REVIEW ARTICLE
Emerging Themes in Uterine Natural Killer Cell Heterogeneity and Function Kota Hatta1,2, R. John MacLeod3, Scott A. Gerber4, B. Anne Croy3 1
Division of Cardiovascular Surgery, Toronto General Hospital Toronto, Toronto, ON, Canada; Institute of Medical Sciences, University of Toronto, Toronto, ON, Canada; 3 Department of Biomedical and Molecular Sciences, Queen’s University, Kingston, ON, Canada; 4 Department of Microbiology and Immunology, University of Rochester Medical Center, Rochester, NY, USA 2
Keywords Angiogenesis, blood pressure regulation, decidua, gestational programming, mouse, pregnancy, ultrasound, Wnt signaling
Problem Understanding of uterine natural killer (uNK) cell functions during normal pregnancy remains incomplete.
Correspondence B. Anne Croy, Department of Biomedical and Molecular Sciences, Queen’s University, Room 924 Botterell Hall, Kingston, ON, Canada K7L 3N6. E-mail:
[email protected]
Method of study Cloud tag analysis of literature was used to document themes addressed experimentally for uNK cells. Immunohistochemistry, including wholemount staining of early implantation sites, separation of uNK cells into molecularly distinct subsets, and physiologic measurements in normal and mutant mice, are further advancing understanding of uNK cell biology.
Submission February 24, 2012; accepted April 23, 2012. Ciatation Hatta K, MacLeod RJ, Gerber SA, Croy BA. Emerging themes in uterine natural killer cell heterogeneity and function. Am J Reprod Immunol 2012; 68: 282–289
Results Literature analyses revealed three key, current uNK cell research themes: angiogenesis, spiral arterial remodeling/pre-eclampsia/hypertension and infertility. UNK cells are being defined as cells potentially regulated by Wnt signaling that are heterogeneous in progenitor source and function and make unique contributions to implantation site development prior to spiral arterial remodeling.
doi:10.1111/j.1600-0897.2012.01160.x
Conclusion Future studies are poised to define uNK cell progenitor cells, identify the signaling pathways supporting established uNK cell functions and move current understanding of mouse uNK cells to clinical research questions.
Introduction Uterine natural killer (uNK) cells are distinct lineage, innate lymphoid cells with phenotypic and functional differences to blood NK cells. Human uNK cells are commonly identified as CD56brightCD16 granulated lymphocytes. Murine uNK cells are commonly identified as Dolichos biflorus agglutinin lectin (DBA)+ cells that become very large and heavily granulated. Periodic acid Schiff’s (PAS) reagent stains
mouse uNK cells more comprehensively than DBA1, but its use prevents downstream RNA study. Although NK cells are minor blood lymphocytes (approximately 7–15%), they are the dominant lymphocytes of early decidua. Their peak abundance in decidual cell suspensions is 56-84% of CD45+ cells at weeks 12–16 in humans2–4 and approximately 50% in gestation day (gd)5-10 mice (Croy BA, Chen Z, Hofmann A, Lord EM, Sedlacek AL, Gerber SA; manuscript in preparation). UNK cell expansion American Journal of Reproductive Immunology 68 (2012) 282-289
282
ª 2012 John Wiley & Sons A/S
MOUSE UTERINE NATURAL KILLER CELLS
requires progesterone-triggered decidualization of endometrial stroma, not recognition of non-self, paternal or embryonic antigens. Many but not all uNK cells express perforin,5 but, contrary to their name, their normal functions contribute to an endometrial environment supportive of pregnancy. Murine uNK cells arise from transplantable (circulating) progenitors,1,6–8 and data have been published that suggest trafficking of human uNK cells to early decidua,9–11 although in both species precursor and progenitor cells remain undefined. While uNK cells express Il22,11 they differ from lymphoid tissue inducer (LTI) cells because they lack CD127 until late in their lifespan.12 The anatomic confinement of uNK cells to transient, dynamically changing endometrial niches makes study and understanding of uNK cells challenging. Human uNK cells proliferate during the menstrual cycle and, by the late secretory phase, comprise at least 30% of all endometrial cells.13 With pregnancy, uNK cell expansion continues, reaching maximum numbers in late first trimester. In non-menstruating species with hemochorial placentation such as mice, uNK cells appear with postimplantation decidualization. Here too, uNK cell expansion does not depend upon the presence of a conceptus because uNK cells proliferate in the deciduomata of pseudopregnancy.14 Human uNK cells are not found in fetal, neonatal15 or postmenopausal uteri.16 NK cells first appear in mouse uteri after birth but prior to puberty. However, mature uNK cells do not differentiate until after decidual induction.17 The rodent cells now known as uNK cells had earlier names, most recently granulated metrial gland (GMG) cells. Eventually, the newer ‘uNK’ designation was adopted, and the transient, anatomic metrial gland became known as the mesometrial lymphoid aggregate of pregnancy (MLAp).18 Earlier research applied blood NK cell theories to uterine biology, leading to the hypothesis that the cytotoxic potential of uNK cells was dangerous for conceptuses and actively suppressed. Cytotoxicity of dysregulated/activated uNK cells was considered to be important in causing pathologies such as implantation failure and spontaneous abortion. More recently, uNK cells, also described as activated, are thought to contribute to major pregnancy-associated changes in endometrial structure, including angiogenesis and spiral artery remodeling. UNK cell actions also modulate placentation through chemokine signaling and other
mechanisms.19 The popular themes in uNK cell research over the past 50 years were identified by cloud tag data searching methodology (http://tagcrowd. com/). Since 1961, 765 manuscripts discussed/ reviewed uNK cells; Table I summarizes the past half century’s activity. Over the past 36 months, three areas received increased attention (i) angiogenesis, (ii) spiral arterial modification/pre-eclampsia (PE)/ blood pressure regulation and (iii) infertility. Here, we review selected aspects of these more active areas of investigation to highlight newer insights from our own work and converging theories concerning uNK cells. Table I PubMed-Queried Publication Distributions. Manuscripts on uNK Cells were Queried Using PubMed, and 765 Papers were Found Since 1961, of Which 157 were Published in the Past 3 Years (January 27, 2009–January 27, 2012). The Fold Increase in Topic Share was Determined by Comparing the Percentage of uNKRelated Papers on a Given Topic Between 1961–2012 and 2009– 2012
PubMed queried topic (‘+’ denotes uNK subtopics) uNK, dNK, eNK, GMG cells +Pregnancy +Decidua, decidualization +Implantation +Trophoblast invasion, placentation +Pregnancy loss, abortion, miscarriage +Angiogenesis +Tolerance +Spiral artery +Cancer +Preeclampsia +Aging, menopause +Non-pregnant, virgin +Endometriosis +Infertility +Stem cells, progenitors +Infectious disease +Deciduoma, pseudopregnancy +Blood pressure, hypertension +Autoimmunity
Manuscripts 1961–2012
Manuscripts past 3 years
Fold increase in topic share
765
157
1.00 (base)
614 426
113 93
0.90 1.06
236 116
44 24
0.91 1.01
113
29
1.25
66 66 61 57 53 50 39 38 37 28 26 23
31 19 22 9 21 7 10 5 13 4 6 1
2.29 1.40 1.76 0.77 1.93 0.68 1.25 0.64 1.71 0.70 1.12 0.21
10
6
2.92
10
2
0.96
dNK, decidual natural killer; eNK, endometrial natural killer; GMG, granulated metrial gland; uNK, uterine natural killer.
American Journal of Reproductive Immunology 68 (2012) 282-289 ª 2012 John Wiley & Sons A/S
283
HATTA ET AL.
