In ovo hyperglycemia causes congenital limb defects in chicken embryos via disruption of cell proliferation and apoptosis

Abstract

While the correlation between diabetes during pregnancy and birth defects is well-established, how hyperglycemia causes developmental abnormalities remains unclear. In this study, we developed a novel “hyperglycemic” chicken embryonic model by administrating various doses of glucose to fertilized eggs at embryonic stages HH16 or HH24. When the embryos were collected at HH35, the LD50 was 1.57 g/Kg under HH16 treatment and 0.93 g/Kg under HH24 treatment, indicating that “hyperglycemic” environments can be lethal for the embryos. When exposed to a dose equal to or higher than 1 g/Kg glucose at HH16 or HH24, more than 40% of the surviving chicken embryos displayed heart defects and/or limb defects. The limb defects were associated with proliferation defects of both the wing and leg buds indicated by reduced numbers of p-H3S10 labeled cells. These limb defects were also associated with ectopic apoptosis in the leg bud and expression changes of key apoptotic genes. Furthermore, glucose treatment induced decreased expression of genes involved in Shh-signaling, chondrogenesis, and digit patterning in the limb bud. In summary, our data demonstrated that a high-glucose environment induces congenital heart and limb defects associated with disrupted cell proliferation and apoptosis, possibly through depressed Shh-signaling.

INTRODUCTION

Diabetes mellitus is a chronic metabolic disease characterized by hyperglycemia associated with insulin deficiency or insulin resistance. As the incidence of diabetes mellitus continues to rise and the age of affected population keeps dropping, women of childbearing age have a higher rate of diabetes mellitus during pregnancy. Diabetes during pregnancy affects the health of both the mother and child. Of the two types of diabetes mellitus during pregnancy, gestational diabetes accounts for 90% while pre-gestational diabetes accounts for 10% (1, 2). In the United States, the prevalence of gestational diabetes among pregnant women is as high as 9.2% (3). With the introduction of insulin, diabetes-associated fetal mortality rate dropped from 70% to nearly 12% (4). Unfortunately, the birth defect rate from diabetic pregnancies (~10%) remains higher than the general population (3%) and appears to be on the rise (3–5). While the positive correlation between diabetes mellitus during pregnancy in women and birth defects is well established, the specific mechanism in which in utero hyperglycemia leads to developmental abnormalities remains unclear. In order to formulate a preventive strategy for diabetes mellitus related birth defects, a better understanding of fetal development under hyperglycemia is urgently needed.

Diabetes during pregnancy can lead to a variety of congenital anomalies, including abnormalities of the craniofacial, cardiovascular, gastrointestinal, urogenital, musculoskeletal, and central nervous systems (4). For over twenty years, limb defects, especially the preaxial polydactyly, have been used as markers for diabetic embryopathy due to their high association with diabetic mothers (6–8). The longitudinal limb deficiencies in infants have also been reported to be associated with diabetes mellitus during pregnancy (OR=7.01 and 95% Crl=1.91–25.68) (9). The Center for Disease Control and Prevention (CDC) estimates that 6 out of every 10,000 babies in the United States are born with upper limb and/or lower limb abnormalities every year (10). Cardiovascular defects are also common congenital abnormalities associated with maternal diabetes. The risk of having a baby with congenital heart defects (CHD) is 2–5 times higher in diabetic mothers compared to non-diabetic mothers (11, 12). A recent systematic meta-analysis study of literature from 1975 to 2012 on the association of diabetes mellitus during pregnancy and CHDs reported that 8.3% of pre-gestational diabetes mellitus/gestational diabetes mellitus women gave birth to infants with CHDs (13). Importantly, diabetic embryopathy has been reported to be linked with VACTERL association, in which an individual exhibits a combination of vertebral, anal, cardiac, trachea-esophageal, renal, and limb defects (14–18).

Previous studies have used both the STZ induced (19–22) and high-fat diet induced (6) pregestational diabetes mouse models to investigate how maternal diabetes causes birth defects. These models contribute a considerable amount of information about the manifestations of fetal cardiovascular and central nervous system defects. In these studies, ontogeny of congenital heart defects in offspring from pre-gestational diabetic pregnancies is explained through mechanisms that involve cell proliferation defects (22, 23), excessive programmed cell death (24–26) and abnormal migration of the cardiac progenitors during early stages of heart development (27). However, no study has examined the molecular mechanisms behind the link between diabetic pregnancies and abnormal limb development or concurrent abnormalities of the limbs and heart.

