GD4.5) (Figure 2j), in contrast to the increase in the proportion of Treg cells (Figure 2d). In the spleen, Tconv cell proportion and number were unchanged across pregnancy (Supplementary figure 6i, j).

Ex‐Foxp3 cells maintain proinflammatory cytokine expression during pregnancy

We next investigated whether the ex‐Foxp3 cells found in the udLNs and spleen expressed T helper cytokines that might mediate pathogenic effects. Expression of the Th1 and Th17 cytokines IFNγ and IL‐17A was measured in ex‐Foxp3 cells as well as in Treg cells and Tconv cells across gestation.

Expression of IFNγ or IL‐17A in udLN Treg cells was rare in non‐pregnant dams and in early pregnancy GD4.5, at <0.8% and 1.6%, respectively (Figure 3a, b). By GD16.5, there was a non‐significant trend to a 5.9‐fold increase in udLN Treg cells expressing IFNγ compared to GD4.5 (P = 0.076). There was no significant change in IL‐17A expression by Treg cells in the udLNs in late gestation, and the mean proportion of Treg cells expressing either of these cytokines at GD18.5 was still relatively small, at 1.5% and 3.1% for IFNγ and IL‐17A, respectively. Similarly low proportions and numbers of Treg cells expressing IFNγ or IL‐17A were observed in the spleen, and this did not significantly change across pregnancy (Supplementary figure 7a, b).

The ex‐Foxp3 population in udLNs is enriched for cytokine‐expressing cells. The experimental design is outlined in Figure 2. (a) Proportion of IFNγ+ and (b) IL‐17A+ Treg cells. (c) Proportion of IFNγ+ and (d) IL‐17A+ ex‐Foxp3 cells. (e) Proportion of IFNγ+ (Th1) and (f) IL‐17A+ (Th17) Tconv cells. N = 4–11 per group. All groups were compared by one‐way ANOVA and Tukey's post‐hoc test.

Compared to Treg cells, a larger proportion of udLN ex‐Foxp3 cells expressed IFNγ or IL‐17A in non‐pregnant dams and early pregnancy (Figure 3c, d). In fact, the mean expression of IFNγ in ex‐Foxp3 cells in the udLNs at GD4.5 was 4.4%, or 19.1‐fold higher than in Treg cells. Similarly, the mean proportion of IL‐17A‐expressing ex‐Foxp3 cells at GD4.5 was 14.8‐fold higher than Treg cells, at 11.3% in the udLNs. In the spleen, a higher proportion of IFNγ expression was observed in the ex‐Foxp3 population, at 6.7% on GD4.5, while the proportion of IL‐17A‐expressing ex‐Foxp3 cells was 5.8% at GD4.5, which was lower than in udLNs (Supplementary figure 7c, d). Expression of these cytokines did not change in ex‐Foxp3 cells across gestation in either the udLNs or spleen.

The frequency of IFNγ^+^ or IL‐17A^+^ cells among Tconv cells was lower than for ex‐Foxp3 cells across all of gestation in both the udLNs and spleen (Figure 3e, f, Supplementary figure 7). At GD4.5, a mean of 1.3% and 1.4% of udLN Tconv cells, respectively, expressed IFNγ or IL‐17A, which was 3.3 and 8.3‐fold lower than ex‐Foxp3 cells at the same time‐point, and not much higher than in Treg cells (Figure 3e, f). In the spleen, 5.4% of Tconv cells were IFNγ^+^ at GD4.5, while 1.3% expressed IL‐17A (Supplementary figure 7e, f). There was no significant increase in cytokine expression among Tconv cells at late gestation in either the udLNs or spleen.

These findings indicate that although ex‐Foxp3 cells represent a much smaller population than Tconv cells, they are enriched for proinflammatory cytokine expression. However, ex‐Foxp3 cell expression of these cytokines does not appear to increase in late gestation in normal pregnancy.

Ex‐Foxp3 cells do not express the canonical Treg cell lineage transcriptional program

The presence of ex‐Foxp3 cells in the udLN across gestation led us to investigate whether these cells have a distinct transcriptional profile compared to Treg cells. Ex‐Foxp3 cells and Treg cells were sorted from the udLN on GD18.5 and compared by RNA‐sequencing (Figure 4a).

A distinct transcriptional program in ex‐Foxp3 cells compared to Treg cells. (a) Experimental design. Ex‐Foxp3 cells and Treg cells were sorted from udLNs of GD18.5 Foxp3 fate‐mapping mice by fluorescence‐activated cell sorting (FACS), then analyzed by RNAseq. (b) Principal‐component analysis showing clustering of samples by cell type. Ovals represent 95% confidence interval of the group mean. (c) Volcano plot showing significantly upregulated and downregulated genes in ex‐Foxp3 cells compared to Treg cells. A log fold‐change (lfc) cut‐off of 1 (equivalent to 2‐fold change in expression) and FDR‐adjusted P‐value <0.05 was used to determine significance. (d) Heatmap of relative expression levels of the top 25 most significantly DE genes in ex‐Foxp3 cells. (e) Comparison of the expression of significantly DE transcription factors, (f) cytokines, (g) granzymes and (h) chemokine receptors in the RNAseq. N = 5 mice pooled per sample, and 5 samples per group. CPM: counts per million.

Principal component analysis showed clustering of samples based on their cellular identity as Treg or ex‐Foxp3 cells on the first component PC1, representing 66.3% of the variation in the data (Figure 4b). Differential gene expression analysis revealed differential expression of 7840 genes (adj.P < 0.05, no fold‐change (lfc) threshold), of which approximately half were upregulated and half downregulated. Among these, 418 genes showed a lfc >1, equivalent to at least 2‐fold change in expression, with 284 of these genes upregulated and 134 downregulated in ex‐Foxp3 cells (Figure 4c, Table 1).