Materials and methods Literature Analyses PubMed-indexed manuscripts were queried using Boolean search for: ‘uterine NK’ OR ‘uterine natural killer’ OR ‘decidual NK’ OR ‘decidual natural killer’ OR ‘endometrial NK’ OR ‘endometrial natural killer’ OR ‘granulated metrial’ OR ‘metrial gland NK’ OR ‘metrial gland natural killer’. Tag cloud was created from abstracts of papers indexed as reviews using the Web site (http://tagcrowd.com/). Mice Animals were ethically used under protocols approved by University Animal Care Committee (UACC), Queen’s University. Timed syngeneic matings of normal (WT) C57BL6/J (B6), CD1 or BALB/ c-Rag2 / Il2rg / (NK , T , B ; alymphoid) females were used with the day of the copulation plug counted as gd0.5. B6 mice were purchased from The Jackson Laboratory, Bar Harbor, ME, CD1 from Charles River Laboratories, St. Constant, QU, and alymphoid mice were bred at Queen’s University. Histology Immunohistochemistry of living whole-mounted20 or fixed, paraffin-embedded implant sites was conducted. Whole-mount studies employed Phycoerythrin-conjugated anti-CD31 antibody (Clone MEC13.3; BD Pharmingen, Toronto, ON, Canada). Studies of paraffin-embedded tissue used DBA lectin (Sigma, St. Louis, MO, USA),21,22 goat-anti-Dkk1 (R&D), mouseanti-Krm2 (R&D, Minneapolis, MO, USA), rabbitanti-Ror2 (Cell Signaling) and rabbit-anti-Wnt5a (Cell Signaling, Burlington, ON, Canada) antibodies. Results and discussion Angiogenesis Uterine natural killer cell study is logistically difficult because of the lineage’s transient nature and terminal differentiation. Mouse uNK cells are difficult to maintain in vitro because they are very large, fragile, hormone and stromal cell–dependent cells that, upon death, spontaneously release granules and molecules toxic to other cells. Earlier work on uNK cells was therefore largely histologic. In the past
15 years, improvements in culture techniques (Il15 supplementation,23 stromal support such as transfected OP9 cells,24 hanging drop cultures25), transcript quantification from laser capture microdissected (LCM) uNK cells22 and the availability of various KO mouse models26 have advanced our understanding of uNK cell biology. The original studies of implant site architecture in mice genetically deficient in uNK cells gave important clues regarding uNK cell functions. Implant sites in these mice had abnormal arterial endothelium and other vascular anomalies from gd6, poorly modified spiral arteries from gd10 and hypomorphic decidual cells.6,27 Although mechanisms were not understood beyond a major role for Ifng in changing gene expression in the various target cell types, as well as in uNK cells themselves,7 the idea emerged that uNK cells made critical contributions to normal decidual structure and overall quality of pregnancy. Uterine natural killer cells produce pro-angiogenic molecules, including vascular endothelial growth factor (Vegf) family members and placenta growth factor (Pgf). This was first reported by Wang et al.28 and has been confirmed by others using histologic techniques, transcript profiling and functional assays.19,29–32 The pro-angiogenic properties of NK cells are not limited to the uterine environment but appear to be an endogenous characteristic of this lineage. Human choriocarcinoma cells mixed with uNK cells grow larger tumors with much greater endothelial cell infiltration in xenogeneic hosts than the same tumor cells mixed with blood NK cells, which are also tumor-promoting relative to placebo conditions.19 VEGF promotes angiogenesis through its actions on endothelial tip cells.33 It additionally acts though VEGFR2/PDGFRB complexes as a negative regulator of vascular smooth muscle cells and pericytes, inhibiting blood vessel maturation.34 Recent studies showed that VEGF acts indirectly on endothelial cells via intermediary molecules delta-like ligand 1 (Dll1) and Dll4 that bind to Notch. Dll1, responsible for induction of migratory, non-dividing tip cells, is not expressed by endothelium.35 Dll4 is expressed by endothelium, and its binding by Notch leads to lateral signals to adjacent endothelial cells, turning them into proliferative stalk cells that lengthen the vessel being led by a tip cell. Dll1 induces the expression of ephrin B2 (Efnb2), an essential cofactor of Vegf for tip cell induction. Vefg+ mouse uNK cells are Efnb2+ until gd8 suggesting they too may respond to Dll1 signals from stromal American Journal of Reproductive Immunology 68 (2012) 282-289