In order to investigate diabetic embryopathy, especially in the limb, we developed a novel chicken embryonic model with a high glucose environment in ovo. These embryos developed limb and/or heart defects that resemble those of infants born to diabetic mothers. Using this model, we studied the molecular and cellular mechanisms of how embryonic development under hyperglycemia leads to limb defects.

METHODS

Maintenance of chicken embryo

Fertilized eggs (Gallus gallus, Single Comb White Leghorn) were obtained from the Poultry Science Department, Texas A&M University (College Station, TX, USA). All chicken embryos were maintained in an egg incubator at 38°C ± 0.5°C.

Preparation of chicken saline and glucose solution

Chicken saline (110 mM NaCl, 10 mM BaCl₂, 0.4 mM MgCl₂, 5.3 mM KCl) was used as a vehicle. Glucose was dissolved in chicken saline at 330 mg/mL then filtered through a 0.22 um syringe filter as a stock solution and stored at −20 °C for up to one month. Before each treatment, the stock solution was diluted with chicken saline to the respective doses.

Chicken saline and glucose solution injection

At embryonic day 2 (HH16 stage, 51–56 hours old) or day 4 (HH24 stage, 4.5 days old), the weight of each egg was measured with a scale. Treatments were conducted under a sterilized culture hood. The eggs were oriented with sharp ends pointing down and cleaned with 70% ethanol. A small window (less than 0.5cm²) on the shell above the air-sac was pierced open, and the shell membrane was carefully peeled away. The pre-made chicken saline or glucose solution of certain doses were pipetted onto the air sac membrane, with a volume adjusted to the egg’s weight. The dosage of the glucose (g/Kg) was calculated as the amount of glucose in grams per kilogram of egg being treated. The window was sealed with medical tape to prevent contamination post-treatment and returned to the incubator.

Yolk glucose measurement

At each harvest, 100 μL of yolk from each sample was added to an EP tube with 500 μL of PBS by pipetting up and down several times to wash out the yolk on the tip, then vortexed and spun down. The glucose concentration of the diluted yolk was measured with Contour NEXT EZ glucose monitoring kit (Ascensia Diabetes Care US Inc, Parsippany, USA).

Real-Time PCR

The leg and wing bud at either HH27 or HH29 were collected and put in a 2 mL EP tube with 500 uL Trizol reagent (Thermo Fisher Scientific, Waltham, USA). The RNA extraction and real-time PCR were performed as previously established (28). The primers used are listed in Supplementary Table 4.

Immunohistochemistry (IHC) for proliferation and apoptosis

Standard procedures were used for IHC. IHC was performed using the rabbit anti-mouse p-Histone-H3 (p-H3S10) (Abcam) as the primary antibody. For colorimetric staining, slides were incubated with rabbit ImmPress reagent (Vector Labs) and stained with a 3,3′-diaminobenzidine-tetrahydrochloride (DAB) substrate kit (Vector Labs), and counterstained with hematoxylin (HE). For TUNEL, an ApopTag Plus Peroxidase in-Situ Apoptosis Detection Kit was used (EMD Millipore) with DAB and HE.

Statistical Analysis

Data of yolk glucose concentration, histology, and real-time PCR were analyzed using Student’s t-test. Data were considered significantly different when p<0.05. The limb and heart defect rates were analyzed using Chi-square test, the difference was considered significant when p<0.05. The two-parameter logistic function was used to analyze the survival rate data. The survival curves were generated using the following equation:

$$y=c+\frac{d-c}{1+exp\{b(x-e)\}}$$

In this equation, c stands for baseline response and d is the largest effect achievable. If b < 0 the curve is increasing, otherwise, curve is decreasing. Lethal-dose 50 (LD50) is equal to e when x = e, $y=c+\frac{d-c}{2}$.

LD50 is the estimated dosage lethal to 50% of the population.

Data and Resource Availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

RESULTS

Glucose treatment successfully induces “hyperglycemic environment” in fertilized chicken eggs.