Table 1: Differentially expressed genes in ex‐Foxp3 cells compared to Treg cells in the udLNs of pregnant Foxp3 fate‐mapper dams at GD18.5, with a logFC >1 and adj.P.Val. <0.05.

The 2 most significantly suppressed genes in ex‐Foxp3 cells were Foxp3 and Il2ra encoding the key Treg cell identifying molecules Foxp3 and CD25 (Figure 4d). A suite of other genes associated with Treg cell identity or function were similarly downregulated, including Ikzf4 (encoding Eos), Ctla4, Tnfrsf18 (GITR), Tnfrsf4 (CD134/OX40), Tnfrsf9 (CD137/4‐1BB), *Lrrc32 (*GARP), Rgs1, Ebi3, and many others (Table 1). There was upregulation of some genes associated with improved Treg cell fitness, function, or survival, including Themis, Sema4a and Vdr. There were 8 identified differentially expressed transcription factors, among which Foxp3, Gata1, Ikzf4 (Eos) and Tcf7l2 (TCF‐4) were downregulated, while Tcf7 (TCF‐1), Tox2, Id2 and Zeb2 were upregulated (Figure 4e).

Among the transcriptional changes in ex‐Foxp3 cells, we also identified downregulation of genes that suppress T helper cell programming including the IL‐4/STAT6‐suppressing genes Socs2 and Cish (Figure 4d). Furthermore, there was upregulation of T helper cell effector molecules including the cytokines Il21, Il2, Il4 and Ifng (Figure 4f). Other effector molecules that were differentially expressed included granzymes, with Gzmb and Gzmc downregulated and Gzmk upregulated in ex‐Foxp3 cells (Figure 4g). Chemokine receptor expression was dysregulated in ex‐Foxp3 cells, such that Cxcr5, Cxcr6 and Xcr1 were upregulated and Ccr3, Ccr6 and Ccr8 were downregulated (Figure 4h).

Pathway enrichment analysis using the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (Kegg) databases for the top 100 most significantly differentially expressed genes using ClusterProfiler63, 64 indicated dysregulation of T cell activity in ex‐Foxp3 cells. GO analysis, with removal of redundant terms using ClusterProfiler ‘simplify’, revealed enrichment for 69 Biological Process, 2 Cellular Component and 5 Molecular Function terms (Table 2). Among the top 10 enriched terms in the Biological Process ontology, 9 were related to T cell or leukocyte activation or differentiation. The other term was ‘regulation of immunoglobulin production’, for which associated differentially expressed genes were related to upregulation of Th2 or T follicular helper or regulatory (T_FH_ or T_FR_) cell genes, Il4, Il21, Cd40lg and Tnfrsf4 (OX40), and the MHC gene H2‐Q2. Other enriched terms included ‘cellular response to interleukin‐2’, ‘negative regulation of T‐helper 17 type immune response’, ‘immune effector process’, ‘regulation of interleukin‐10 production’, ‘cell fate commitment’ and ‘regulation of inflammatory response’. Examination of the genes annotated with these terms showed they were mostly either down‐regulated, or would result in suppression of the process. Enriched terms in the Molecular Function ontology were related to cytokine or signaling receptor activity, highlighting the dysregulated chemokine receptors and cytokine transcriptional changes described above. This was consistent with the enriched terms in the Cellular Component ontology, which were ‘cell surface’ and ‘external side of plasma membrane’ and in which the genes annotated with these terms were mainly cytokines, cytokine receptors and other cell‐surface receptors such as those from the TNFR superfamily. Overall, these changes indicate that ex‐Foxp3 cells exhibit a lack of Treg cell programming and function and de‐repression of the T helper cell differentiation programs and their inflammatory effector functions.

Kegg‐enrichment analysis revealed 21 enriched terms, including 7 being in the Immune system or Immune disease subcategories, 5 terms related to signaling molecules or signal transduction and 3 terms related to metabolism (Table 3). The immune‐related terms hinted at inflammatory potential, which is consistent with poor Treg cell immunosuppressive identity and function. These terms were related to differentiation to T helper subsets (‘Th17 differentiation’ and ‘Th1 and Th2 cell differentiation’) and inflammatory conditions (‘Inflammatory bowel disease’ and ‘Allograft rejection’).

Treg and ex‐Foxp3 cells are refractory to late‐gestation LPS or IL‐1β challenge

Finally, although Treg cells appeared to be stable over the course of normal pregnancy, our finding that Treg cells expressed higher levels of Th17‐related molecules IL‐17A and CCR6 after IL‐1β‐administration in C57Bl/6 mice (Figure 1) prompted us to test whether inflammatory challenges that induce preterm birth can destabilize Treg cells and increase their conversion to ex‐Foxp3 cells. Here, we utilized IL‐1β as described above, as well as a second model of preterm birth induced by systemic administration of the gram‐negative bacterial cell wall component lipopolysaccharide (LPS). LPS induces inflammation via the TLR4 pathway, in which IL‐1β is a downstream mediator. DCs activated with LPS have demonstrated capacity to induce Treg cell lineage conversion in vitro.37 The timing and dose of LPS or IL‐1β were each optimized to ensure they effectively induced preterm birth yet maintained relatively high fetal viability (Supplementary figure 8). Dams were injected intraperitoneally with 4 μg LPS at 0900 h on GD17.5, and a separate group of dams was injected with 4 μg IL‐1β at 1900 h on GD17.5 (Figure 5a). Control dams received vehicle (PBS/0.1% BSA). The later injection time was used for IL‐1β since it is downstream in the inflammatory cascade and therefore causes faster progression to preterm birth than LPS. In these experiments, we tracked ex‐Foxp3 cells in the udLNs and spleen and also included the uterine decidua to examine local changes at the placental‐maternal interface where effects on uterine contractility and fetal expulsion are manifested.

