Immunological Adaptations to Pregnancy in Women with Type 1 Diabetes

Despite adequate glycemic control, pregnancy outcome of women with type 1 diabetes (T1D) is still unfavorable as compared to healthy women. In a rat-model of T1D under normoglycemic conditions, adverse pregnancy outcome was also observed, which was associated with aberrant immunological adaptations to pregnancy. Because similar processes may occur in women with T1D we studied the systemic immune response in non-pregnant and pregnant women with and without T1D. The systemic immune response was assessed by using flow cytometry to evaluate the number and activational status of subpopulations of lymphocytes, Natural Killer cells and monocytes in peripheral blood of non-pregnant and pregnant women with and without T1D. An increased white blood cell count, an increased Th1/Th2 ratio, increased Natural Killer cell expression of CD335 and enhanced activation of intermediate and non-classical monocytes was observed in pregnant women with T1D vs. healthy pregnant women. Also, the pregnancy outcome (i.e. incidence of preterm delivery and macrosomia) of women with T1D was unfavorable as compared to healthy women. This study showed that in T1D, the immunological adaptations to pregnancy are disturbed. In addition to hyperglycemia, these different immunological adaptations may be responsible for the greater frequency of complications in pregnant women with T1D.

Despite tight glycemic control, frequent episodes of hyperglycemia still occur during pregnancy in type 1 diabetic (T1D) patients 1 . Up to now, these hyperglycemic episodes have been thought to be responsible for the increased incidence of pregnancy complications 2 . However, recent work suggests that more factors may be involved 3 . We have shown that BBDP-rats, a rat-model of T1D, undergo aberrant immunological adaptations during pregnancy, which are associated with pregnancy complications 3 . Similar events may occur in autoimmune T1D pregnant women.
Normal pregnancy is accompanied by various immunological adaptations facilitating implantation, placentation and tolerance of the semi-allogeneic fetus [4][5][6] ; aberrations in these adaptations are associated with pregnancy complications, like miscarriages, preeclampsia and preterm delivery [7][8][9] . Normal systemic adaptations to pregnancy include a shift towards a type 2 immune response in T-lymphocytes 4,10 and Natural Killer (NK) cells 11,12 and an increased frequency of regulatory T-cells (Treg) during the first and second trimester of pregnancy 10,11 . Moreover, normal pregnancy is accompanied by a generalized activation of the inflammatory response, characterized by activation of monocytes and granulocytes, showing increased expression of various activation markers and altered cytokine production 13,14 . The proinflammatory condition of pregnancy is further supported by the fact that an increased amount of intermediate monocytes are found in pregnancy. Intermediate monocytes are a separate monocyte subset, a transitional state between classical and non-classical monocytes, of which the function is not exactly known yet 15 . However, their numbers are increased in pro-inflammatory conditions, such as preeclampsia and other inflammatory diseases 15,16 . It is unknown which immunological adaptations to pregnancy occur in women with autoimmune T1D.
The immune response in T1D patients is characterized by a type 1 immune response 17 , impaired function of regulatory T-lymphocytes 18 and increased mRNA expression of interferon-gamma (IFN-γ ) 19 , when compared to healthy individuals. In view of this different immune response in T1D women, we hypothesized that pregnancy-induced immune adaptations differ between T1D and healthy pregnant women. The aim of this study was therefore to assess the systemic immunological adaptations to pregnancy in women with T1D. Understanding these aberrant changes may be a first step toward designing new means, in addition to glycemic control, for preventing pregnancy complications in T1D pregnant women.

Materials and Methods
Patients. This study was approved by and carried out in accordance with the guidelines of the ethic committees of the University Medical Center Groningen (UMCG) and Martini Hospital Groningen (NL30779.042.09) and registered at the Dutch trial register (NTR2195). All participants provided informed consent.