Chicken saline or different doses of glucose (0.3, 0.5, 0.75, 1, 1.25, 1.5 and 2 g/Kg of egg weight) were administered to the air-sac of fertilized eggs at either Hamburger Hamilton stage 16 (HH16) or HH24 (Supplementary Table 1). The glucose levels in the yolk of eggs injected with chicken saline, or various doses of glucose solution were then measured at HH29 and HH35. There was only a marginal difference in yolk glucose at HH29 (1 g/Kg: 159.1±21.3 mg/dL vs. saline: 118.7±6.2 mg/dL, p=0.084; Fig. 1A). However, injections of 1g/Kg glucose into fertilized eggs at HH16 resulted in a significantly higher yolk glucose compared to the control at HH35 (1 g/Kg: 196.8±5.9 mg/dL vs. saline: 139.1±10.1 mg/dL, p=0.0002; Fig. 1B). Injections of 1 g/Kg glucose into fertilized eggs at HH24 caused significantly higher levels of yolk glucose checked at both HH29 (1 g/kg: 168±6.1mg/dL vs. saline: 112±6.9 mg/dL, p=0.00393) and HH35 (1 g/Kg: 199.7±4.7 mg/dL vs. saline: 132.0±7.4 mg/dL, p=0.0000002). At HH35, fertilized eggs treated with 0.75 g/Kg glucose solution at HH24 also had significantly higher yolk glucose levels compared to the control (0.75 g/Kg: 159.1±6.2 mg/dL vs. saline: 132.0±7.4 mg/dL, p=0.025; Fig. 1B). Therefore, we confirmed the success of creating a hyperglycemic environment in ovo through glucose treatment that simulates “in utero hyperglycemia”.

Fig 1. Embryo survival rate decreases with increasing glucose dose, especially in later exposures.

A) At HH16 or HH24, eggs were treated with either glucose (1g/Kg) or saline (control). The yolk glucose levels was measured at HH29.

B) At HH16 or H24, eggs were treated either glucose (0.5g/Kg, 0.75g/Kg, or 1g/Kg) or saline (control). The yolk glucose levels was measured at HH35.

C) For embryos at HH35 with treatment at HH16, most of the embryos treated with saline or 1g/kg glucose group survived, 57% embryos treated with 1.5g/kg glucose died.

D) For embryos at HH35 with treatment at HH24, most of the embryos treated with saline or 1g/kg glucose group survived, all embryos treated with 1.5g/kg glucose died.

E) Dose-survival rate curve of HH16 injection. LD50 is 1.57g/kg at this treatment stage.

F) Dose-survival rate curve of HH24 injection. LD50 is 0.93g/kg at this treatment stage.

G-R) H&E staining of HH35 embryo sections shows VSD (G, K, O), DORV (H, L, P), PTA (I, M, Q), tricuspid valve regurgitation (J, N, R). AO, aorta; DORV, double outlet right ventricle; LV, left ventricle; RV, right ventricle; PTA, persistent truncus arteriosus; VSD, ventricular septal defect. Read arrows indicates defects.

Data is presented as Mean±SE, N=6–30; #p<0.1, *p<0.05, **p<0.01, ***p<0.001.

Embryo survival rate decreases with increasing glucose treatment, especially in later exposures.

The survival rates of glucose treated chicken embryos were determined at both HH29 and HH35 (Supplementary Table 1). At the time of embryo collection, all the embryos treated with saline were normally developed, but some of the embryos treated with glucose solution were found dead at an earlier stage, especially at higher glucose doses (Fig. 1C, D). A dose-survival rate curve was fitted with a sigmoid function for logistics model analysis for HH35 embryos. Treatment at both stages resulted in reverse “S” shaped curves which indicated that embryos survived at lower doses of glucose while their survival rate dropped within a narrow range of dosage until it plateaued (Figure 1E) or approached to zero as glucose levels increased (Fig 1F). The effective dose that caused the death of 50% of treated embryos (LD50) was higher in embryos treated at HH16 (LD50=1.57 g/Kg) compared to HH24 (LD50=0.93 g/Kg). This trend implies that the sensitivity to glucose varies at different developmental stages. The sharp decrease in survival rate of HH24 treated embryos indicates that embryos in later stages of development are more susceptible to the effects of hyperglycemia.

Hyperglycemia increases rate of OFT heart defects.

