Immunology and Inflammation

TGF-β drives the conversion of conventional NK cells into uterine tissue-resident NK cells to support murine pregnancy

Josselyn D Barahona
Liping Yang
D Michael Nelson
Wayne M Yokoyama author has email address

Division of Rheumatology, Department of Medicine, Washington University School of Medicine, St. Louis, United States
Department of Obstetrics and Gynecology, Washington University School of Medicine, St. Louis, United States

https://doi.org/10.7554/eLife.109878.2

Open access
Copyright information

Figures and data

Peripheral cNK cells extravasate into the pregnant uterus and acquire a uterine trNK cells phenotype.
(A) Representative flow plots depicting the presence of non-vascular CD49b⁺ Eomes⁺ cNK cells within the gravid uterus of wildtype mice intravascularly labeled with anti-CD45.2 antibody in vivo at gds 6.5 and 14.5 (gd 6.5: C57BL/6 dams, n=3, implantation sites n=9; gd 14.5: C57BL/6 dams, n=3, implantation sites n=9). Gating strategy: Live, Single Cells; CD3^- CD19^- CD45.1^- CD45.2–PE-Cy7^- CD45.2–Pacific Blue⁺ NK1.1⁺ NKp46⁺ cells. (B) Absolute cell counts of non-vascular CD49b⁺ Eomes⁺ cNK cells within the gravid uterus of wildtype mice at gds 6.5 and 14.5. (C) Concatenated flow plots of implantation sites showing that adoptively transferred cNK cells in pregnant uterus of wildtype dams upregulate CD49a and down regulate CD49b by gd 10.5, acquiring a CD49a⁺ CD49b^- Eomes⁺ phenotype characteristic of uterine trNK cells (C57BL/6 dams n=4). Here, 2.5x10⁶ CD45.2⁺ CD3^- CD19^- NK1.1⁺ NKp46⁺ CD49b⁺ splenic cNK cells were adoptively transferred into pregnant C57BL/6-CD45.1 dams at gd 0.5, and the receptor profile of these cells was subsequently assessed at gd 10.5. Gating strategy: Live, Single Cells; CD3^- CD19^- CD45.1^- CD45.2–PE-Cy7^- CD45.2–PE⁺ NK1.1⁺ NKp46⁺ cells. (D) Proportion of uterine ILC subsets derived from adoptively transferred splenic cNK cells in the pregnant uterus of wildtype dams. Statistics were calculated using unpaired t tests with the Mann-Whitney correction. Error bars indicate SEM; *** p < 0.001.

Loss of TGF-β Signaling in Ncr1-expressing cells impairs uterine trNK cell differentiation in pregnant mice.
(A) Representative histograms depicting TGF-β Receptor II expression on splenic NK cells from virgin TGF-βRII^Ncr1Δ and wildtype mice as well as splenic and uterine NK cell subsets from pregnant wildtype mice at gd 10.5 (virgin TGF-βRII^Ncr1Δ mice, n=2; virgin mice: C57BL/6, n=5; gd 10.5: C57BL/6 dams, n=8, implantation sites n=8). MFI, median fluorescent intensity. Gating strategy: Live, Single Cells; CD3^- CD19^- CD45.1^- CD45.2⁺ NK1.1⁺ NKp46⁺ cells. (B) Representative flow plots showing the expression of CD49a, CD49b, and Eomes across ILC subsets in the pregnant spleens of littermate control and TGF-βRII^Ncr1Δ dams at gd 6.5 (Littermates, n=6; TGF-βRII^Ncr1Δ, n=5). Gating strategy: Live, Single Cells; CD3^- CD19^-CD45.1^- CD45.2⁺ NK1.1⁺ NKp46⁺ cells. (C) Absolute cell counts of CD49a⁺ Eomes⁺ trNK cells, CD49a⁺ Eomes^- ILC1s, and CD49b⁺ Eomes⁺ cNK cells in the spleens of pregnant littermate control and TGF-βRII^Ncr1Δ dams at gd 6.5. (D) Representative flow plots showing the expression of CD49a, CD49b, and Eomes across ILC subsets in the gravid uterus of littermate control and TGF-βRII^Ncr1Δ dams at gd 6.5 (Littermates, n=6, implantation sites n=54; TGF-βRII^Ncr1Δ, n=5, implantation sites n=15). (E) Absolute cell counts of CD49a⁺ Eomes⁺ trNK cells, CD49a⁺ Eomes^- ILC1s, and CD49b⁺ Eomes⁺ cNK cells in the gravid uterus of littermate control and TGF-βRII^Ncr1Δ dams at gd 6.5. Statistics were calculated using unpaired t tests with the Mann-Whitney correction. Error bars indicate SEM; *** p < 0.001; and **** p < 0.0001.

Impaired TGF-β–dependent uterine trNK cells differentiation leads to adverse pregnancy outcomes characterized by reduced litter sizes.
(A) Number of live pups at first parturition from littermate control and TGF-βRII^Ncr1Δ dams (Littermates, n=7; TGF-βRII^Ncr1Δ, n=7). (B) Pup birth weight in grams (g) from pups birthed by littermate control and TGF-βRII^Ncr1Δ dams (Littermates, n=68; TGF-βRII^Ncr1Δ, n=31). (C) Gestational period in days for littermate control and TGF-βRII^Ncr1Δ dams (Littermates, n=7; TGF-βRII^Ncr1Δ, n=7). Statistics were calculated using unpaired t tests with the Mann-Whitney correction. Error bars indicate SEM; *** p < 0.001.

