Endocrine Reviews, doi:10.1210/er.2007-0029
Endocrine Reviews 29 (3): 265-291
Copyright © 2008 by The Endocrine Society
Neonatal Diabetes Mellitus
Lydia Aguilar-Bryan and
Joseph Bryan
Pacific Northwest Diabetes Research Institute, Seattle, Washington 98122
Correspondence: Address all correspondence and requests for reprints to: Lydia Aguilar-Bryan and Joseph Bryan, Pacific Northwest Diabetes Research Institute, 720 Broadway, Seattle, Washington 98122. E-mail: lbryan{at}pnri.org or jbryan{at}pnri.org
 |
Abstract
|
|---|
An explosion of work over the last decade has produced insight into the multiple hereditary causes of a nonimmunological form of diabetes diagnosed most frequently within the first 6 months of life. These studies are providing increased understanding of genes involved in the entire chain of steps that control glucose homeostasis. Neonatal diabetes is now understood to arise from mutations in genes that play critical roles in the development of the pancreas, of β-cell apoptosis and insulin processing, as well as the regulation of insulin release. For the basic researcher, this work is providing novel tools to explore fundamental molecular and cellular processes. For the clinician, these studies underscore the need to identify the genetic cause underlying each case. It is increasingly clear that the prognosis, therapeutic approach, and genetic counseling a physician provides must be tailored to a specific gene in order to provide the best medical care.
- I. Introduction
- II. Diagnosis of Neonatal Diabetes Mellitus
- III. Frequency of TNDM vs. PNDM
- IV. Etiology of Transient Neonatal Diabetes Mellitus (OMIM 601410, 600937, 600509)
- V. Chromosome 6q Anomalies (OMIM 601410) Are the Main Cause of TNDM
- VI. Hepatic Nuclear Factor 1β (OMIM 189907)
- VII. Permanent Neonatal Diabetes Mellitus
- VIII. Abnormal Pancreatic Development
- A. Pancreas transcription factor 1,
subunit (PTF1a) (OMIM 607194, 609069)
- B. Pancreatic and duodenal homeobox 1/insulin-promoter-factor 1 (PDX1/IPF-1) (OMIM 600733)
- C. GLIS subfamily of Kruppel-like zinc finger proteins - 3 (GLIS3) (OMIM 610199)
- IX. PNDM Attributable to Increased β-Cell Apoptosis or Necrosis: Pancreatic Eukaryotic Initiation Factor 2 Kinase (EIF2AK3); Wolcott-Rallison Syndrome (OMIM 226980)
- X. Insulin
- XI. Immunodysregulation, Polyendocrinopathy, and Enteropathy, X-Linked (IPEX), FOXP3 (OMIM 304790/304930)
- XII. PNDM Attributable to β-Cell Dysfunction and Impaired Glucose-Stimulated Insulin Secretion
- XIII. Glucokinase (GCK) (OMIM 606176)
- XIV. TNDM and PNDM Attributable to Reduced Insulin Secretion Secondary to Overactivity of ATP-Sensitive Potassium Channels, KCNJ11 and ABCC8 (OMIM 600509, 600937, 610374)
- A. Clinical aspects
- B. Channel aspects
- XV. Location of KCNJ11 and ABCC8 Mutations
- XVI. Sulfonylureas Are an Effective Therapy for NDM Caused by Overactive KATP Channels
- XVII. Summary
"With few exceptions, diabetes in childhood knows no cure, no matter how mild it may appear in the beginning, nor how gradual its development in the first months or even years."
Carl von Noorden, great European authority, 1913
"All cases which have come to my attention of youthful patients with diabetes living for many long periods of time have been hereditary."
E. P. Joslin, 1934
|
I. Introduction
|
|---|
DIABETES MELLITUS PRESENTING as uncontrolled hyperglycemia during the first 6 months of life is a rare disorder that affects all races and ethnic groups. The majority of the cases present with intrauterine growth retardation (IUGR), failure to thrive, decreased sc fat, and low or undetectable C-peptide levels (1). This form of hyperglycemia has been termed "early-onset" or neonatal diabetes mellitus (NDM) and is commonly of genetic origin. Because the neonatal period is defined as the first 4 wk of life, whereas diagnosis of these cases extends through 6 months of age, a reviewer pointed out that "congenital diabetes mellitus" is a more valid descriptor for a disorder that is present at birth although not always clinically apparent immediately. However, the term "neonatal diabetes" has become established in the literature; we use the terms interchangeably. Kitselle (2) in 1852 is credited with the first clinical description of the disorder that was present in his son (3, 4). Temple and Shield (5) comment on the history of the disease, noting an early report by Ramsey (6) of a low birth weight boy who required insulin to control his transitory diabetes. Hutchinson et al. (7) were the first to distinguish the permanent (PNDM) vs. the frequently relapsing transient (TNDM) forms of congenital or neonatal diabetes, a term attributed to Gentz and Cornblath (8). A follow-up study of published cases by von Muhlendahl and Herkenhoff (9) in the New England Journal of Medicine established that a high percentage of children with TNDM relapsed and developed type 2 diabetes years after the initial hyperglycemic period. The etiology of NDM is genetically heterogeneous, producing abnormal development or absence of the pancreas or islets, decreased β-cell mass secondary to increased β-cell apoptosis or destruction, and β-cell dysfunction that limits insulin secretion. Recent studies defining the multiple underlying mechanisms that give rise to TNDM and PNDM continue to illuminate the developmental aspects and basic physiology of glucose homeostasis as well as increasing understanding of individual genes that may play a role in the etiology of more common polygenic forms of diabetes mellitus. Although the physiology underlying some of the genetic causes of NDM is relatively well understood, the mechanism(s) behind the remission, characteristic of TNDM, and subsequent relapse remain unexplored.