Treg cells maintain stability after late‐gestation inflammatory challenge to induce preterm birth. (a) Experimental design. Pregnant Foxp3 fate‐mapping mice were administered 4 μg LPS on the morning (0900 h) of GD17.5 or 4 μg IL‐1β on the evening (1900 h) of GD17.5. Controls received vehicle PBS/0.1% BSA. Flow cytometry was performed on the udLNs and decidua on the following morning on GD18.5 to track Treg cell stability by fluorescent reporter protein expression. (b) Total number of udLN and (c) decidual CD4+ T cells. (d) Proportion of Treg cells in the udLNs and (e) decidua. (f) Number of Treg cells in the udLNs and (g) decidua. (h) Proportion of ex‐Foxp3 cells in the udLNs and (i) decidua. (j) Number of ex‐Foxp3 cells in the udLNs and (k) decidua. (l) Proportion of Tconv cells in the udLNs and (m) decidua. (n) Number of Tconv cells in the udLNs and (o) decidua. N = 6–8 per group. The LPS‐ and IL‐1β‐treated groups were compared to the control group by One‐way ANOVA and Dunnett's post‐hoc test.

The total number of CD4^+^ T cells in the udLN, decidua and spleen was not significantly different after LPS or IL‐1β challenge (Figure 5b, c, Supplementary figure 9). RFP^+^GFP^+^ Treg cells were also unchanged both as a proportion of CD4^+^ T cells and in cell number (Figure 5d‐g, Supplementary figure 9c, d). Likewise, RFP^+^GFP^−^ ex‐Foxp3 cells were not significantly different as a proportion of CD4^+^ T cells or as cell number after either LPS or IL‐1β (Figure 5h‐k, Supplementary figure 9g, h). This shows there is no specific induction of ex‐Foxp3 cells following inflammatory challenge in either model. There were also no changes in RFP^−^GFP^−^ Tconv cell proportion or number (Figure 5l–o, Supplementary figure 9k, l). Although the total number of Treg cells, ex‐Foxp3 cells and Tconv cells appeared to increase slightly after LPS administration in udLNs, decidua and spleen, this was not statistically significant (Figure 5g, k, o, Supplementary figure 9).

The expression of IFNγ and IL‐17A in Treg, ex‐Foxp3, and Tconv cells was also measured to evaluate inflammatory potential. In mice treated with LPS, but not IL‐1β, IFNγ^+^ Treg cells appeared to increase in the udLN and decrease in the decidua, but this was not statistically significant (Figure 6a, b). There were no changes in the expression of IFNγ by ex‐Foxp3 cells or Tconv cells in the udLN or decidua after LPS or IL‐1β challenge in late gestation (Figure 6c–f). IL‐17A expression by udLN and decidual Treg, ex‐Foxp3, or Tconv cells was also not significantly altered by LPS or IL‐1β administration (Figure 6g–l). Treg cells, ex‐Foxp3 cells, and Tconv cells also showed no changes in the proportion of IFNγ or IL‐17A expressing cells in the spleen of LPS or IL‐1β‐treated mice (Supplementary figure 9). Overall, this indicates that although ex‐Foxp3 cells present in the decidua and systemically express T helper cytokines, this expression is not amplified by proinflammatory challenge with LPS or IL‐1β in late gestation.

Expression of cytokines by T cell subsets after late‐gestation inflammatory challenge. The experiment was carried out as shown in Figure 5. (a) Proportion of IFNγ+ Treg cells in the udLNs and (b) uterine decidua. (c) Proportion of IFNγ+ ex‐Foxp3 cells in the udLNs and (d) decidua. (e) Proportion of IFNγ+ Tconv cells in the udLNs and (f) decidua. (g) Proportion of IL‐17A+ Treg cells in the udLNs and (h) decidua. (i) Proportion of IL‐17A+ ex‐Foxp3 cells in the udLNs and (j) decidua. (k) Proportion of IL‐17A+ Tconv cells in the udLNs and (l) decidua. N = 7–8 per group. The LPS‐ and IL‐1β‐treated groups were compared to the control group by One‐way ANOVA and Dunnett's post‐hoc test.

DISCUSSION

Abundant populations of uterine Treg cells with potent immunosuppressive function are essential for pregnancy success, yet we have discovered that a small population of ex‐Foxp3 cells lacking key Treg cell functional genes are present in the uterine decidua, udLNs and spleen during pregnancy in mice. These cells most likely arise from Treg cells which have lost expression of Foxp3 and undergone cell lineage conversion. Alternatively, since activation of CD4^+^ T cells can transiently induce Foxp3 expression without leading to Treg cell differentiation,34 these ex‐Foxp3 cells may be a subpopulation of Tconv cells derived through this pathway. Regardless of their ontogeny, elevated expression of inflammatory cytokines by ex‐Foxp3 cells compared with Treg cells and Tconv cells, and their deficit in expression of immunoregulatory molecules, leaves open the potential for them to contribute to immune‐mediated pregnancy pathology.