The presence of heart defects was examined in surviving HH35 embryos treated with 0.75 g/Kg or higher glucose dosage at HH16 or HH24. The heart structure was examined via histological evaluation on serial sections.

Embryos treated at HH16 with 1 g/Kg glucose displayed ventricular septum defects (VSD) where the ventricular septum is partially missing (Fig. 1K and supplementary Table 2, 5/12, p=0.037), double outlet right ventricle (DORV) where both the aorta and pulmonary artery connect to the right ventricle (Fig. 1L and supplementary Table 2, 4/12, p=0.093), and persistent truncus arteriosus (PTA) where the truncus arteriosus failed to separate into the aorta and pulmonary artery (Fig. 1M and supplementary Table 2, 1/12, p=1). The total incidence of outflow tract (OFT) defects, including VSD, DORV and PTA, was five out of 12 embryos (Fig. 1 and supplementary Table 2, p=0.037). The percentile of myocardium area relative to the area of ventricular chamber and the thickness of compact myocardium on the ventricular wall of the glucose group were similar to the control group, suggesting neither hypoplastic nor hyperplastic myocardium in these embryos (Supplementary Figure 1).

When 1 g/Kg glucose was given at HH24, we observed VSD (Fig. 1O and supplementary Table 2, 6/9, p=0.0015) and DORV (Fig. 1P and supplementary Table 2, 5/9, p=0.0062) at HH35. PTA was not observed. Valve malformation was found in one of the embryos (Fig. 1R and supplementary Table 2, 1/9, p=0.429). This embryo displayed four leaflets instead of a tricuspid valve. The total incidence of outflow tract defects (OFT), including VSD and DORV, and valve abnormality was six cases out nine embryos (Fig. 1 and supplementary Table 2, p=0.0015). There was no significant difference in percentile of myocardium area relative to the area of ventricular chamber, nor thickness of compact myocardium between the glucose and the control group (Supplementary Figure 1). This indicates that neither hypoplastic nor hyperplastic myocardium was induced by glucose treatment at HH24, similar to observations of HH16 treated embryos. These results suggest that hyperglycemic conditions during development leads to increased rates of congenital heart defects, especially at the OFT.

Hyperglycemia increases the rate of limb defects.

Among the surviving HH35 embryos exposed to high glucose at HH24, the presence of leg/wing reduction was observed with an incident rate of 13% when exposed to 0.75 g/Kg glucose (Fig. 2 and Supplementary Table 3, 4/30, p=0.2091) or 44% when exposed to 1 g/Kg glucose (Fig. 2 and Supplementary Table 3, 4/9, p=0.0058). Specifically, in the 0.75 g/Kg group, limb reduction applied to the leg only in three of the four observed cases, and one case had reductions in both wing and leg. At the dose of 1 g/Kg glucose, two of the four cases were in the leg only, and the two other cases had both wing and leg reductions. To be noted, we did not observe any embryos with only wing defects but with normal legs. We then speculated if hyperosmolarity, instead of hyperglycemia, caused these defects. Thus, we treated eggs with L-glucose at two different does, 0.75 g/kg and 1 g/kg, at HH24 (Supplementary Table 1) to induce hyperosmolarity. Neither limb defects nor heart defects were observed in these chicken embryos. Thus, we ruled out the possibility of hyperosmolarity to be the cause of the observed limb and heart defects with glucose treatment.

Fig 2. High glucose treatment at HH24 causes limb defects.

A) The embryos displayed normal limbs with saline injection.

B-E) For HH24 injection of 0.75 or 1 g/Kg glucose, embryos showed defects at only leg (B, D) or at both leg and wing (C, E). Red arrows show the malformed limbs.

High level of glucose causes proliferation defects in the limb bud at HH27–29.