|
II. Diagnosis of Neonatal Diabetes Mellitus
|
|---|
Diagnosis of "early-onset" diabetes can occur within the first days or months of life with presentation of hyperglycemia. In rare cases there are neural complications. The time of presentation is variable, and a potential diagnostic problem is the differentiation of a monogenic cause vs. autoimmune type 1 diabetes in these early-onset children. In a study of 111 diabetic children who required insulin within the first year of life, Iafusco et al. (10) reported a greater frequency of protective HLA antigens and less frequent autoimmune markers in children with diabetes onset before 180 d vs. those with onset greater than 180 d. Many of these newborns were small for gestational age, a finding strongly correlated with diabetes onset before 180 d. Several studies indicate that the preponderance of diabetic cases identified before 6 months of age are of monogenic origin (11, 12, 13) with no evidence for autoimmune markers of β-cell destruction (13, 14, 15). Although sero-conversion has been reported in some patients with long-standing congenital diabetes (16), the available data strongly support the argument that cases of diabetes diagnosed before 6 months of age are probably of monogenic origin and thus are candidates for genetic screening.
|
III. Frequency of TNDM vs. PNDM
|
|---|
NDM is rare, variously quoted as one case per 300,000 to 500,000 live births (see Refs. 17 and 18 for reviews); Stanik et al. (19) have estimated the frequency of PNDM in Slovakia at one in 215,417 live births. Table 1
summarizes data from published studies where NDM cases were stratified into TNDM vs. PNDM. The number of cases continues to accumulate, but the available, combined data indicate that somewhat over half (
57%) of NDM cases are transient, require insulin treatment initially, and spontaneously resolve in less than 18 months, only to relapse in later years.
In a follow-up of cases in the literature, von Muhlendahl and Herkenhoff (9) noted that a reduced birth weight for gestational age was characteristic of NDM (Fig. 1). This finding has been confirmed in multiple studies that included several racial and ethnic groups by Metz et al. (20), Babenko et al. (21), and Vaxillaire et al. (22) for cases from the French Network for the Study of Neonatal Diabetes; by Flanagan et al. (12) for a cohort of 97 cases with TNDM; by Flanagan et al. (11) for 37 cases of PNDM resulting from mutations in KCNJ11 (potassium inwardly-rectifying channel, gene identifier for KIR6.2); by Suzuki et al. (23) in 31 Japanese individuals; by Rica et al. (24) for 22 Spanish cases; and by Massa et al. (25) for 12 cases of Spanish origin. There is considerable overlap in the birth weights of TNDM and PNDM individuals as shown in Fig. 1, thus limiting the diagnostic potential of this parameter.
View larger version (29K):
[in this window]
[in a new window]
|
FIG. 1. Birth weights of 45 patients with NDM. The percentiles are those for normal girls from the study by Weller and Jorch (275 ). The third percentile for boys is higher by 20 g at 24 wk gestation, 125 g at 34 wk, and 150 g at term. Closed symbols denote girls; open symbols, boys; circles, infants with TNDM; triangles, infants with TNDM with later recurrence; and squares, infants with PNDM. [Reproduced with permission from K. E. von Muhlendahl and H. Herkenhoff: N Engl J Med 333:704–708, 1995 (9 ). © Massachusetts Medical Society.]
|
|
The IUGR can be attributed, in part, to insulin acting as a fetal growth factor, coupled with the failure of maternal insulin to cross the placental barrier (26) unless it is bound to an antibody (27, 28). The growth effects of fetal insulin are broadly supported by the incidence of macrosomia in neonates born to mothers that are hyperglycemic secondary to poorly controlled diabetes or gestational diabetes (Ref. 29 ; reviewed in Refs. 30 and 31), by the large birth weight associated with cases of familial hyperinsulinism (HI) (32, 33), and by studies on knockout mice in which insulin action is impaired by altering expression of proteins in the insulin signaling pathway (34).