Treg cells are thought to exhibit plasticity and instability in various in vitro and in vivo inflammatory or lymphopenic settings. Lineage tracing methods have identified that ex‐Foxp3 cells can gain effector T cell functions,33, 34, 36, 65 and indicate a contribution to loss of allotolerance in transplantation,65 as well as autoimmune disease pathology in inflammatory arthritis and Type I diabetes.33, 36 However, it is unclear if ex‐Foxp3 cells arise after conversion from Treg cells rendered unstable by inflammation or whether ex‐Foxp3 cells are an accumulated population of Tconv cells that transiently express Foxp3 in their ontogeny. Whether Treg cells exhibit instability in normal pregnancy, and importantly whether this might have a causal role in inflammatory pregnancy complications, such as miscarriage, preeclampsia and preterm birth, are important questions that have not been explored.

During pregnancy, induction of Treg cells mediating tolerance of paternally inherited alloantigens is controlled by pregnancy hormones, seminal fluid and paternal and fetal antigens. The most vulnerable periods of gestation in which Treg cell switching reasonably might occur are during the post‐coital response to seminal fluid priming, during embryo implantation, upon completion of placental development when the maternal blood and circulating immune cells make direct contact with placental antigens, and during parturition when inflammatory cytokines are released. Seminal fluid exposure is particularly important in the generation of Treg cells in the udLN and uterus,66, 67 as this fluid delivers male partner alloantigens in the context of abundant TGF‐β and other pro‐tolerogenic factors, and its composition has potential to affect the stability of Treg cells.16 Progesterone (P4), which is high throughout pregnancy until luteolysis in late gestation, also regulates Treg cell abundance and phenotype,68, 69 and acts to reduce IFNγ expression in both Treg and Tconv cells68 and inhibit Th17 cell differentiation. P4 has been demonstrated to increase the stability of Foxp3 expression in Treg cells induced in vitro,69 so Treg cell lineage instability due to luteal phase P4 insufficiency may be a key factor affecting risk of reproductive and pregnancy disorders.

The apparent stability of Treg cells demonstrated in the current study is consistent with our previous finding in mice that hypomethylation of the CNS1 region of the Foxp3 promoter occurs in tTreg cells after seminal fluid exposure to reinforce and stabilize the Treg cell lineage program.16 pTreg cells do not exhibit this same hypomethylation adaptation in early pregnancy, which may predispose them to greater instability. Indeed, we found that the ex‐Foxp3 cell pool was enriched for Nrp1^−^ cells compared to the Treg cell pool, indicating that pTreg cells convert to ex‐Foxp3 at a greater rate than tTreg cells. However, this was equivalent in non‐pregnant and pregnant females, and no increased propensity for either tTreg or pTreg conversion was seen over the course of gestation either locally in the udLNs or systemically in the spleen.

Evidence from in vitro differentiation experiments highlights the necessity of several environmental cues for proper F oxp 3 TSDR demethylation and Treg cell lineage commitment.70, 71 It is possible that in some pregnancies, absence of certain signals or the presence of signals that skew lineage commitment, such as IL‐6 which favors Th17 cell generation rather than Treg cells,72 may occur in the udLN and uterus when P4 and/or seminal fluid factors are suboptimal. This could reasonably lead to Treg cell instability and conversion to ex‐Foxp3 cells with effector functions, to compromise maternal immune tolerance and contribute to pathologies associated with a T cell imbalance such as miscarriage, preeclampsia and preterm birth. Alternatively, women predisposed to Treg cell instability due to genetic or environmental factors may have elevated risk of infertility and/or pregnancy pathologies. Consistent with this, we recently identified that women experiencing recurrent implantation failure or recurrent miscarriage have fewer Treg cells and bear hallmarks of lineage instability, in the form of lower Foxp3 levels and an aberrant transcriptional signature including proinflammatory cytokines upon activation in vitro, compared to fertile women.73 As Foxp3 TSDR hypermethylation in recurrent miscarriage patients negatively correlates with Treg cell abundance,54 both the scale and stability of the Treg cell response initiated at conception are likely to be instrumental for pregnancy tolerance.

The ex‐Foxp3 cells observed in udLNs in late‐gestation were transcriptionally distinct from Treg cells, including downregulation of four major genes that show Treg cell‐specific CpG hypomethylation during lineage commitment: Foxp3, Tnfrsf18 (GITR), Ctla4 and Ikzf4 (Eos).26, 27 This was accompanied by other widespread transcriptional changes suggesting a deviation away from the Treg cell phenotype, including upregulation of proinflammatory Th cell cytokines and downregulation of genes for Treg cell fitness and immunosuppressive functions. These changes are consistent with findings from other studies of the transcriptome of ex‐Foxp3 cells,34, 74 including after deletion of Foxp3 in mature Treg cells.75 Although Foxp3 is important for conferring immunosuppressive capacity and repressing effector cytokine production,23, 76, 77 transcriptional factors other than Foxp3 are required for Treg cell lineage‐specification,25 and Id2, Zeb2, Gata1 and Tcf7 were also differentially expressed in ex‐Foxp3 cells in a direction that could contribute to either failure to commit to or loss of the Treg cell transcriptional program.78, 79, 80 A direct transcriptional comparison of ex‐Foxp3 cells to Tconv cells was not performed, limiting capacity to determine if ex‐Foxp3 cells bear greater similarity to Tconv or Treg cells. However, the marked absence of Treg cell functional markers and gain of some Tconv cell molecules including effector cytokines is reminiscent of Tconv cell types, though this must be investigated in more detail and with functional assays. Freuchet et al,74 showed that human ex‐Foxp3 cells, though clonally expanded from Treg cells, transcriptionally clustered closely to and overlapped with effector memory T cells on a uniform manifold approximation and projection (UMAP) plot, supporting probable Tconv cell functions for this subset.

