Dynamic states of cervical epithelia during pregnancy and epithelial barrier disruption

Abstract

The cervical epithelium undergoes changes in proliferation, differentiation, and function that are critical to ensure fertility and maintain pregnancy. Here, we identify cervical epithelial subtypes in non-pregnant, pregnant, and in labor mice using single-cell transcriptome and spatial analysis. We identify heterogeneous subpopulations of epithelia displaying spatial and temporal specificity. Notably in pregnancy, two goblet cell subtypes are present in the most luminal layers with one goblet population expanding earlier in pregnancy than the other goblet population. The goblet populations express novel protective factors and distinct mucosal networks. Single-cell analysis in a model of cervical epithelial barrier disruption indicates untimely basal cell proliferation precedes the expansion of goblet cells with diminished mucosal integrity. These data demonstrate how the cervical epithelium undergoes continuous remodeling to maintain dynamic states of homeostasis in pregnancy and labor, and provide a framework to understand perturbations in epithelial health that increase the risk of premature birth.

Keywords: Biological sciences; Genetics; Genomics; Pregnancy.

Graphical abstract

Figure 1

Morphological changes in cervical epithelia during pregnancy (A) Schematic longitudinal section of mouse uterus and cervix. Blue: stratified squamous epithelia of endo- and ectocervix. Red: columnar epithelium of uterus. Transition zone (TZ): Boundary comprising both types of epithelia. Blue boxed area: Indicates cervical region below TZ collected for single-cell libraries. (B) Morphological changes in cervical epithelia in non-pregnant, pregnant (GD6, GD12, GD15, and GD18), and in labor mice. Shown are H&E staining (top), Ki67 immunostaining for proliferation (middle), and TUNEL staining for cell death (bottom) (red: nuclei, green: TUNEL⁺). E: Epithelia; Str: Stroma. Scale bar 50 μm, Representative images from three independent samples per each group.

Figure 2

Remodeling of cervical epithelia is associated with alterations in transcriptional and open chromatin status (A) (top) Design of scRNA-Seq experiments (gray circles: sequencing time points). (bottom) Distribution of epithelia and non-epithelial cells captured per time point. Each time point is a single replicate. (B) UMAP visualization of all cells captured from all time points of scRNA-Seq. (C) Feature plots indicate the expression of genes used to identify different cell types. Epcam and Ehf: epithelia; Col1a2: stroma; Tyrobp: immune; Pecam1: endothelia, and Hbb-bt: red blood cells. (D) Heatmap showing the expression of cell type-specific genes. (E) Distribution of epithelia and non-epithelial cells captured per time point of scATAC-Seq. Each time point is a single replicate. (F) UMAP visualization of all cells captured from all time points of scATAC-Seq. (G) Feature plots indicate the open chromatin status of genes used to identify different cell types. (H) Genome browser snapshots of open chromatin status for the genes in (G), for each cell type. (I) Heatmap showing the cell type-specific open chromatin status of genes. (J) (top) Feature plot of open chromatin status for Esr1, Esr2, and Pgr. (bottom) Motif deviation scores for these transcription factors indicate cell type-specific enrichment of TF-binding motifs at open chromatin regions.

Figure 3

Shifts in epithelial subtypes and proliferation in early softening (A) Schematic of lineage relationships of cervical epithelial cell types. Basal cells give rise to luminal non-secretory and luminal secretory cells. Luminal secretory cells can mature into non-goblet (Spdef⁻ Muc1⁺) or goblet cells (Spdef⁺, Muc5b⁺). (B) UMAP visualization of epithelial cells from the non-pregnant time point with three major types of epithelia: basal, luminal, and secretory. (C) Feature plots showing the expression of genes differentially expressed between the NP epithelia. Jag2: basal cells; Mki67: cycling basal cells; Dsg1a: luminal cells; Ifitm1, Avil, and Muc1: secretory cells. (D) UMAP visualization of epithelial cells from GD6. (E) Feature plots showing the expression of genes differentially expressed between the different types of GD6 epithelia. Jag2: basal cells; Mki67: cycling basal cells; Krt12: luminal cells; Muc5b, Pigr, and Spdef: secretory cells. (F) Bar chart quantifying epithelial subtypes in NP and GD6. Each time point is a single replicate. (G) Bar chart quantifying proliferating (Mki67⁺) epithelial cells in NP and GD6. (H) IF and IHC showing protein expression of Krt5 (green), and Krt10 (brown) in mouse endocervical epithelia from NP (estrus and diestrus) and GD6. Krt5: basal cells; Krt10: luminal epithelia. RNAscope analysis of Dsg1a (yellow) with Avil (pink) mRNA in mouse endocervical epithelia from NP and GD6. DAPI (blue): nuclei. E: Epithelia; Str: Stroma. Scale bar 50 μm, objective lens 40×. Representative images from two independent samples per group. (I) Expression of Muc5b (yellow) mRNA by RNAscope and Cxcl15 (brown) protein by IHC in mouse endo and ectocervical epithelia from NP (estrus and diestrus) and GD6. DAPI: nuclei. E: Epithelia; Str: Stroma. Representative images from three independent samples per group. (J) Dot plot showing the expression of luminal markers. (K) Heatmap highlighting the different genes expressed in secretory clusters from NP and GD6.