Does the rapid "catch-up" growth characteristic of TNDM contribute to relapse? Rapid "catch-up growth" has been reported in patients with NDM after the initiation of insulin treatment and a high-caloric diet (35). This is consistent with the idea that insulin acts as a growth factor in the fetus, but mainly as a regulator of energy metabolism after birth. Similar catch-up growth has been reported in individuals born small for gestational age as a consequence of IUGR, and multiple studies have linked the velocity of early growth with a predisposition for subsequent development of type 2 diabetes (for review, see Refs. 36 and 37) and the metabolic syndrome (38). The molecular mechanism(s) underlying this predisposition are controversial and not firmly established, but increased insulin resistance (39), perhaps as a consequence of reduced adiponectin levels (Ref. 40 ; but see Ref. 41), has been suggested as a contributing factor to the subsequent development of obesity and diabetes. Although the IUGR in these studies is secondary to maternal malnutrition, placental insufficiency, and/or other environmental factors, the low birth weight and catch-up growth in NDM of genetic origin might increase insulin resistance and contribute to the later relapse in these children. In this regard, Valerio et al. (42) report no evidence for insulin resistance in four patients with TNDM resulting from 6q24 anomalies, whereas others (43) described impaired insulin sensitivity in four cases of PNDM resulting from mutations in KCNJ11 (R201H and K170N) and showed improved sensitivity after a switch to sulfonylurea therapy (44). Studies using mouse models that lack the KCNJ11/ABCC9 type skeletal muscle KATP channels exhibit an improved insulin-dependent glucose uptake (45, 46). This suggests that the increased channel activity that arises from KCNJ11 NDM mutations might underlie the impaired insulin sensitivity observed by Skupien et al. (43) and contribute to their diabetes. In this regard, it will be of considerable interest to determine whether NDM cases resulting from mutations in ABCC8, a subunit of neuroendocrine, but not skeletal muscle KATP channels, exhibit comparable impaired insulin sensitivity and whether there are differences in the frequency of relapse of TNDM cases attributed to these various genetic causes.
|
IV. Etiology of Transient Neonatal Diabetes Mellitus (OMIM 601410, 600937, 600509)
|
|---|
The hyperglycemia characteristic of TNDM is the result of reduced or absent insulin output during the fetal period that extends for a variable time into postnatal life. The genetic origin for more than 90% of TNDM cases has been established. Recent work suggests that the deficit in insulin output can arise either from delayed maturation of pancreatic islets and β-cells as a consequence of the altered expression of imprinted genes on chromosome 6 or from β-cell dysfunction that impairs insulin secretion. In the first case, islets and β-cells are poorly developed with reduced or absent insulin; in the latter instance, insulin is present but glucose sensing is defective, thus abrogating insulin release. In either case, reduced fetal insulin, acting as a growth factor, is expected to slow fetal growth.
Figure 2 summarizes pooled data from three studies of children diagnosed with TNDM where the genetic basis was identified. The majority of cases (68%) are due to abnormalities in the 6q24 region, whereas 10 and 13% of cases are attributable to mutations in KCNJ11 and ABCC8, respectively. In a study of 97 TNDM cases (12), it was noted that neonates with 6q24 anomalies had a lower average birth weight (1950 vs. 2570 g) and were diagnosed earlier (0 vs. 4 wk) compared with those with KATP channel mutations. In addition, the average period of neonatal hyperglycemia attributed to KATP channel mutations was longer (35 vs. 13 wk). During the hyperglycemic period, the insulin secretory response to glucose and other secretagogues was impaired, and insulin therapy was required to achieve normoglycemia. Ketoacidosis was present in some cases. After several weeks on insulin and high-caloric feeding, there was striking weight gain, and after more than 3 months, a range of 3–18 months, insulin could be discontinued. The remission period can last several years. For example, Schiff et al. (35) describe normal glucose tolerance (iv glucose tolerance test) in a TNDM case in remission over a 2-yr span. Shield et al. (47) analyzed a group of TNDM patients reporting relatively or entirely normal measures of β-cell function and insulin sensitivity in five individuals over several years; a sixth case exhibited a deficient insulin secretory response to iv glucose. Subsequently, perhaps as a consequence of developing insulin resistance and increased demands on β-cells during puberty, about 40% of cases relapse and present with common features of type 2 diabetes mellitus. The percentage of cases developing diabetes increases with age (12), and the 40% figure will undoubtedly be a minimal estimate as follow-up studies progress. Further work will determine whether relapse is secondary to the low birth weight and restricted intrauterine growth and can be connected to studies associating low birth weight in the general population with the development of type 2 diabetes, and to determine whether clinical markers can be identified to distinguish TNDM from PNDM.