Surprisingly, ex‐Foxp3 cells did not accumulate in late gestation, nor were they increased after inflammatory challenge with LPS or IL‐1β. The lack of influence of the inflammatory interventions we evaluated on ex‐Foxp3 generation is difficult to reconcile with findings in other experimental settings of increased Treg cell instability and loss of Foxp3 in the presence of inflammation,33 or proinflammatory mediators such as IL‐6.37, 81 Il6 null mutant mice show impaired induction of normal labor and are less susceptible to LPS‐induced preterm birth, while IL‐6 signaling inhibitors are effective in preventing LPS‐induced preterm birth.57, 58 We considered this to suggest potential for IL‐6‐mediated changes in T cells contributing to pathology at this critical phase of pregnancy.55 However, the current data do not support this postulate, and instead indicate that Treg cells are a highly stable lineage with unique ability to withstand lineage conversion during inflammatory states. This state may be entrained by Treg cell exposure to the high levels of P4 characteristic of pregnancy.68

It is possible that ex‐Foxp3 cells in udLNs and uterus arise from expansion of rare cells that transiently express Foxp3 without committing to the Treg cell program.34, 84 An alternative interpretation to Treg cell instability or transient Foxp3 upregulation by uncommitted cells is that the ex‐Foxp3 population contains latent Treg cells that have transiently downregulated Foxp3. It has been shown that the ex‐Foxp3 population is heterogeneous, containing cells that do not permanently lose ‘memory’ of their Treg cell identity, but can re‐express Foxp3 upon stimulation.34 Consistent with this possibility, it was recently demonstrated using cell transfer experiments that ex‐Foxp3 latent Treg cells with fetal specificity are retained postpartum, and act to regenerate the Treg cell pool in subsequent pregnancies sired by genetically similar males.85 Unfortunately, the rarity of ex‐Foxp3 cells in the udLN and decidua makes it difficult to perform in vitro restimulation, sorting and TSDR methylation analysis that could confirm whether the ex‐Foxp3 cells in these tissues are, in fact, latent Treg cells. The limited number of decidual ex‐Foxp3 cells was also a constraint for sorting and analyzing them by RNAseq.

In late gestation, ex‐Foxp3 cells were enriched in the uterine decidua compared to the udLNs and spleen. A limitation of this study is that we did not evaluate decidua or other regions of the uterus across the course of gestation, so whether there are gestation‐dependent changes in uterine ex‐Foxp3 abundance remains unknown. However, their relative enrichment in the decidua raises the question of whether this is due to the presence of cognate antigen or other physiological cues that promote lineage switching, homing, or expansion of existing ex‐Foxp3 cells. For example, increased loss of Foxp3 likely as a consequence of strong TCR engagement with self‐antigen in an inflamed setting has been observed in the pancreas of NOD Type I diabetic mice relative to draining LNs.33 While we did not observe increased Treg lineage switching in decidua, cognate antigen may contribute to the accumulation of ex‐Foxp3 cells in the decidua relative to the udLNs and spleen. Furthermore, since Shao et al.,85 found that only fetal‐specific ex‐Foxp3 cells were expanded postpartum, it is possible that examining bulk ex‐Foxp3 cell abundance also limited our ability to detect a change in fetal‐specific ex‐Foxp3 cell generation during first pregnancy. Alternatively, this conversion may occur postpartum during contraction of the immune response.

Whether the ex‐Foxp3 cells we observed in gestational tissues during pregnancy can adopt Th cell or other detrimental functions in pregnancy is an important consideration. Overwhelmingly, the enriched Kegg terms for ex‐Foxp3 cell DE genes were related to Th cells and inflammatory conditions, suggesting that the ex‐Foxp3 cells would be detrimental to immune tolerance. While the exact contributions of IFNγ and IL‐17A to preterm birth pathogenesis are not defined, due to a lack of studies investigating the potential efficacy of blocking these cytokines in preventing preterm birth, numerous studies link increased IFNγ, IL‐17A, Th1 cells or Th17 cells with gestational disorders such as recurrent miscarriage.73, 86, 87, 88, 89 Increased expression of proinflammatory cytokines IFNγ or IL‐17A is often observed in ex‐Foxp3 cells compared with Treg cells,33, 34 which may result directly from loss of Foxp3, which suppresses inflammatory cytokine expression during Treg cell differentiation. Indeed, ex‐Foxp3 cells in the udLNs of pregnant mice in our study expressed increased inflammatory cytokines Ifng, Il4 and IL21. Also, flow cytometry showed that in the udLNs and spleen across the whole of gestation and including in non‐pregnant females, ex‐Foxp3 cells contained the highest proportions of cells expressing IFNγ or IL‐17A relative to both Treg cells and Tconv cells, highlighting their inflammatory potential. However, the Tconv cell population in udLNs and spleen largely consisted of naïve CD4^+^ T cells, so the relatively higher frequency of cytokine‐expressing cells among ex‐Foxp3 cells may be an artifact of selectively analyzing a population that is already enriched for cells likely to have undergone TCR stimulation in the past. Despite this, in the decidua, the proportions of IFNγ^+^ or IL‐17A^+^ ex‐Foxp3 cells were contrastingly low. Instead, Treg cells were the dominant expressers of these cytokines in the decidua, even more than Tconv cells. As such, it is likely that mechanisms to keep ex‐Foxp3 cell activity in check are at play in the uterus, though whether these are the same mechanisms that constrain Tconv activity during pregnancy is unknown.

Contrary to reports that ex‐Foxp3 cells can express granzymes and adopt cytotoxic functions,39, 74 we found that Gzmb and Gzmc genes were downregulated in ex‐Foxp3 cells, although Gzmk, a proinflammatory granzyme,90, 91, 92 was upregulated. Therefore, the ex‐Foxp3 cells in the udLNs of pregnant mice do not appear to be cytotoxic cells. Consistent with our results, deletion of Foxp3 in mature Treg cells has been shown to induce the loss of Gzmb,75 and since Granzyme B is used by Treg cells to constrain NK and CD8^+^ T‐cell responses,93 this is consistent with the loss of immunosuppressive function in ex‐Foxp3 cells.