We then investigated if the limb reduction was due to decreased numbers of proliferating cells. Immunohistochemical (IHC) staining for p-H3S10 were performed on limb buds at HH27 or HH29 (24 or 48 hours post-treatment) of HH24 1 g/Kg glucose treated embryos (Fig. 3A–D). At HH27, there were significantly lower numbers of p-H3S10+ cells in the wing bud of 1 g/Kg glucose treated embryos compared to the saline (Fig. 3E, 1 g/Kg: 666±56/mm² vs. saline: 916±66/mm², p=0.045). In the leg, the numbers of p-H3S10+ cells were not different between glucose treated and saline treated embryos at HH27 (Fig. 3F). At HH29, treatment with 1 g/Kg glucose resulted in a two-fold decrease of p-H3S10+ cells in the leg bud compared to the saline (Fig. 3H, 1 g/Kg: 181±5.7/mm² vs. saline: 583±25/mm², p =0.00009). However, there was no significant difference in the number of p-H3S10+ cells observed in the wing buds of glucose treated and saline controls at HH29 (Fig. 3G).

Fig 3. High glucose treatment at HH24 cause proliferation defects in limbs at HH27–29.

A-B) IHC staining for p-H3S10 in wing and leg of 1g/Kg glucose injected embryo at HH27.

C-D) IHC staining for p-H3S10 in wing and leg of 1g/Kg glucose injected embryo at HH29.

E-H) The number of p-H3S10+ cells in 1mm² of wing or leg tissue.

I-J) Expression levels of cell cycle related gene in wing and leg tissue at HH27.

K-L) Expression levels of cell cycle related gene in wing and leg tissue at HH29.

Data is presented as Mean±SE, N=3–5, #p<0.1, *p<0.05, **p<0.01, ***p<0.001.

The expression levels of several key genes involved in cell proliferation, including p21, p16, CycD and CycE, were measured in both wings and legs at HH27 and HH29 after 1 g/Kg glucose injection at HH24. In the HH27 wings, the expression of CycD in the glucose treated embryos was significantly lower than that of the control. However, this level of difference faded over time and the glucose induced reduction of Cdk1 expression in the wing was only marginal at HH29 (p<0.1). These expression changes are consistent with the decrease in proliferation rates marked by lower pH3S10+ count observed in HH27 wing, but not the HH29 wing (Fig. 3I and K). Unlike the wing, expression levels of these genes in the HH27 leg were not statistically different between glucose and saline treated embryos (Fig. 3J). However, CycD expression was significantly lower in the HH29 leg exposed to glucose compared to saline (Fig. 3L). CycE expression in the glucose treated HH29 leg was also lower with a marginal significance (p<0.1).

Taken together, these results suggest that “hyperglycemia” hinders proliferation in the wing at an earlier stage (HH27) and the leg at a later stage (HH29).

Hyperglycemia causes abnormal patterns of cell survival in the limb bud at HH27.

One of the crucial steps of limb development, specifically the digit formation, is the degeneration of interdigital tissue via controlled cell death patterns (29). We hypothesized that glucose-induced limb truncation could be accounted by abnormal cell survival. TUNEL staining was performed to evaluate if glucose treatment (1 g/Kg) result in changes in programmed cell death patterns in the developing HH27 limb buds.

In the control wing bud, three regions were shown to be undergoing programmed cell death: the apical ectodermal ridge (AER) (Fig. 4A and E, region “a”), the posterior necrotic zone (PNZ) (Fig. 4A, E; region “b”), and the opaque patch (OP) (Fig. 4A, E; region “c”). Apoptotic cells were also observed in the same three regions of glucose treated wing buds. The number of apoptotic cells was quantified at each region in all the serial sections through the wing. There were no significant differences in the numbers of apoptotic cells in each individual region of the wing or in total between the saline controls and glucose treated embryos (Fig. 4B, E; 1 g/Kg: 56±7 vs. saline: 49±9 , p=0.597).

Fig 4. High glucose treatment at HH24 caused ectopic apoptosis in the leg at HH27.

A-D) TUNEL staining in leg and limb tissue at HH27 embryo. a, b, c or d indicate corresponding region of apoptotic cell of different tissue.

E-F) The number of TUNEL+ cells in each region. Data is presented as Mean±SE, N=3–5, *p<0.05.

G-H) Expression levels of apoptosis related gene in wing and leg tissue at HH27. Data is presented as Mean±SE, N=3–4; , #p<0.1,*p<0.05, **p<0.01, ***p<0.001.