284
ª 2012 John Wiley & Sons A/S
MOUSE UTERINE NATURAL KILLER CELLS
cells and may establish a Vegf gradient for tip cells (Fig. 1). UNK cells lose Efnb2 expression approximately 24 hr before endothelial cells in spiral arteries show its loss as an antecedent to modification.36 Although angiogenesis is widely considered a hallmark of uNK cells, Bany et al.37 recently reported contradictory data. Implantation sites were compared between gd7.5 Il15 KO (NK cell deficient) and WT mice using gene array profiling and quantitative PCR. Expression of many key genes was not Il15 dependent, leading the authors to conclude that uNK cells do not have major roles in decidualization or angiogenesis before spiral arterial remodeling.37 Others have previously described38,39 spiral arterial pathology (decreased lumen size and thicker spiral arterial walls) in gd10 and 12 Il15 KO mice. Thus, study timing may partly explain the differing conclusions. Alternatively, genome profiling may miss key features that can be revealed by other techniques. Electron microscopic (EM) studies of early gd6-8 implantation sites of a different uNK cell-deficient mouse clearly showed that organization of vessels from gd6 was abnormal.27 Our current work (Croy
BA, Chen Z, Hofmann A, Lord EM, Sedlacek AL, Gerber SA; manuscript in preparation) using wholemount staining19 of viable, early WT and Rag2 / / Il2rg / implantation sites supports the EM study and the conclusion that uNK cells have important roles in early implant site angiogenesis. From studies that included morphometric measurement and were conducted on five strains of mice between gd4.5 and 9.5 (n = 2–5 dams/time point and multiple implant sites/dam), we found that lymphocytes contribute to early angiogenesis in the decidua basalis and to subsequent pruning of these nascent vessels as they mature (Fig. 2). Many authors have suggested that uNK cells are not functionally homogeneous but represent distinct subsets.11,40,41 We found that PAS stain identified all mouse uNK cells and that PAS+ uNK cells could be subdivided based on DBA reactivity.1 The four microscopically defined mouse uNK cell maturation steps21 occur in both PAS+DBA and PAS+DBA+ populations. PAS+DBA uNK cells are the major
(a)
(b)
(c)
Fig. 1 Proposed model for tip cell growth in uNK cell-endriched decidua basalis. Angiogenesis is tightly regulated in a Vegf gradient (red dots) by the Notch signaling pathway. Endothelial cells express the receptor Notch. Extrinsic Dll1, usually from stroma, binds to endothelial cell Notch. This induces the cell’s differentiation into a non-proliferative tip cell. Tip cells form filopodia that arborize from the induced cell. Dll1 binding also induces Dll4, Vegfr and Efnb2 (green dots) in the tip cell that regulate differentiation of adjacent endothelial cells into proliferative stalk cells that lengthen the new vessel. In this model, uNK cells contribute from gd5.5 to the Vegf gradient and to Efnb2 signaling in decidua basalis where the most extensive and complex decidual angiogenesis is seen using whole-mount staining. .
Fig. 2 Alterations of vascular morphology between WT and Rag2 / / Il2rg / decidua as illustrated by whole-mount microscopy. (a) Paraffin-embedded WT implant site reveals details of large but not fine vessels. The boxed area indicates the approximate location of the whole-mount images. In (b) WT and in (c), Rag2 / /Il2rg / implant sites were stained with anti-CD31-PE to label blood vessels and imaged using fluorescence microscopy. WT decidua (b) shows CD31 staining over areas with webbed angiogenesis (wider upper, more mesometrial vessels). Lower vessels in this image show evidence of pruning into a network of finer vessels. Conversely, vessels in Rag2 / Il2rg / decidua basalis (c) measured using IMAGE J software (Media Cybernetics Inc., Bethesda, MD, USA) were larger (39 wider), lacked evidence of pruning and remained immature. This is especially noticeable in the lower part of (c) compared to (b).
American Journal of Reproductive Immunology 68 (2012) 282-289 ª 2012 John Wiley & Sons A/S
285
HATTA ET AL.