Four groups of women (age 18-40 years) were included in the study: healthy non-pregnant (n = 16), healthy pregnant (n = 19), non-pregnant with T1D (n = 19) and pregnant with T1D (n = 21). Healthy non-pregnant women were recruited from personnel at the UMCG and healthy pregnant women from the midwifery clinic. Non-pregnant and pregnant women with T1D were recruited from the diabetes outpatient clinics of the UMCG and the Martini Hospital Groningen. Non-pregnant women filled in a questionnaire to obtain information about their cycle, contraception and last menstrual bleeding. Blood samples were obtained by venous punction into a 10 ml EDTA tube (BD-Plymouth, UK) from non-pregnant women within 10 days after the start of the last cycle and from pregnant women at 30 weeks of gestation. All pregnant women were followed up during pregnancy and 6 weeks postpartum to evaluate pregnancy outcome. For the healthy pregnant women, exclusion criteria were gestational diabetes mellitus, known autoimmune, cardiovascular or other active diseases, > 2 previous miscarriages, maternal/fetal complications during pregnancy (i.e. preeclampsia/HELLP, intra-uterine growth restriction, prematurity, perinatal death, congenital malformations and neonatal death). For the healthy non-pregnant women, exclusion criteria were known active diseases. For both diabetic groups, exclusion criteria were HbA1c > 7.5% (at the time of sampling), renal failure (serum creatinine > 120 μ mol/l) and active treatment for autoimmune disease.
Outcome measures. The basic characteristics and pregnancy outcome variables are shown in Table 1.
Sample processing. Human blood leukocyte counts. 100 μ l of blood was used to measure total white blood cell numbers (WBC), using a microcell counter (Sysmex PocH 100i, Sysmex, Etten-Leur, The Netherlands).
Monocytes antibody cocktail. PerCP/Cy5.5 labeled anti-human CD14 (clone HCD14), Pacific Blue labeled anti-human CD16 (clone eBioCB16, eBioscience), FITC labeled mouse anti-human HLA-DR, DP, DQ (i.e. MHC-II; clone Tu39, BD Pharmingen, Breda, Netherlands), PE labeled anti-human CD62L (clone DREG-56) and PE labeled mouse IgG1 isotype control (clone MOPC-21). Sample labeling. We used antibody cocktails for T-lymphocytes, NK-cells and monocytes. Immediately after sampling, whole blood was mixed with RPMI-1640 (1:1) and aliquoted, centrifuged and aspirated. Subsequently, all tubes were incubated with different antibody-cocktails in the dark for 30 min and washed with 500 μ l washing buffer, followed by red blood cells lysis with lysing buffer in the dark for 30 minutes. After centrifugation, aspiration and washing, tubes for lymphocyte/lymphocyte isotype were incubated with fixperm (from FoxP3-staining buffer set) in the dark for 30 minutes. To the other tubes, 300 μ l washing-buffer was added and stored. After 30 min incubation in the dark, tubes for lymphocyte and lymphocyte isotype staining were washed with perm (from FoxP3-staining buffer set) followed by incubation with the FoxP3, Tbet and GATA-3 antibodies or their isotype controls for 30 min in the dark at RT and washed with perm. Finally, 300 μ l washing buffer was added to these tubes. All tubes were stored in the dark at 4 °C until measurement by flow cytometry within 24hr.
Flow cytometry. Cells were analyzed using the BD LSR-II flow cytometer (BD Biosciences, Breda, the Netherlands). At least two hundred fifty thousand events were measured and saved. Analysis was performed using FlowJo 7.6.1 (Tree Star, Inc., Ashland, OR, USA).
Data analysis. Differential cell counts. First, different leukocyte populations (i.e. lymphocytes, monocytes and granulocytes) were assessed, using a forward-/sideward scatterplot (FSC/SSC) of the tube stained for NK-cells. A gate was set around the leukocytes and secondary gates were set around the three different subpopulations in a new FSC/SSC plot (Fig. 1, panel A).  Statistics. Continuous parameters were expressed as mean ± SEM in case of normally distributed data, or as median (Q1-Q3) in case of skewed distribution. Independent T-tests and Chi-square-tests were used for categorical variables (corrected for multiple testing). Two-way ANOVA (with interaction effects) was used to analyze diabetes-and/or pregnancy-induced differences in immune cells. Before performing this two-way ANOVA, normality of the data was tested using the Kolmogorov-Smirnov test. If data were distributed abnormally, log-transformation was used. For all analysis, p-values < 0.05 were considered as statistically significant and p-values between 0.05-0.1 were considered as a statistical trend. Statistical analyses were performed using SPSS for windows version 20.0.0.1 (SPSS, Inc., Chicago, IL).