In the control leg bud, two major regions were shown to be undergoing programmed cell death: the PNZ (Fig. 4C, F; region “a”) and the OP (Fig. 4C, F; region “b”). Interestingly, in the glucose treated embryos, the PNZ region showed no apoptotic cells while the OP region had an increased number of apoptotic cells (Fig. 4D, F; region “a” and “b”). In control leg buds, no apoptotic cells were observed in the AER (Fig. 4C, region “c”) or in the region opposite to AER (Fig. 4C, region “d”). In contrast, the glucose treated leg bud had a significant count of apoptotic cells in both of these regions (Fig. 4D, “c” and “d”; Fig. 4F). Despite the presence of ectopic apoptosis, the total number of programmed cell deaths was not different between the glucose treated leg bud and control leg bud (Fig. 4D and F, 1 g/Kg: 103±6 vs. saline: 70±29, p=0.33).

Next, we examined the expression of several key genes involved in cell survival, including Casp2, Casp3, Casp8, Apo2L, Api5, Bcl, Bid, Bcl2L Bak1, Bax and Grp78. In the wing buds, the expression levels of Grp78 and Apol2L significantly decreased in the glucose treated embryos compared to their saline controls (Fig. 4F). In the leg buds of glucose treated embryos, all genes, except for Casp2 and Casp3, had significantly decreased expression levels compared to the control embryos (Fig. 4G). Interestingly, these include Bak1, Bax, Casp8, Apo2L and Bid that promote apoptosis as well as Grp78, Bcl, Bcl2L and Api5 that are anti-apoptosis.

Hyperglycemia affects the integrity of Shh signaling in the limb bud at HH27.

Sonic hedgehog (Shh) is required for both posterior-distal limb skeleton and the posterior digit development in vertebrate animals (30–32). Therefore we investigated if high glucose (1 g/Kg) would affect the integrity of Shh-signaling in the developing limb buds by measuring changes in expression of its key modulator genes. In the wing buds, glucose treatment significantly lowered the expression of Shh (Fig. 5A, 1 g/Kg: 0.58±0.0349 vs. saline: 1.00±0.0017, p=0.000159), but not its downstream key modulator genes (Fig. 5A). In the leg buds, the Shh modulator genes, including Smo, Gli1, Gli3 and Ptch1 were expressed at significantly lower levels in the glucose treated leg buds compared to the control leg buds (Fig. 5B). These results suggest that hyperglycemia disrupts the integrity of Shh-signaling in limb buds.

Fig 5. High glucose treatment at HH24 disrupted the integrity of Shh signaling pathway and the expression of genes for chondrogenesis and digit patterning in HH27 limbs.

A-B) Expression levels of Shh, Smo, Gli1, Gli3, Ptch1, Foxf1 in limb tissue of saline or glucose treated embryos.

C, D) Expression levels of Sox9, Col2a1, Runx2, Hoxd10, Hoxd11, Hoxd12, Hoxd13 in limb tissue of saline or glucose treated embryos.

Data is presented as Mean±SE, N=3–4, *p<0.05, **p<0.01, ***p<0.001.

Hyperglycemia induces expression changes in genes involved in chondrogenesis and digit patterning at HH27.

In order to understand how high glucose causes defects in limb outgrowth, we further assessed expression changes in chondrogenic gene markers (Sox9, Col2a1), early osteogenic gene markers (Runx2), and patterning gene markers (Hoxd10, Hoxd11, Hoxd12 and Hoxd13) by RT-PCR. At HH27, the expression of Col2a1 decreased in both the wing and leg bud of glucose treated embryos (Fig. 5C, D). Unique to the wing bud, Hoxd13 expression was significantly lower in glucose treated embryos compared to the control (Fig. 5C). There were significant decreases in expression of all patterning genes including Hoxd10, Hoxd11, Hoxd12 and Hoxd13 in the leg bud of glucose treated embryos compared to the control (Fig. 5D). There was no difference in Runx2 expression in either wing or leg buds between glucose and saline treated embryos (Fig. 5C, D).

DISCUSSION

In the current study, we established a novel “in ovo hyperglycemia” model using chicken embryos to mimic fetuses exposed to hyperglycemia in mothers with diabetes. By administering the glucose solution into the air-sac of fertilized eggs in ovo, the chicken embryos were exposed to higher glucose levels in the yolk that simulates an in utero hyperglycemic environment.