View larger version (19K):
[in this window]
[in a new window]
|
FIG. 2. Breakdown of the genetic causes of TNDM derived from studies on French, English, and Japanese cohorts. The data are from Metz et al. (20 ), Babenko et al. (21 ), Flanagan et al. (12 ), and Suzuki et al. (23 ).
|
|
|
V. Chromosome 6q Anomalies (OMIM 601410) Are the Main Cause of TNDM
|
|---|
A report (48) that insulin was preferentially expressed from the paternal allele in the mouse embryo yolk sac led Haig (49) to propose that a loss-of-function mutation in the paternal insulin gene, coupled with silencing (imprinting) of the maternal allele, might provide a mechanism to account for TNDM—the thought being that the hyperglycemia would resolve after postnatal activation of the maternal allele. This idea did not prove to be correct; however, the importance of imprinting for TNDM was shown in the seminal observation by Temple and Shield (50) of paternal uniparental isodisomy (UPD) of an imprinted region at chromosome 6q in two cases of TNDM. This finding has been confirmed extensively (reviewed in Refs. 5, 20 , and 51). Overexpression of genes in the 6q24 locus either as a consequence of loss of imprinting at 6q24 by UPD (50, 52, 53, 54, 55), by duplication of this region (paternal duplication) (56), or by loss of DNA methylation, and thus activation of the maternal allele (20, 57), is the most common cause of TNDM (Fig. 3). There are multiple imprinted genes in this region (see the imprinted gene catalog at www.otago.ac.nz/IGC), including ZAC [zinc finger protein which regulates apoptosis and cell cycle arrest—also called LOT1, for lost on transformation (58)] and HYMAI (hydatiform mole-associated and imprinted—also called PLAGL1, for pleomorphic adenoma of the salivary gland gene like 1). HYMAI is an untranslated RNA of undetermined function. ZAC is a C2H2 zinc-finger transcription factor (59) with multiple functions, including acting as a coactivator with p53 (60) of Apaf1 (apoptotic protease activating factor 1) (61) transcription, regulating the histone acetyl transferase activity of p300 (62), and serving as a coactivator or corepressor of several nuclear hormone receptors (63). ZAC and HYMAI are transcribed from overlapping genes, and a differentially methylated cytosine and guanine separated by a phosphate (CpG) island important for regulation lies at their 5' end. The DNA of the ZAC gene is unmethylated in sperm and methylated in oocytes. The maternal alleles of ZAC and HYMAI are both silent, thus the loss of this imprinting is expected to increase the expression of both gene products. Little is known about the specific role(s) of ZAC or HYMAI during pancreatic islet development, but studies in other cell systems provide insight and suggest that the hypothesis that increased expression of tumor suppressor genes contributes to the IUGR associated with the disorder. In this hypothesis, up-regulation of a tumor suppressor gene(s) would lead to a reduction of fetal β-cell mass and thus IUGR secondary to reduced insulin. The 6q24–25 locus is a cancer hot spot, and ZAC has been described independently as a tumor suppressor gene that is reduced or absent in several types of tumors including nonfunctional pituitary tumors (64), primary breast tumors, and in cell lines isolated from breast and ovarian tumors (65, 66, 67). Overexpression of ZAC in cell lines resulted in a decreased rate of cell replication, increased apoptosis, and G1 arrest (58, 68). In short, overexpression of ZAC could be anticipated to reduce growth rate and subsequently β-cell mass.
View larger version (16K):
[in this window]
[in a new window]
|
FIG. 3. Potential roles of ZAC in chromosome 6q24 anomalies. A, The state of ZAC expression in control and TNDM cases due to paternal UPD or loss of imprinting (LOI) at the maternal locus. B, Four potential mechanisms by which overexpression of ZAC could reduce cell proliferation or increase apoptosis.
|
|
The targets responsible for the antiproliferative effects of ZAC are under active investigation, and peroxisome proliferator-activated receptor 
(PPAR), an insulin sensitizer and tumor suppressor gene expressed in islets (69), is a potential candidate. Barz et al. (70) have shown that ZAC can bind to the PPAR1 proximal promoter and activate transcription in human colon carcinoma cells. Somatostatin analogs have an antiproliferative effect on pituitary tumors, and the treatment of pituitary tumor cells with the somatostatin analog, octreotide, induces the transcription of ZAC, which in turn up-regulates the expression of PPAR to slow proliferation. The targeted deletion of PPAR in β-cells results in marked islet hyperplasia (71). PPAR β knockout mice exhibit no significant changes in glucose homeostasis when maintained on lab chow, but interestingly show an impaired increase in β-cell mass in response to a high-fat diet (71). Administration of thiazolidinediones to islets and β-cell lines is reported to potentiate glucose-stimulated insulin secretion (72), enhance fatty acid oxidation (73), and reduce cell proliferation (71). Although a role for PPAR in islet/β-cell development has not been examined, an increase in ZAC expression during embryogenesis could increase PPAR expression and slow β-cell proliferation.