Results
Basic characteristics. The mean age, BMI and smoking status were similar in all 4 groups (Table 1).
Mean HbA1c was decreased in pregnant women with T1D as compared to non-pregnant women with T1D. In pregnant women with diabetes, we observed a significantly higher incidence of prematurity, macrosomia and preeclampsia (not significant) as compared to healthy women.
Peripheral immune response. WBC: T1D pregnant patients showed increased WBC-counts. WBCcount increased significantly in both diabetic and healthy women during pregnancy (Fig. 3A). However, WBC-count was significantly higher in diabetic pregnant women vs. healthy pregnant women (interaction diabetes and pregnancy; p = 0.018). The number of lymphocytes did not change during pregnancy in women with T1D, whereas a decreased lymphocyte number was found in healthy pregnant woman vs. their non-pregnant controls (Fig. 3B). The number of monocytes and granulocytes was increased in both groups during pregnancy (Fig. 3C,D).
Lymphocytes: Increased Th1/Th2-ratio in pregnant women with T1D. The percentage of T-lymphocytes was similar in non-pregnant diabetic patients and healthy women (Fig. 4A). During pregnancy, the percentage of T-lymphocytes increased in healthy women only. The ratio of helper T-(Th) and cytotoxic T-cells (Tc) and the number of effector T-cells (Teff) cells were not affected by either diabetes or pregnancy (Fig. 4B,D). The percentage of CD4+ CD25-was lower, while the percentage of CD4+ CD25+ FoxP3+ cells was higher in non-pregnant diabetic vs. non-pregnant healthy women; these differences were not observed during pregnancy (Fig. 4C,E). The percentage of Tbet positive Th cells was significantly higher in both non-pregnant and pregnant diabetic women as compared to their respective healthy controls (Fig. 4F). The percentage of GATA-3 positive Th-cells was similar in the non-pregnant groups, while it was increased in healthy pregnant women only (Fig. 4G). This resulted in an increased Th1/Th2 ratio (p = 0.013) in pregnant women with diabetes vs. healthy pregnant controls (Fig. 4H).
Scientific RepoRts | 5:13618 | DOi: 10.1038/srep13618 healthy women and tended to decrease in diabetic women. The percentage of NKT-cells was influenced neither by pregnancy nor diabetes (Fig. 5C).
To show functional changes of NK cells, we studied the expression of the activating receptor NKp46 (CD335). We choose to study this receptor, since studies have shown that the expression of NKp46 may not only be associated with diabetes development 20 , but also with infertility 21 and preeclampsia 22 .
Most of the ctNK-cells were positive for CD335 (> 80%). The percentage of positive cells and the MFI of CD335 for ctNK-cells was significantly higher in non-pregnant diabetic women than in non-pregnant healthy women (Fig. 5D,E). During normal pregnancy, both the numbers and the MFI of CD335+ ctNK-cells increased as compared to non-pregnant controls. In T1D, we only observed a trend towards an increased MFI of CD335.
Almost all csNK-cells expressed CD335; the percentage of positive cells was not affected by diabetes or pregnancy (Fig. 5F). The MFI of CD335 on csNK-cells was similar in both non-pregnant groups (Fig. 5G). However, we observed an increased (trend) MFI in pregnant diabetic women as compared to non-pregnant diabetic women. This resulted in a significantly increased MFI of CD335 in pregnant women with T1D vs. healthy pregnant women.