Recent studies report that diabetes during the first trimester, which expose the early embryo to hyperglycemia, closely relates to birth defects. Higher fasting glucose level in the first trimester (5, 33–37), even within the normoglycemic range (34, 37), is an early predictor of gestational diabetes mellitus. Indeed, fetuses can be exposed to amniotic fluid with elevated glucose levels as early as gestational week 15 which is before screening takes place for gestational diabetes mellitus (33). Thus, fetuses can already be exposed to a high-glucose environment in the early developmental stage far before diagnosis of gestational diabetes mellitus in the third trimester. Unlike STZ-induced Type-I diabetes (19–21, 38) or high-fat diet induced Type-II diabetes (6) models of pre-gestational diabetes mellitus, our chicken embryonic model allows for the specific investigation of dose and temporal effects of elevated glucose levels. In our study, post-fertilization hyperglycemia was induced at HH16 or HH24, which is equivalent to the human embryonic day 24 or 42, respectively (at gestational week 5 or 8). Moreover, treatment of 1 g/Kg glucose at HH16 or 0.75 g/Kg glucose at HH24 increased glucose levels in the HH35 egg yolks by 21% and 51%, respectively. Considering HH35 as equivalent to human embryonic day 56 or gestational week 10, our data mimics the 44% glucose elevation detected in amniotic fluids of gestational diabetic pregnancies compared to normal glycemic pregnancies (33). Thus, the chicken embryonic model used in this study resembles both temporal and clinical glucose levels that embryos are exposed to during the first trimester of gestational diabetic pregnancies.

Although a handful of studies report an increased risk of congenital defects in infants born to women with gestational diabetes (39–42), conflicting reports exist (43–46). A recent systematic review of 768 potential articles to assess observed associations between gestational diabetes mellitus and birth defects found no evidence for consistent association of gestational diabetes mellitus with birth defects (47). There is only a weak association between gestational diabetes mellitus and congenital heart defects (47). Considering that gestational diabetes mellitus is always diagnosed after the critical developmental period of various organs, the data in these studies could be confounded by factors such as obesity and undiagnosed prediabetes before conception. Consistently, the same systematic review reiterated this paradox by reporting an increased risk of selected birth defects among offspring of women with both obesity and gestational diabetes mellitus (47). With this discrepancy, the direct link between in utero hyperglycemia and birth defects remains to be addressed. In our study, we showed that glucose treatment (0.75 g/Kg at HH16), even at a dose unable to maintain a prolonged high glucose level in egg yolks, was substantial enough to be teratogenic and induce embryopathy. At treatment doses that maintain high glucose levels in yolks, embryos displayed increased incidences of birth defects. Our data demonstrates that exposure to abnormal glucose levels during development is an important risk factor for heart and limb abnormality. Importantly, both the STZ induced (19–21, 38) and high-fat diet induced (6) pre-gestational diabetes mellitus mouse models have contributed a considerable amount of information about the manifestations of fetal cardiovascular defects and central nervous system defects. However, neither of the two mouse models nor other vertebrate animal models reported phenotypes of limb defects in offspring of gestational diabetes mellitus mothers. From our knowledge, this study is the first to report phenotypes of limb abnormality caused by exposure to “hyperglycemia” during development.

Common features of birth defects resulting from gestational diabetes mellitus pregnancies are variable expressivity and incomplete penetrance (9). The basis for this variability is presently unclear. In our study, a hyperglycemic environment was induced at two different time points that were selected based on a previous study designed to investigate the effect of maternal hyperglycemia on offspring myocardium development. The stage HH16 is when chicken embryos start myocardium trabeculation, and at the stage HH24, myocardium compaction starts (48). Neither hypoplastic nor hyperplastic myocardium were induced with the glucose treatment. However, structural heart defects including VSD and OFT defects were observed in embryos treated at either timepoint. Although DORV was observed in HH16 glucose treated embryos, the rate was not statistically significant. Consistently, limb defect rates were higher with glucose administration at HH24, but not HH16. These results might suggest that the OFT and limb development were more sensitive to “hyperglycemia” at a later stage than an earlier stage. In addition, we reported association of “hyperglycemia” with a certain spectrum of CHD phenotypes, which is consistent to previous findings in human (49). Similar as their reports, we observed much more outflow tract malformations than the inflow tract malformations in our model, suggesting that the cardiac precursor cells located at the anterior second heart field are more sensitive to “hyperglycemia” than the posterior second heart field. Unfortunately, Oyen et al. did not provide data to evaluate if GDM is associated with hypoplastic or hyperplastic myocardium. Because hyperglycemia may consistently affect the ventricular wall thickness through the entire embryonic stage and after birth, we could not conclude that hyperglycemia is not associated with hypoplastic or hyperplastic myocardium based on our observation at HH35. Furthermore, several key points should be noted. First, the major OFT defects induced by glucose were DORV which are caused by disruptions in alignment. During chicken heart development, the OFT septation starts at HH27 (48). Secondly, the glucose administration was only given once, therefore, the glucose levels at HH27 could be different between the HH16 and HH24 treated groups. Our results and previous knowledge of OFT developmental timeframe suggest the mechanistic etiology of embryopathy under gestational diabetes mellitus to involve both dose and timing of “hyperglycemia” as elements of variability. Thus, our data provides a potential explanation why clinical gestational diabetes mellitus is associated with incomplete penetrance and variability of amongst offspring phenotypes.