ZAC has also been implicated in the regulation of an imprinted region on human chromosome 11 (11p15.5) near the Beckwith-Wiedemann fetal overgrowth syndrome (BWS) locus and two classic imprinted growth regulatory genes, igfr2 and igfr2r. Specifically, ZAC is reported to bind to and regulate a differentially methylated region, KvDMR, in the KvLQT1/KCNQ1 gene. Loss of methylation and activation of the maternal allele of KvDMR are frequent alterations associated with BWS (74). KvDMR is a complex multipartite, multifunctional regulatory region with distinct promoter, repressor, and enhancer modules (75). Paternal inheritance of a deletion of KvDMR results in the loss of expression of a noncoding RNA, Kcnq1ot1 (also called LIT1), and the derepression in cis of imprinted, normally silent, paternal genes on both sides of the deletion (76, 77). Two mechanisms have been suggested to account for the gene silencing in this region: 1) blocking of transcription enhancer activity by a process called "CTCF-mediated insulation" which involves a ubiquitously expressed zinc-finger protein, CTCF; or 2) a process that involves the noncoding RNA, Kcnq1ot1. A recent dissection of the KvDMR locus suggests that both mechanisms may be operating (for example, see Refs. 77 and 78). One gene that is normally expressed exclusively from the maternal chromosome is Cdkn1c, a cyclin-dependent kinase inhibitor also called P57KIP2. Cdkn1c/P57KIP2 is a tumor suppressor gene whose down-regulation contributes to the fetal overgrowth characteristic of BWS. Mutations in the maternal copy of Cdk1c/P57KIP2, while infrequent, are a cause of BWS (79). In rodents, overexpression of Cdkn1c/P57KIP2 significantly reduced fetal growth, whereas deletion of the Cdkn1c/P57KIP2 gene resulted in overgrowth (80). The data are consistent with the idea that an increase in the level of ZAC, secondary to UPD or loss-of-imprinting at the 6q24 locus, can up-regulate tumor suppressor genes including PPAR and Cdkn1c and thus contribute to the growth restriction characteristic of TNDM (summarized in Fig. 3
).
The molecular link(s) between 6q24 anomalies and islet development have not been elucidated directly. It seems reasonable to speculate that the effects of increased ZAC and its downstream targets would lead to reduced fetal β-cell mass and concomitant reduction in fetal insulin. Potentially, the combined effect of tumor suppressor gene activity and reduced fetal insulin could synergize to produce a greater growth restriction than reduced insulin secretion alone. This synergism would be consistent with the lower average birth weights associated with 6q24 anomalies vs. overactive KATP channels. The potential overlap between genes or chromosomal regions involved in BWS and TNDM is interesting and could help to explain the overlapping features, macroglossia and umbilical hernia, observed in both disorders.
To develop an animal model of TNDM, Ma et al. (81) generated mice carrying multiple copies of the human 6q24 region in an effort to enhance expression from the 6q24 locus. These animals recapitulate the TNDM phenotype seen in human neonates. TNDM mouse neonates were hyperglycemic, whereas older adults shown impaired glucose tolerance. Neonatal hyperglycemia occurred only with paternal transmission of the transgene, consistent with the paternal dependence of TNDM in humans. Significantly, pancreata of TNDM mouse embryos showed fewer positive structures staining for insulin, glucagon, somatostatin, and pancreatic polypeptide and had reduced expression of several endocrine differentiation factors consistent with impaired islet development. At postnatal stages, β-cell mass was normal or increased, although neonatal pancreatic insulin content and adult peak serum insulin levels in response to glucose infusion were reduced. The mouse phenotype is consistent with the hypothesis that overexpression of a gene or genes in the 6q TNDM locus restricts or slows islet development.
|
VI. Hepatic Nuclear Factor 1β (OMIM 189907)
|
|---|
Mutations in hepatocyte nuclear factor-1β (HNF1β), also called transcription factor-2 (TCF2), are responsible for two syndromic diabetes phenotypes, maturity-onset diabetes of the young (MODY) 5 and TNDM. In these patients, hyperglycemia cosegregates with renal abnormalities and genital malformations including vaginal and Müllerian aplasia (82, 83, 84). HNF1β/TCF2 is a member of the POU-homeobox family of basic helix-loop-helix proteins that bind to DNA as dimers. HNF1β is structurally similar to HNF1 (TCF1), with greatest identity in their DNA-binding domains. Although termed hepatic factors, their tissue distribution is not restricted to the liver, being also present in the kidney, gut, genital tract, thymus, lung, and pancreas. N-terminal dimerization domains allow these transcription factors to homo- or heterodimerize and transactivate a variety of genes including those for insulin (85), polycystic kidney hepatic disease 1 (PKHD1) (86), and suppressor of cytokine signaling-3 (SOCS-3) in kidney epithelial cells (87). These presumably represent a small subset of the genes potentially regulated by HNF1β as the up- and down-regulation of a large number of genes, more than 200, in mouse hepatoma cells treated with HNF1β-targeted RNA interference has been reported (88).