Monocytes: Increased activation of intermediate and non-classical monocytes in pregnant women with
T1D. The percentage of intermediate monocytes was decreased in non-pregnant women with diabetes as compared to healthy non-pregnant women, while percentages of classical and non-classical monocytes did not differ between these 2 groups (Fig. 6A-C). In healthy pregnant women we found a decreased percentage of classical monocytes and an increased percentage of intermediate monocytes vs. healthy Gates were set in the isotype control in such a way that < 1% of the cells stained for the isotype control were positive (percentage positive cells are shown in the graphs). This gate was then copied to the samples containing the antibodies.
non-pregnant women. In women with diabetes, only an increased percentage of intermediate monocytes was found during pregnancy as compared with non-pregnant women.
In all groups of women, the three monocyte subsets were 100% positive for MHC-II (data not shown); however, the MFI of MHC-II differed between the subsets (being lowest in classical monocytes and highest in intermediate monocytes) and was affected by diabetes and pregnancy (Fig. 6D In all four groups, almost all classical monocytes were positive for CD62L. This was not the case for the intermediate and non-classical subsets. In these subsets the percentage of CD62L positive monocytes was higher in non-pregnant women with diabetes vs. healthy non-pregnant controls, while the opposite was observed during pregnancy (interaction pregnancy and diabetes; p = 0.00006 and p = 0.0003). In non-pregnant women, the MFI of CD62L on classical monocytes was not affected by diabetes, while this MFI was increased in intermediate monocytes and decreased in non-classical monocytes in women with diabetes as compared with healthy women. During healthy pregnancy, MFI of CD62L was not affected in classical and intermediate monocytes, but was decreased in non-classical monocytes. By contrast, MFI of CD62L was decreased in all monocyte subsets during a diabetic pregnancy.

Discussion
This is the first study showing differences in the systemic immune response in pregnant women with T1D as compared with healthy pregnant women. The strength of our study is that we evaluated various subpopulations of immune cells, of the specific and the innate arm, and their activation status, giving us a broad overview of changes in different cells populations. Our data showed that pregnant women with T1D have an increased WBC-count, Th1/Th2 ratio, and expression of NKp46 on NK-cells and enhanced activation of intermediate and non-classical monocytes as compared to healthy controls. The exact mechanisms are unclear and probably multifactorial. The immunological adaptations in normal pregnancy are suggested to be driven by factors such as increased sex hormone concentrations 23 , presence of the placenta 24 and the production of factors by the placenta, like placental microparticles or  cytokines 25 . Aberrant adaptations in diabetic pregnancy may be due to changes in either of these factors. Alternatively, it may also be due to the presence of autoimmune disease, inducing differences in immune cells, resulting in a different response to pregnancy factors.
It is well known that the immune response of non-pregnant diabetic patients differs from that of healthy subjects [17][18][19]26,27 . Our data confirmed an increased Th1 response and an increased activation status of monocytes in diabetic women vs. healthy women [26][27][28] . Moreover, we augment previous knowledge by showing that only intermediate and non-classical monocytes are activated. We also showed an increased percentage of Treg in non-pregnant diabetic women, which may act as a compensatory mechanism for impaired Treg function 18 . Moreover, we found decreased NK-cell numbers in non-pregnant diabetic women, which may be related to this defective Treg function 29 . CD335 (or NKp46) , a member of the cytotoxicity receptor-family 30 , is essential in the development of T1D 20 . This may explain why this receptor is upregulated on ctNK-cells in non-pregnant diabetic women.
Various studies observed a decreased Th1/Th2 ratio during healthy pregnancy 4,10 . Although we did not find this, we observed an increase in GATA-3 positive Th lymphocytes in healthy pregnant women. The reason for this discrepancy is not clear, but may be due to methodological differences 31 . The decreased percentage of Treg during pregnancy is in line with earlier reports 32,33 . This decrease in Treg appeared to be more pronounced in diabetic women, suggesting that Treg do adapt to a diabetic pregnancy. The increase in Th1 response in diabetic pregnancy was also found in non-pregnant diabetic women, suggesting that this is due to the diabetic state. We observed an increased Th1/Th2 ratio in diabetic pregnancy. As such an increased ratio is associated with pregnancy complications, such as preeclampsia and miscarriage 7,10 , it may be suggested that the increased Th1/Th2 ratio in diabetic pregnancy is be involved in the increased numbers of pregnancy complications in this condition. In contrast to other studies, pregnancy did not affect the total percentage of NK-cells in healthy controls 11,34 . However, in line with former studies, the absolute numbers of NK-cells tended to decrease in healthy pregnancy (1.73·10 8 vs. 1.28·10 8 , p = 0.053), but not in T1D pregnancy (1.21·10 8 vs. 1.12·10 8 , p = 0.652). This may be due to the already low frequency of these cells in non-pregnant diabetic women; a further decrease may not be required during pregnancy. We observed increased CD335 expression on  (D-F), and percentage and MFI of CD62L positive monocyte subsets (G-L) in nonpregnant (NP) and pregnant (P) women with type 1 diabetes and healthy women. *Significant difference between the marked groups, independent T-test, p < 0.05. I: Significant interaction between diabetes and pregnancy, p < 0.05, two-way ANOVA.