Figure 4

Distinct populations of secretory cells during pregnancy (A) UMAP visualization of epithelial cells from GD12, GD15, and GD18. (B) Feature plots show the markers used to identify the different epithelia clusters. (C) Abundance of goblet subtypes shift between early pregnancy to late pregnancy in scRNA-Seq libraries. Pigr⁺ goblet cells are at their highest level in GD6. Rbp2⁺ goblet cells peak at GD18. Each time point is a single replicate. (D) RNA velocity of the GD12-18 clustering shows strong directionality from luminal cells to Rbp2⁺ goblet cells, but not to Pigr⁺ goblet cells. (E) Heatmap highlighting changes in gene expression across the different secretory clusters identified in the NP (Sec non-goblet and Sec-goblet) and GD6-18 clustering (Goblet 1 and Goblet 2). (F) Spatial analysis of cervical epithelia subtype markers at GD6-18. Basal (Krt5), luminal (Krt10 and Krt12), goblet (Spdef, Muc5b, Muc1, Pigr, and Rbp2). Detection of Krt5 (green) and Krt 10 (brown) protein by IF and IHC. Detection of Krt12, Muc5b (yellow); Pigr, Muc1 (green); Spdef (teal blue) and Rbp2 (red) by RNAscope. DAPI(blue)-nuclei. In contrast to basal/luminal markers, expression of Spdef, Pigr, Rbp2, Muc5b, and Muc1 is restricted to cells close to the lumen. E: Epithelia, Str: Stroma; Scale bar 50 μm, objective lens 40×. Representative images from three independent samples per group. (G) RNAscope images showing co-analysis of Spdef (pink) with Pigr or Rbp2 (yellow) mRNA in mouse endocervical epithelia from GD15. DAPI (gray)-nuclei. E: Epithelia; Str: Stroma.Scale bar 50 μm, objective lens 40×. Representative images from two independent samples per group.

Figure 5

Goblet cell expansion in mid-late pregnancy is followed by induction of epithelial transcription programs in labor to repopulate NP-specific subtypes following birth (A) UMAP visualization of epithelia from all time points. Cells are colored by time point. (B) As in (A), but cells are colored by epithelial type. (C) Feature plots showing the expression of genes that identify epithelial subtypes. (D) Heatmap highlighting temporal and subtype-specific changes in keratin gene expression from basal to luminal to secretory populations across pregnancy. (E) Dotplots showing the expression of secretory markers and mucins in the different secretory clusters. (F) Gene Ontology analysis highlights the functional changes in the luminal and secretory populations. (G) Detection of Dsg1a (yellow) and Spdef (teal blue) mRNA expression by RNAscope and Ifitm1(brown) protein expression by IHC on GD18, IL, and NP (estrus and diestrus). DAPI(blue) - nuclei. Inserts show a magnified view of selected areas. E: Epithelia; Str: Stroma.Scale bar 50 μm, objective lens 40×. Representative images from two independent samples per group.