Heterozygous mutations in HNF1β are responsible for MODY5 as shown initially by Horikawa et al. (89). More than 30 nonsense, missense, and frame-shift mutations are currently known, and the structural basis for their effects involve both impaired DNA-binding and loss of association with transcriptional coactivators (90, 91). The loss of HNF1β expression in different tissues is consistent with the syndromic phenotype. HNF1β is expressed in visceral endoderm at the onset of gastrulation and is essential for the differentiation of visceral endoderm. Consistent with a role in early visceral development, there are several reports of MODY 5 patients with pancreatic atrophy (90, 92). The complete loss of HNF1β in TCF2 null mice results in lethality in utero at embryonic day 7.5 (E7.5), resulting in a fetus with disorganized visceral endoderm (93). Haumaitre et al. (94) partially rescued TCF2 null animals and showed that the complete lack of HNF1β, normally expressed in pancreatic buds, results in the absence of a pancreas. In vitro studies have identified an HNF1 binding site in the promoter of the PKHD1 gene and showed that HNF1 and HNF1β bind and stimulate PKHD1 transcription (86, 95). The PKHD1 gene encodes a large, single transmembrane-spanning protein called fibrocystin, which is located on the primary cilia of renal cells and is present in fetal and adult kidney cells and to a lesser extent in liver and pancreas (96, 97). Loss of fibrocystin produces abnormal cilia and leads to the formation of renal cysts. HNF1β has also been shown to bind to the SOCS-3 promoter and repress SOCS-3 transcription (87). Overexpression of a dominant-negative HNF1β increases SOCS-3 mRNA levels significantly and suppresses tubule formation (87). Morphogenetic growth factors including hepatocyte growth factor are important for kidney tubule formation via stimulation of epidermal growth factor/ hepatocyte growth factor receptors that activate JAK/STAT and MAPK pathways. Increased SOCS-3 can suppress this activation and impair formation of kidney tubules. Therefore, mutations in TCF2 that impair the ability of HNF1β to stimulate transcription can affect renal development by both up- and down-regulation of gene expression.
HNF1β is a key member of the network of transcription factors controlling the differentiation of the endodermal pancreatic precursor cells that assemble the exocrine and endocrine pancreas. HNF1β has been shown to regulate expression of pancreas transcription factor 1 (PTF1a)/P48, part of the PTF1 transcription factor critical for the formation of the exocrine pancreas (see Section VIII), to bind to and activate transcription of HNF6 (98), a ONECUT transcription factor required for expression of Ngn3, a factor critical for β-cell differentiation. The partial loss of HNF1β in mice with a targeted deletion of TCF2 in β-cells resulted in increased expression of HNF1 and pancreatic and duodenal homeobox 1 (PDX1), decreased expression of HNF4 and HNF4, and impaired glucose-stimulated insulin release (99).
Yorifuji et al. (84) were the first to report a mutation in HNF1β that produced transient NDM. The S148W mutation was identified in heterozygosity in two siblings presenting variable phenotypes. The first had NDM and a few small renal cysts, but normal renal function. The second had a short episode of hyperglycemia, neonatal polycystic kidneys, and early renal failure. The parents were asymptomatic. Genetic analysis indicated that the mother was a low-level mosaic of normal and mutant TCF2, suggesting that the phenotypic heterogeneity of the children may have resulted from germline mosaicism. A second heterozygous mutation, S148L, was identified during screening of 27 NDM patients for whom no known mutation was found (100). The patient was diagnosed at 17 d of age with transient diabetes that relapsed 8 yr later. Low birth weight, pancreatic atrophy, and exocrine insufficiency were also present, consistent with the role of HNF1β in pancreatic development.
|
VII. Permanent Neonatal Diabetes Mellitus
|
|---|
As with the transient form of NDM, the hallmark of PNDM is hyperglycemia early in life, but without the period of remission that defines TNDM. To date, 10 genes involved in pancreatic development, β-cell apoptosis, or dysfunction have been identified as being able to give rise to PNDM. In most cases, TNDM and PNDM cannot be distinguished clinically, and genetic analysis needs to be performed because the identification of a known molecular defect will determine the clinical prognosis, treatment, and genetic counseling. Although hyperglycemia is the main manifestation, the majority of PNDM cases also present IUGR and failure to thrive (reviewed in Ref. 101). Diabetes has also been reported in syndromic disease reflecting the roles the affected genes have beyond the β-cell, for example as described for HNF1β. Table 2 summarizes the genes identified to date. The etiologies are grouped loosely in terms of abnormal pancreatic development, reduction in β-cell mass due to increased apoptosis or necrosis, and β-cell dysfunction. It is worth noting that mutations in two genes, PDX1/IPF-1 and glucokinase (GCK), that cause MODY in the heterozygous condition produce PNDM in the homozygous state, whereas TNDM has been seen in children with two MODY5 (HNF1β) mutations as described in Section VI.