both NK-cell subpopulations in pregnant women with diabetes. The expression of CD335 (or NKp46) is negatively correlated between csNK cell CD335 expression and type 1 cytokines, such as TNF-α 21,35 in a fertile population, but not in women with recurrent pregnancy loss or implantation failure 21 . This may suggest that signal transduction via CD335 on NK cells is changed in these pregnancy complications. If this is also the case in diabetic pregnancy, upregulation of CD335 may be a compensatory mechanisms.
We found increased expression of MHC-II and decreased expression of CD62L on intermediate monocytes in diabetic pregnant women, suggesting activation of these cells. Decreased MHC-II expression on monocytes is associated with inadequate presentation of antigens to T-cells and disturbed T-cell stimulation 36 . Thus, the ability of these monocytes to present antigens to T-cells is increased in pregnant women with diabetes, which could lead to enhanced activation of the systemic immune response. Shedding of CD62L is usually associated with the activation of monocytes 37 , indicating that the monocytes of diabetic women may be further activated during pregnancy. Intermediate monocytes have been suggested to play a role in the pathophysiology of preeclampsia 16,38 . It may therefore be suggested that increased activation of intermediate monocytes during diabetic pregnancy, may be involved in the increased incidence of preeclampsia in these women.
Our observations of aberrant immunological changes in diabetic pregnancy raise further questions. Various adaptations of the immune response are required for adequate implantation, placentation and tolerance of the semi-allogeneic fetus [4][5][6] . We hypothesize that the increased pregnancy complications in diabetic pregnancy are partly the result of the disturbed immunological adaptations, immunological maladaptations may lead to pregnancy complications [7][8][9] . We cannot completely rule out a role for hyperglycemia, since, although the mean HbA1c was significant decreased in pregnant T1D women as compared to non-pregnant T1D women (which is most likely the result of the preconceptional care of these women 39 ), it is still increased as compared to healthy pregnant women (mean HbA1c 4.7% for healthy pregnant women with a gestational age of 30 weeks 40 ). However, our suggestion of immunological involvement is corroborated by our findings in an animal model for T1D, in. In this model, in the face of normoglycemia, similar immunological maladaptations as in the present study, were associated with increased pregnancy complications 3 . Our hypothesis is further endorsed by the fact that in the present study women with T1D and complicated pregnancy outcome (i.e. preeclampsia or preterm delivery) showed even higher Th1/Th2-ratios and more pronounced activation of monocytes as compared to diabetic women with normal pregnancy outcome (data not shown). Moreover, our suggestion that disturbed immunological adaptations in an autoimmune environment contributes to pregnancy complications is in line with findings in other autoimmune diseases, such as rheumatoid arthritis or systemic lupus erythematosus 41,42 . Unfortunately, due to the relatively small sample size in this study, we were not able to correlate immunological parameters to pregnancy outcome parameters.
In conclusion, this study showed that in T1D the pregnancy induced immunological adaptations are disturbed. Importantly, these different immunological adaptations may be responsible, as also shown for other autoimmune diseases, for the increased frequency of complications in pregnant women with T1D. Further mechanistical insights into the role of various immune cells in implantation as well as placental and fetal development in healthy and diabetic pregnancy are warranted and may lead to new therapeutic approaches to improve pregnancy outcome of women with T1D.