The significance of RORγt, IL‐17A and IFNγ expression in Treg cells in pregnancy is not clear. The expression of Th transcription factors by Treg cells has been suggested to endow them with properties that increase their stability,94 promote their suppressive capacity more specifically to cognate Th subsets95 or with specialized transcriptional programs and functions such as prevention of excessive Type 1 and Type 2 immune responses by Tbet^+^ and RORγt^+^ Treg cells.29, 31, 96 However, while Th transcription factors are expressed by mature Treg cell subsets, this does not necessarily lead to expression of Th cytokines that are usually repressed by Foxp3,95, 97 indicating complex and context‐dependent transcription factor effects. Further investigation into the significance of Treg cell inflammatory cytokine expression will be important for understanding whether they are dysfunctional or lineage unstable. Of note, expression of Th cytokines has been reported to be higher in Treg cells that show lineage instability and convert to ex‐Foxp3 cells.36 Despite this, the lack of increase in ex‐Foxp3 cells despite increased expression of these cytokines by Treg cells in our study again indicates Treg cell stability in pregnancy and raises the prospect of functions for Treg cell adoption of Th cytokines that remain to be elucidated. Unfortunately, there is currently no method available for specific depletion of ex‐Foxp3 cells, which limits the feasibility of definitively demonstrating their functions.

Due to the importance of Treg cells in pregnancy for inducing and maintaining immune tolerance and promoting fetal and placental growth, there is interest in therapeutic interventions that increase Treg cell abundance or function in pregnancy to remediate recurrent implantation failure, recurrent miscarriage, preeclampsia and preterm birth. However, the stability of Treg cells is an important consideration for assessing the risk of such therapeutic strategies. Our study suggests that the risk of lineage switching in endogenous Treg cells in pregnancy may be low, at least provided that the Treg cells exhibit normal transcriptional and phenotypic qualities. Promisingly, therapeutically targeting IL‐2‐signaling, which is effective in mitigating fetal loss in mice,98, 99 can further stabilize Treg cells and reduce their conversion to ex‐Foxp3 cells.100

A key limitation of this study is that it is not possible in laboratory animals to fully replicate the range of genetic and environmental factors that contribute to potential Treg cell instability in humans. Therefore, it is possible that under other specific conditions, Treg cells may be less stable than demonstrated in the mouse. However, whether intrinsic defects in Treg cells of some individuals lead to Treg cell instability that contributes to pregnancy pathologies should be examined. Nevertheless, caution must be exercised in extrapolating our results to human pregnancy and preterm labor, as there are limitations in the extent to which these models accurately reflect preterm birth in humans. The proinflammatory cascade that precedes preterm labor in humans is not fully understood and is unlikely to be entirely recapitulated by delivery of LPS or IL‐1β alone. Further, it is increasingly clear that the underlying causal factors for spontaneous preterm birth emerge from perturbations arising in early pregnancy, especially to insufficient maternal immune tolerance.82, 83 Therefore experiments involving a longer period of inflammatory exposure, or exposure to inflammatory signals during Treg cell induction in early pregnancy, may reveal an impact on Treg cell stability not apparent in healthy gestation or under acute inflammatory conditions. Importantly, the short duration between inflammatory induction and analysis of the T‐cell response in these models may have limited our ability to detect measurable changes. Onset of labor after LPS or IL‐1β administration was rapid, limiting the timeframe in which T cell analysis could be performed before labor and birth. Other studies investigating Treg cell lineage‐conversion have analyzed results after 3 days in vitro 37 or >1 week post‐stimulation or transfer in vivo.33, 36

In summary, this study shows that under normal circumstances, Treg cells in late gestation display a high degree of stability in late gestation and resistance to inflammatory challenge without the expected increase in conversion to ex‐Foxp3 cells. Our findings suggest that, although insufficiency in Treg cells is associated with preterm labor in women, and mouse studies directly link Treg cell deficiency with increased risk of both spontaneous and inflammation‐induced preterm labor, these declines in Treg cell activity are unlikely to be due to lineage switching. As such, other mechanisms to override Treg cells to enable on‐time birth, and to link pathological Treg cell deficiency with preterm labor, must be at play. Further studies are required to expand understanding of the origins and physiological significance of ex‐Foxp3 cells in pregnancy.

METHODS

Mice

All mice were held and bred in specific‐pathogen‐free conditions at the Adelaide Health and Medical School (AHMS) Animal Facility, University of Adelaide, Australia. All experiments were performed in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes 8th edition 2013 and were approved by the University of Adelaide Animal Ethics Committee (approval numbers M‐2015‐115, M‐2015‐115A, M‐2018‐035, M‐2019‐087). C57Bl/6 female mice and Balb/c males for mating were purchased from Ozgene (Perth, WA, Australia). B6129S‐Tg(Foxp3‐EGFP/cre)1aJbs/J (Foxp3^eGFPCre^) mice62 and B6.Cg‐Gt(ROSA)26Sor^tm(CAG‐tgTomato)Hze^/J (Rosa26^RFP^) mice were purchased from Jackson Laboratories (Bay Harbor, ME, USA; IDs #023161 and #007914, respectively). Rosa26^RFP^ mice were bred from homozygous (RFP/RFP) pairs. Foxp3^eGFPCre^ mice were maintained by crossing hemizygous mice (GFP^+^) with wildtype littermates, since hemizygous x hemizygous breeding was often unsuccessful, consistent with Jackson Laboratories reports. Mice carrying the transgene were noticeably smaller and took longer to reach reproductive age than wildtype littermates. The Foxp3 fate‐mapping strain, Foxp3^eGFPCre^.Rosa26^RFP^, was created by crossing Foxp3^eGFPCre^ mice to Rosa26^RFP^. Experimental females were obtained from the F1 generation of this cross, or from the F2 generation (Foxp3^eGFPCre^.Rosa26^RFP^ x Rosa26^RFP^) after screening for the presence of the GFP transgene either by genotyping in‐house or using Transnetyx genotyping service (transnetyx.com).