Our data suggests that “hyperglycemia” during development causes limb defects by disrupting cell proliferation and apoptosis in the limb bud. First, glucose treatment caused more leg reduction than wing reduction, as wing reduction was always concurrent with the leg reduction. Consistent with the aforementioned phenotype, the leg buds showed both ectopic apoptosis and disrupted cell proliferation, while the wing bud only displayed proliferation defects. These results suggest to us that normal regulation of both cell proliferation and cell survival patterns is required for proper limb development. Second, the glucose treatment disrupted normal patterns of apoptosis rather than reducing the total number of apoptotic cells during limb development. This finding is consistent with previous knowledge that suggest major areas of programmed cell death in the limb bud mesenchymal cells to be a part of programmed limb growth (29). Third, exposure window to “hyperglycemia” is critical to affect cell proliferation in the wing bud vs. the leg bud. The proliferating cells in the leg are more sensitive to elevated glucose between HH27 and HH29, evidenced by lower expression of CycD and CycE and lower count of pH3S10+ cells observed only at HH29, but not HH27. In contrary, stages before HH27 are more important to wing bud cell proliferation as lower pH3S10+ cell count and suppression of CycD was observed only in the HH27 wing, but not the HH29 wing.

Shh signaling is required for the limb bud outgrowth by controlling the width of the limb bud and by regulating the antero-posterior length of the AER (30–32). Mechanisms involve regulation of mesenchyme cell proliferation and apoptosis (31). In our study, the hyperglycemia-induced abnormalities in proliferation and apoptosis of limb buds were associated with depressed Shh-signaling pathway. Glucose treatment disrupted the integrity of Shh-signaling which depressed expressions of cell cycle regulators such as CycD, consistent with previous reports (31, 50). This signaling is likely to be independent of AER, the signaling region required for laying down structures that drive limb bud outgrowth along the proximal-to-distal axis (31, 50). However, decreased cell survival in the AER of the glucose treated leg bud directly affected the proximal-distal growth evidenced by limb reduction in our samples. In addition, the interaction between AER and the polarizing region, PNZ, is required for the formation of both anterior-posterior and proximal-distal patterning (31). The number of cells in the polarizing region is controlled by Shh signaling which regulates BMP2 signaling, that in turn determines the size of the PNZ (51). In our study, decreased Shh signaling could have promoted apoptosis in PNZ, possibly via inhibition of BMP2 signaling, leading to fewer cells in that region. Our data points to potential candidates for preventing diabetic embryopathy. Future studies can test if applying a chemical agonist for Shh-signaling could prevent hyperglycemia induced limb defects by rescuing the Shh/BMP and Shh/CycD levels.

In summary, our study established a novel chicken embryonic model with a high glucose environment in ovo. Using this model, the important mechanisms of how gestational diabetes mellitus increases the risk for birth defects were investigated. Findings of this study will guide efforts to explore interventional approaches to reduce diabetes-induced embryonic malformations.

Highlights.

• Administration of glucose induces hyperglycemia in ovo, which shows dose-dependent lethality to the embryos and causes limb or heart defects.

• Hyperglycemia causes proliferation defects and disrupted apoptosis pattern in the limb.

• Hyperglycemia affects the integrity of Sonic hedgehog signaling in the limb.