gd6.5 population but by gd10.5 represent only one in ten uNK cells. PAS+DBA+ uNK cells were the only phenotype to differentiate from WT marrow in pregnant Rag2 / /Il2rg / recipients.1 Thus, PAS +DBA+ uNK cells arise from cells that home the uterus. To determine whether homed PAS+DBA+ uNK cells differed functionally from PAS+DBA uNK cells, flow sorting with a CD3 CD122+ strategy was used to isolate equivalent cells as PAS staining destroys RNA. We found that DBA+ uNK cells had much higher Vegfa and Pgf relative transcript abundance than DBA- uNK cells, which preferentially made Ifng (Chen Z, Zhang J, Hatta K, Lima PDA, Yadi H, Colucci F, Yamada AT, Croy BA, Submitted). Thus, origins and functions of uNK cell subsets differ. A molecular family that may play roles in uNKpromoted angiogenesis is Wnt. The importance of Wnt signaling crosstalk during implantation,42 decidualization and placentation43 has been reviewed by others. Using immunohistochemistry, we observed in gd6.5, 10.5 and 12.5 implantation sites (n = 3 dams/time point and multiple implant sites/dam) that uNK cells express ligand Wnt5a, its receptor tyrosine-protein kinase transmembrane receptor (Ror2)44 and inhibitor of Wnt signaling, Dickkopfrelated protein 1 (Dkk1), but not Kremen 2 (Krm2), the receptor for Dkk1 that together with Wnt receptor lipoprotein receptor-related protein 6 (Lrp6) blocks Wnt-b-catenin signaling (gd6.5 results shown in Fig. 3). These and other components of Wnt signaling may contribute to uNK cell function and add another layer of regulatory complexity. For example, Dkk1 increases angiogenic potential in endothelial progenitor cells45 and is expressed across ectoplacental cone trophoblasts. Trophoblast additionally, as an epithelium, may be regulated by uNK-derived Ifng in a manner similar to intestinal epithelial cells where proliferation and apoptosis are maintained by Ifng through converging AKT-b-catenin and Wnt-bcatenin pathways.46 Spiral Arterial Remodeling, Pre-Eclampsia and Hypertension Continuous telemetric recording of carotid mean arterial blood pressure (MAP) in pregnant mutant mice suggests roles for NK cells in MAP and for T cells in maternal cardiac adaptations to pregnancy.47 These findings may correlate with observations that lymphocytes have regulatory effects on vascular
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 3 Expression of Wnt signaling molecules in mouse implantation sites. B6 implantation sites were stained with Dolichos biflorus agglutinin (DBA) lectin (green) and Wnt molecules (red). Immunohistochemistry localized Ror2 (a), Wnt5a (b), Dkk1(c) and Krm2 (d) in gd6.5 mouse implantation sites. Arrowheads (a–c) mark uNK cells. Decidual endothelial cells (asterisks) are also marked in (b). Krm2 was restricted to extracellular spaces that occasionally overlapped with uNK cells (asterisks) (d). Krm2 localization in gd6.5 ectoplacental cone trophoblasts (ECP) (e). At gd12.5 it was clearer that DBA+ uNK cells (arrow) did not display detectable Krm2, although other large, apparently granular lymphocyte-like cells were reactive (arrowhead) (f).
structure and vasoconstrictive responses in skeletal muscle.48 Others report49 central T-cell roles in suppressing responses to angiotensin II administration and encourage the study of renal stromal lymphocytes to advance understanding of underlying mechanisms. In women with pre-eclampsia, skewed ratios of Treg to Th17 cells have been linked to their hypertension.50 To advance our work in a clinically relevant manner, we examined normal and Rag2 / / Il2rg / mice across pregnancy using ultrasound. The Rag2 / /Il2rg / mice that lacked spiral arterial remodeling had greater uterine arterial flow velocity to gd16. Fetally driven placental circulation was also disturbed over this interval. Linked outcomes were significant gains in maternal heart weight, stroke American Journal of Reproductive Immunology 68 (2012) 282-289
286
ª 2012 John Wiley & Sons A/S
MOUSE UTERINE NATURAL KILLER CELLS
volume and cardiac output compared to congenic pregnancies with spiral arterial remodeling. Additionally, fetal hearts were impacted. Their rate of beating was much lower than in the congenic control fetuses, suggesting altered structure to provide increased cardiac output. Placental thickness was also diminished.51 These data link incomplete spiral arterial modification and, indirectly, actions of uNK cells to gestational programming risks.52 Analyses of kidney, including renal stromal lymphocytes,53 are again necessary for full understanding of gestational circulatory control. Infertility There is very little infertility reported among various mutant strains that lack uNK cells. However, these mice are reared in artificial conditions that restrict commensal and pathogenic microbial exposure. Most of the publications addressing uNK cells in relation to infertility are from human studies. This literature suggests the major outcome of uNK cell function is pregnancy success. From a series of genotyping analyses, Hiby and her collaborators have strongly advanced the hypothesis that the genetic match between the mother and conceptus is most healthy when trophoblast expresses HLA-C alleles that activate maternal uNK cells.54 The same genetic relationship also protects against later pregnancy complications such as pre-eclampsia.55 A subset of patients with recurrent pregnancy loss is now recognized as superfertile. These women conceive easily (