Matings

C57Bl/6 females aged 8–12 weeks were used for mating. Foxp3 fate‐mapper females aged 10–16 weeks were used due to their smaller size and delayed reproductive age. The estrus cycle of experimental female mice was tracked by phase‐contrast microscopy of vaginal lavage to determine the cycle stage, and those in proestrus and estrus were mated to proven fertile Balb/c males aged 10 weeks–12 months overnight and checked on the following morning between 0800 and 1100 h for the presence of a copulatory plug, designated GD0.5. The plugged females were separated and housed with other females mated on the same night, with a maximum of 3 females per cage.

Lipopolysaccharide or IL‐1β treatment and monitoring for preterm birth

Females remaining pregnant at GD15.5 as determined by increase in weight were housed individually for inflammatory challenge and preterm birth induction. For optimization of IL‐1β‐induced preterm birth in C57Bl/6 mice, 0.5–8 μg of recombinant human IL‐1β (PeproTech, Rocky Hill, NJ, USA) or 0.1% bovine serum albumin (BSA; Gibco) in PBS (vehicle) control was injected i.p. at 0800–1000 h on GD15.5 in pregnant females. Timing of birth was determined by recording with a Sony HandyCam Video Camera with nightshot mode capability and tripod set up above the mice. Dams were given adequate food and a NECTA H20 gel pack for hydration (Able Scientific, Adelaide, SA, Australia) for the duration of filming. Timing of birth was designated after delivery of the first pup in the litter, and delivery before GD18.5 was considered preterm. The viability of pups was recorded and pup birth weight was measured at 12–24 h after delivery.

For optimization of LPS or IL‐1β‐induced preterm birth in Foxp3 fate‐mapping mice, 2 or 4 μg of LPS (Escherichia coli 0111B4 serotype; Sigma‐Aldrich) or IL‐1β was injected i.p. at 0900 h on GD16.5, or 0900 or 1900 h on GD17.5. Controls received vehicle (0.1% BSA in PBS).

For flow cytometry of the T cell response after IL‐1β challenge in C57Bl/6 mice, 6 μg was injected i.p. at 1900–2200 h on GD15.5 instead, thus delaying labor slightly on GD16.5 to enable collection of tissues prior to signs of parturition. For flow cytometry of the T cell response in Foxp3 fate‐mapping mice, the females were injected with 4 μg of LPS at 0900 h on GD17.5 or 4 μg of IL‐1β at 1900 h on GD17.5.

Pregnancy outcomes and tissue collection

Mice were placed under deep anesthesia by i.p. injection of 15 μL g^−1^ body weight 2% Avertin (2.2.2‐tribromoethanol, Sigma‐Aldrich) to enable blood to be harvested by cardiac puncture using a 20 gauge needle and 1 mL Tuberculin syringe (BD Biosciences) for future analysis. The mice were immediately humanely killed by cervical dislocation. The blood was placed in Eppendorf tubes and rested at room temperature for a minimum of 10 min before centrifuging at 380 g for 10 min. For late‐gestation time‐points, pregnancy outcomes were also assessed. The number of implantation sites was recorded and classified as viable, non‐viable or resorbing/non‐viable. Non‐viable implantations were identified by small size and pale color of fetuses. Resorptions were characterized as small hemorrhagic masses. All viable and non‐viable fetuses were weighed following separation from the fetal membranes and umbilical cord, and placentas were also weighed. The udLN and decidua were collected for flow cytometry. Briefly, the uterine tissue was cut away from placentae and fetuses, and decidual tissue was gently scraped from the uterine myometrium to facilitate resolution of decidual T cells.

Preparation of cells for flow cytometry

Uterine decidua was minced with scissors then digested in 1× Hank's Balanced Salt Solution (HBSS; Gibco) containing 1 mg mL^−1^ collagenase factor VIII (Sigma‐Aldrich), 0.28 mg mL^−1^ hyaluronidase (Sigma‐Aldrich) and 1 mg mL^−1^ BSA for 20 min shaking at 37°C. The suspension was filtered through a 70 μm cell strainer (Corning, NY, USA) and fresh digestion buffer was added to the remaining undigested tissue for a further 20 min before filtering and adding to the first round of digestion. Cells were then incubated with red cell lysis buffer (0.155 m NH_4_Cl, 10 mm KHCO_3_, 99.2 μm ethylenediaminetetraacetic acid (EDTA) disodium salt in reverse osmosis (RO) water) at 37°C for 5 min then rinsed with PBS before enumerating the cells with a hemocytometer and Trypan Blue (Sigma‐Aldrich). udLNs and spleens were gently homogenized to single cell suspensions in PBS through a 70 μm cell strainer using a syringe plunger, or by gently crushing between the frosted ends of two SuperFrost glass slides (HD Scientific Supplies Pty Ltd, Wetherill Park, NSW, Australia). Splenocytes were treated with red cell lysis buffer for 5 min at 37°C then rinsed with PBS.