Figure 6

Epithelial dysfunction in HAKO mice is associated with E. coli ascension and inflammation in maternal and fetal tissue (A) H&E staining (top), Ki67 (brown) immunostaining for proliferation (lower). H&E and Ki67 were done in WT and HAKO on GD15 and GD18. The representative images are shown at 60× magnification from three independent samples per group. E: Epithelia; Str: Stroma. Scale bar 50 μm. (B) TUNEL imaging for cell death (red: nuclei, green:TUNEL⁺). TUNEL staining was done using WT and HAKO at GD15, 16, 17, and 18. Thymus from juvenile WT mice was used as positive (with terminal transferase) and negative (without transferase) control. The representative sections are shown at 20× magnification. E: Epithelia; Str: Stroma Scale bar 50 μm. (C) Schematic of the vaginal E.coli inoculation and assessment for E.coli ascension and inflammation. (D) Bacterial growth in uterine cultures from HAKO but not WT 24 h after the vaginal inoculation of E.Coli demonstrates ascension of E.Coli in HAKO. (E) Quantitative real-time PCR analysis of proinflammatory genes (IL-1β, TNF-α, IL-6,and Mmp8) and a marker of uterine contractility (Cx43) expressed in tissues collected from WT (black) and HAKO (red). Expression was evaluated in cervix-vagina, uterus, fetal membrane, and placenta. N = 5 mice per genotype. Bar graphs depict the average relative gene expression ±SEM Target gene expression was normalized to the housekeeping gene Ppib using the 2ˆ-ddCt relative gene expression method (User Bulletin no.2; Applied Biosystems).∗ indicates p < 0.05 (2-tailed ratio-paired Student’s t test).

Figure 7

Altered cell transition states characterize epithelial barrier disruption (A) UMAP visualization of WT and HAKO mice at GD15 and 18. Cells are colored by genotype and time point. (B) As in (B), but cells are colored by epithelial type. (C) Dotplot showing the expression of basal and luminal markers in WT and HAKO cells. Cycling basal, Goblet 1 and secretory subtypes are not distinguished by genotype. (D) Feature plots showing the expression of goblet markers. (E) Violin plots for Krt5 expression. (∗ indicates p < 0.01). (F) Comparison of the spatial and temporal pattern of epithelial cell markers in WT and HAKO at GD15 and GD18. Basal and luminal (Krt5, Jag2, and Trp63), or goblet (Spdef, Olfm4, and Rbp2) transcripts were analyzed by chromogenic or fluorescent RNAScope. Panel one: Krt5 (red). Panel two: Jag2 (pink) and Spdef (teal blue). Panel three: Trp63 (yellow) and Olfm4 (green). Panel four: Rbp2 (yellow). DAPI stains nuclei (blue or gray). E:Epithelia; Str: Stroma. Scale bar 50 μm, objective lens 40×. Representative images from three independent samples per each group. (G) The abundance of epithelial subtypes in WT and HAKO mice at GD15–18. (Hypergeometric test p values: p_{BasalCycling(D15 vs WT)} = 9.48E–121, p_{Goblet Rbb2 (D15 vs WT)} = 7.41E–18, p_{BasalCycling(D18 vs WT)} = 5.83E–247, p_{Goblet pigr (D18 vs WT)} = 9.63E–74, ∗ indicates p < 0.01.). (H) Feature plot of Olfm4 expression. (I) Olfm4 protein (red) expression in the HAKO cervical epithelia on GD15 and GD18 compared to wild type by IF staining. DAPI stains nuclei (blue). E: Epithelia; Str: Stroma.Scale bar 50 μm, objective lens 20×. Representative images from three independent samples per each group. (J) Violin plot showing loss of Serpina1e expression in goblet 2 cluster in HAKO cells.

Figure 8

Diverse goblet cell populations in mouse cervix during non-pregnant, pregnancy, and in labor Transcriptionally, non-pregnant goblet cells are distinct from the two goblet populations identified in pregnancy. Goblet cells in NP diestrus express the marker Avil. In early pregnancy (GD6), goblet cells are Pigr⁺ and lack Avil expression. Two distinct goblet populations are identified on gestation days 12 and 15. One is Pigr⁺ and the other is Rbp2⁺. On GD18, Goblet 2 population is more abundant than Goblet 1. Consistent with terminal differentiation of goblet cells in labor, transcriptional programs that drive goblet cell differentiation are lost and transcriptional programs in progenitor cells revert back to NP cell subtypes.