|
VIII. Abnormal Pancreatic Development
|
|---|
A. Pancreas transcription factor 1, subunit (PTF1a) (OMIM 607194, 609069)
Hoveyda et al. (102) identified a recessively inherited syndrome in three children from an inbred Pakistani family that was characterized by NDM, severe IUGR, microcephaly, facial dysmorphism, respiratory distress, and hypoplastic cerebellar tissue. The family had a strong history of type 2 diabetes mellitus, with more than five relatives presenting the disease in their early thirties in the absence of obesity and one case of gestational diabetes. Sellick et al. (103) reported on a second family of Caucasian northern European ancestry with the same features, including low, but detectable C-peptide levels and the absence of pancreatic tissue upon postmortem examination. Genome-wide linkage analysis identified a new locus for NDM on 10p13-p12.1, and positional cloning uncovered two mutations, 705insG
R296X and C886TP236fsX270, in the PTF1a (or PTF1a/P48) gene in these families.
PTF1a/P48 is a member of the basic helix-loop-helix family of transcription factors and is essential for the development and maintenance of the adult pancreas (104, 105) and for development of the cerebellum. PTF1a/P48 is a subunit of the unusual, heterotrimeric transcription factor, PTF1, that consists of one PTF1a/P48 subunit, one of several ubiquitous class A basic helix-loop-helix proteins [e.g., E12, E47, HEB reported to activate expression of several cyclin-dependent kinase inhibitors (106)], and a third subunit identified as either the mammalian Suppressor of Hairless (RBP-J) or its paralogue, RBP-L (107). The composition of PTF1 changes with development, with RBP-L directing high-level transcription in the adult exocrine pancreas. Two conserved tryptophan-containing motifs are important for the binding of the RBP isoforms to PTF1a/P48. The two disease mutations truncate one or both motifs and thus impair the binding of PTF1a/P48 to RBP-J and to a lesser extent to RBP-L (107). Deletion of PTF1a/P48 in mice results in neonatal death with pancreatic and cerebellar agenesis (103, 105, 108), consistent with the human phenotype. PTF1a/P48 is expressed early in embryogenesis (E9.5 in mice) throughout the developing pancreas. The ventral pancreatic bud is absent in PTF1a/P48 null mice, outgrowth of the dorsal bud is impaired, and exocrine cells fail to differentiate. Interestingly, the patients have low, but detectable levels of C-peptide (103), and in PTF1a/P48 null mice the pancreatic endocrine cells are present, albeit in reduced numbers and mislocalized to the spleen. PTF1 regulates the expression of multiple genes, but recent work demonstrates that it binds to a conserved element, termed area III, in the promoter of PDX1, a key player in the network of transcription factors that regulate the development and maintenance of the adult endocrine and exocrine pancreas (109, 110). The results are consistent with the idea that PTF1 transactivates the expression of the PDX1 gene via binding to area III and that this expression generates PDX1+ cells required for the differentiation of the exocrine pancreas (Fig. 4). The persistence of endocrine cells in both patients and PTF1a/P48 null mice suggests that there is a parallel or synergistic pathway controlling PDX1 expression and that regulation by PTF1 is not essential for islet cell differentiation and maintenance. Further studies would be needed to determine whether the reduced endocrine mass is a consequence of reduced PDX1 level vs. loss of pancreatic tissue architecture.
View larger version (28K):
[in this window]
[in a new window]
|
FIG. 4. Schematic representation of the central role of PDX1/IPF-1 within the network of transcription factors critical for pancreas differentiation and development. The accompanying table relates MODY classification with the affected protein. Glut2, Glucose transporter 2; L-PK, L-pyruvate kinase.
|
|
The focus of this review is diabetes; therefore our discussion of PTF1a has centered on the pancreas. Impaired cerebellar development is also a feature of cases with PTF1a/P48 mutations, and the reader is urged to see Ref. 111 for a review of the role of PTF1a in that arena.
B. Pancreatic and duodenal homeobox 1/insulin-promoter-factor 1 (PDX1/IPF-1) (OMIM 600733)
Initially cloned by Thomas Edlund’s group (112), PDX1, also known as IPF-1 or insulin-promoter factor 1, is a homeodomain transcription factor that plays a critical role as noted above in the formation of the pancreas by determining the fate and regulating the propagation of both pancreatic exocrine and endocrine precursor cells (113). Mice with a targeted mutation in the PDX1/IPF-1 gene expressed in the homozygous state lack a pancreas and die within a few days of birth (114). Subsequent work in a number of laboratories has shown that PDX1/IPF-1 is a central regulator in the interacting network of transcription factors that govern islet cell differentiation and development (115, 116, 117, 118). Mutations in several of the genes encoding these transcription factors are known to cause MODY (Fig. 4; for review, see Ref. 119).