Flow cytometry

One to two × 10^6^ cells were plated per well in a 96‐well plate (Corning). The cells were stimulated with 20 ng mL^−1^ phorbol 12‐myristate 13‐acetate (PMA; Sigma‐Aldrich, Bayswater, Vic, Australia), 1 nm ionomycin (Invitrogen) and 1/1500 Golgi Stop (BD Biosciences, Macquarie Park, NSW, Australia) for 4 h at 37°C and 5% CO_2_ in Iscove's Modified Dulbecco's Medium (IMDM) containing 10% fetal bovine serum (FBS; Gibco, Themo Fisher Scientific, Waltham, MA, USA), 100 U mL^−1^ penicillin/streptomycin, and 0.002 m L‐glutamine (Gibco, Thermo Fisher Scientific). After stimulation, the cells were rinsed in PBS and stained with viability dye (Fixable Viability Stain 620 or 700, BD Biosciences) for 20 min in the dark. After rinsing again, Fc receptors were blocked with 1/100 Fc block (BD Biosciences) in PBS/0.1% BSA/0.04% NaN_3_ for 15 min at room temperature before adding a master‐mix of fluorophore‐conjugated antibodies against surface antigens (Table 4), then continuing incubation for 25 min on ice. The cells were rinsed then fixed and permeabilized using the Foxp3 Staining Buffer Set (eBioscience) according to the manufacturer's instructions. Fixation was either 30 min at room temperature or overnight at 4°C. Intracellular staining was performed for 30 min on ice. Antibody master‐mixes were prepared in Brilliant Stain Buffer (BD Biosciences). The samples were acquired the same day or fixed in 1% methanol‐free formaldehyde in PBS and acquired the following day on a BD LSRFortessa X‐20. The gating strategy used for measuring the different T cell subsets and their expression of cytokines is shown in Supplementary figure 10.

RNAseq of Treg and ex‐Foxp3 cells

On GD18.5, udLNs were dissociated to single cell suspensions in PBS/5% fetal bovine serum (FBS) using scissors and a plunger through a 70 μm cell sieve (Corning). Cells were kept on ice in the dark at all times. They were stained and blocked with Fixable Viability Stain FVS780 (1/1000, BD Biosciences) and FC receptor block (1/100, BD Biosciences) in PBS/2% FBS, followed by staining with anti‐CD45:BUV395 and anti‐CD4:BV421 antibodies (Table 4) prepared in Brilliant Staining Buffer Plus (BD Biosciences). The cells were rinsed and resuspended in IMDM/5% FBS for sorting using a BD FACSAria Fusion. Treg cells and ex‐Foxp3 cells were sorted into IMDM/20% FBS, then centrifuged, rinsed with PBS and counted before finally resuspending in Buffer RLT Plus (RNeasy Micro Plus kit, Qiagen) with 10 μL mL^−1^ β‐mercaptoethanol for storage at −80°C. The samples were thawed for RNA extraction using the RNeasy Micro Plus kit (Qiagen) according to the manufacturer's instructions. RNA purity was checked with a Nanodrop and quality was checked with an Agilent Tapestation. RNA from five mice with RIN >8.3 was pooled to generate each sample. >6 ng of RNA per sample was sent to NovogeneAIT Genomics Singapore Pte Ltd (Novogene (HK) Company Limited) for library preparation with the Clontech™ SMARTer Ultra Low Input RNA kit and paired‐end 150 base‐pair sequencing on an Illumina NovaSeq 6000.

RNAseq data analysis

Read quality was checked with FastQC v0.12.1 (Babraham Bioinformatics, Cambridge, UK) before and after adapter trimming with bbduk v39.03. Reads were aligned to the GRCh39 mm39^−1^ mouse reference genome using STAR v2.7.10b_alpha_230301‐GCC‐11.2.0 and indexed with SAMtools v1.17‐GCC‐11.2.0. Median library size was 36 million reads (range 33–41 million reads), and with a threshold of 0.5 counts per million in at least 5 samples, 13 396 total genes were measured. DGE analysis was performed in R v4.3.3 and R studio v2023.12.1 + 401 using the limma v3.58.1101 and voom 102 packages and a log fold‐change of 1 (equivalent to 2‐fold difference in expression) and adj.P <0.05 were used to determine significance. The top 100 most significantly DE genes were used for pathway analysis. The R package ClusterProfiler v4.10.164 was used to measure enrichment of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (Kegg) terms, and the simplify (cut‐off 0.6) function was used to reduce redundancy in the list of enriched terms. A BH‐adjusted P value of ≤0.05 was used to determine statistical significance.

Statistical analysis

Flow cytometry data and preterm birth outcomes data were analyzed using two‐sided unpaired t‐tests or one‐way ANOVAs as appropriate in GraphPad Prism 10.1.0 (GraphPad Software, San Diego, CA, USA). Fetal, placental and pup weight data were analyzed using a mixed model linear repeated measures ANOVA with post‐hoc LSD test with mother as subject and litter size as designated covariate in SPSS software (IBM Corporation, Armonk, NY, USA).

AUTHOR CONTRIBUTIONS

Kerrie L Foyle: Data curation; formal analysis; investigation; visualization; writing ‐ original draft, review and editing; Ella S Green: Conceptualization; funding acquisition; formal analysis; investigation; writing ‐ review and editing; Jessie R Walker‐Rogers: Formal analysis; investigation; writing ‐ review and editing; Ha M Tran: Data curation; formal analysis; writing ‐ review and editing; David M Olson: Conceptualization; funding acquisition; Lachlan M Moldenhauer: Conceptualization; writing ‐ review and editing; Sarah A Robertson: Conceptualization; funding acquisition; project administration; supervision; writing ‐ review and editing.

CONFLICT OF INTEREST

The authors have declared that no conflict of interest exists.

Supporting information

Supplementary figure 1–10