TGF-β–dependent uterine trNK cell differentiation required for proper spiral artery remodeling and fetal survival.
(A) At gd 10.5, TGF-βRII^Ncr1Δ dams had fewer implantation sites than littermate control dams (Littermates, n=6; TGF-βRII^Ncr1Δ, n=7). (B) Fetal resorption rates in littermate control and TGF-βRII^Ncr1Δ dams at gd 10.5, showing increased resorptions in conditional knockout dams (Littermates, n=6; TGF-βRII^Ncr1Δ, n=7). Resorption rates (RR) were calculated as: RR(%) = (number of resorbed implantation sites/number of total implantation sites) X 100. (C) Representative images of gd 10.5 decidual spiral arteries from three littermate control and three TGF-βRII^Ncr1Δ dams stained with Masson’s Trichrome (Littermates, n=6; TGF-βRII^Ncr1Δ, n=7; Scale bar, 100μm). (D) Spiral artery wall-to-lumen ratio at gd 10.5 implantation sites from littermate control and TGF-βRII^Ncr1Δ dams. Increased wall-to-lumen ratio in TGF-βRII^Ncr1Δ dams indicative of impaired spiral artery remodeling. (Littermates, n=6, decidual spiral arteries n=257; TGF-βRII^Ncr1Δ, n=7 decidual spiral arteries n=305). Statistics were calculated unpaired t tests with the Mann-Whitney correction. Error bars indicate SEM; ** p < 0.01; and **** p < 0.0001.

Adoptive transfer of splenic cNK cells partially reconstitutes uterine trNK cells and rescues midgestational pregnancy defects in TGFBRII^Ncr1Δ dams.
(A) Representative flow plots showing the expression of CD49a, CD49b, and Eomes across CD45.1⁺ ILC subsets in gd 10.5 implantation sites of TGF-βRII^Ncr1Δ dams reconstituted with splenic CD45.1⁺ cNK cells. Briefly, 3.0x10⁶ splenic CD45.1⁺ cNK cells were adoptively transferred into TGF-βRII^Ncr1Δ dams at gd 0.5. By gd 10.5, a portion of adoptively transferred cNK cells in pregnant uterus of TGF-βRII^Ncr1Δ dams upregulated CD49a and downregulated CD49b, acquiring a CD49a⁺ CD49b^- Eomes⁺ phenotype characteristic of uterine trNK cells (Reconstituted (R)–TGF-βRII^Ncr1Δ, n=3, implantation sites, n=25). Gating strategy: Live, Single Cells; CD3^- CD19^- CD45.1⁺ CD45.2^- NK1.1⁺ NKp46⁺ cells. (B) Absolute numbers of CD45.1⁺ ILC subsets in gd 10.5 implantation sites from reconstituted TGF-βRII^Ncr1Δ dams (R–TGF-βRII^Ncr1Δ, n=3, implantation sites, n=25). (C) Proportion of uterine CD45.1⁺ ILC subsets derived from adoptively transferred splenic cNK cells in gd 10.5 implantation sites from TGF-βRII^Ncr1Δ dams (R–TGF-βRII^Ncr1Δ, n=3, implantation sites n=25). (D) Number of implantation sites and (E) fetal resorption rates in reconstituted TGF-βRII^Ncr1Δ dams at gd 10.5 were comparable to those measured in littermate control dams injected intravascularly with HBBS (HBSS–Littermates, n=4; R–TGF-βRII^Ncr1Δ, n=3). Resorption rates (RR) were calculated as: RR(%) = (number of resorbed implantation sites/number of total implantation sites) X 100. Statistics were calculated unpaired t tests with the Mann-Whitney correction.

Ncr1^iCre expression does not alter uterine ILC composition or early pregnancy outcomes.
(A) Representative flow plots showing the expression of CD49a, CD49b, and Eomes across ILC subsets in the spleens of wildtype C57BL/6 CD45.1 and Ncr1^iCre dams at gd 6.5 (C57BL/6 CD45.1, n=2; Ncr1^iCre, n=3). Gating strategy: Live, Single Cells; CD3^- CD19^- CD45.1^- CD45.2⁺ NK1.1⁺ NKp46⁺ cells. (B) Absolute cell counts of CD49a⁺ Eomes⁺ trNK cells, CD49a⁺ Eomes^- ILC1s, and CD49b⁺ Eomes⁺ cNK cells in the spleens of pregnant wildtype C57BL/6 CD45.1 and Ncr1^iCre dams at gd 6.5. (C) Representative flow plots showing the expression of CD49a, CD49b, and Eomes across ILC subsets in the gravid uterus of wildtype C57BL/6 CD45.1 and Ncr1^iCre dams at gd 6.5 (C57BL/6 CD45.1, n=3, implantation sites n=21; Ncr1^iCre, n=3, implantation sites n=21). (D) Absolute cell counts of CD49a⁺ Eomes⁺ trNK cells, CD49a⁺ Eomes^- ILC1s, and CD49b⁺ Eomes⁺ cNK cells in the gravid uterus of wildtype C57BL/6 CD45.1 and Ncr1^iCre dams at gd 6.5. (E) Number of implantation sites and (F) fetal resorption rates in Ncr1^iCre dams at gd 6.5 were comparable to those measured in wildtype C57BL/6 CD45.1 dams (C57BL/6 CD45., n=3; Ncr1^iCre dams, n=3). Resorption rates (RR) were calculated as: RR(%) = (number of resorbed implantation sites/number of total implantation sites) X 100. Statistics were calculated unpaired t tests with the Mann-Whitney correction.

Sign up for email alerts