The importance of PDX1/IPF-1 in human pancreas development was confirmed when a child with a single-nucleotide deletion in exon 1 (Pro63fsdelC) of the human PDX1/IPF-1 gene was described (120). This mutation produces a frame shift that truncates PDX1 prematurely and results in pancreatic agenesis. The child was homozygous for the mutation and required insulin treatment and pancreatic enzymes to replace pancreatic function. Within the child’s extended family, eight relatives in six generations were heterozygous for the mutation and developed early-onset type 2 diabetes (MODY4) (121, 122). The diabetes in these cases was generally treatable with diet, oral hypoglycemic agents, and in a few instances with insulin. The findings support the idea that haploinsufficiency of PDX1 impairs β-cell function and the increasing demand for insulin with age results in hyperglycemia, whereas the lack of PDX1 blocks endocrine and exocrine tissue differentiation. This would suggest that other MODY genes, if present in the homozygous state, would produce NDM, an idea supported by the finding discussed below that homozygous loss of GCK is a cause of NDM.
A second case was reported by Schwitzgebel et al. (123) in a patient with compound heterozygosity of two mutations in exon 2, E164D and E178K. This patient presented with low birth weight and length and was diagnosed at 12 d of age with a glycemia of 854 mg/dl (47 mmol/liter). DNA binding was unaffected, but the two mutations significantly decreased the half-life of PDX1 (36 and 27%). The reduced PDX1 content altered transcriptional activity, implying that the level of expression is important for pancreatic development. The parents, each a carrier of one mutation, had slightly elevated fasting glucose levels with normal oral glucose tolerance test. The severity of the phenotype correlated with the mutation, the reduced activity attributed to a single heterozygous mutation being insufficient to give a severe diabetes phenotype.
C. GLIS subfamily of Kruppel-like zinc finger proteins-3 (GLIS3) (OMIM 610199)
Taha et al. (124) reported a family with a novel autosomal recessive syndrome in two infants affected with multiple organ involvement. Both siblings presented with IUGR, nonautoimmune congenital diabetes, severe congenital hypothyroidism, cholestasis and subsequent hepatic fibrosis, congenital glaucoma, polycystic kidneys, and minor facial abnormalities. The children were the product of a consanguineous family of Saudi Arabian descent with no history of diabetes, thyroid, liver, and eye or kidney disease. The two infants, a girl and a boy, were the family’s first and fourth children (two normal siblings) and were born after full-term and 37-wk pregnancies without complications. The index cases presented with IUGR and hyperglycemia in the first days after birth, supporting the idea that they had intrauterine insulin deficiency secondary to defects in pancreatic development. The rest of the manifestations appeared in subsequent weeks. Computed tomography scan of the abdomen demonstrated hepatosplenomegaly with a small pancreas in the first case and no visualization of the organ in the second case. The hyperglycemia and hypothyroidism were brought under control, but the infants died at ages 14 and 4 months, respectively, after pneumonia and Escherisia coli sepsis.
Senee et al. (125) undertook a genome-wide scan in the available members of the original family plus two other consanguineous families with a similar but more heterogeneous syndrome where congenital diabetes and congenital hypothyroidism were present. Mutations in GLI Similar-3, a novel transcription factor of the GLIS subfamily of Kruppel-like zinc finger proteins, were identified as responsible for the syndrome. This novel transcription factor was identified and characterized by Kim et al. (126) as a member of the GLI family of proteins that play an important role in neuronal and skeletal development in mammals. GLIS3 can function as both a repressor and activator of transcription via binding to GLI-response element consensus sequences. Both N and C termini are necessary for optimum transcriptional activity, including the zinc finger motif and the nuclear localization signal. Mouse studies suggest that GLIS3 plays an important role in multiple cellular processes during development, including a possible role in apoptosis in the interdigital regions (126). GLIS3 is expressed in the pancreas mainly in β-cells at an early developmental stage. The multiorgan involvement implies a wider role for GLIS3 in thyroid, eye, liver, and kidney development.
In the first family reported by Taha et al. (124), patients were homozygous for an insertion (2067insC) that caused a frame shift and resulted in a truncated protein (625fs703stop). In the other two families, the probands had two distinct deletions affecting the 11 or 12 most 5' exons of the gene. The individuals studied in the second family were homozygous for a 426-kb deletion encompassing part of the 5' untranslated region of GLIS3 and SLC1A1, the high-affinity glutamate transporter. In the third family, the homozygous probands carried a 149-kb deletion that overlapped a small portion of the GLIS3 5' untranslated region. These changes reduced the expression of GLIS3 by more than 90%. Senee et al. (125) also looked at expression in human and rodent tissues in the developing and adult pancreas and showed that GLIS3 is expressed as a major transcript as early as E15.5, increases after birth, and is preferentially expressed in β-cells, consistent with an important role in β-cell development.
|
IX. PNDM Attributable to Increased β-Cell Apoptosis or Necrosis: Pancreatic Eukaryotic Initiation Factor 2 Kinase (EIF2AK3); Wolcott-Rallison Syndrome (OMIM 226980)
|
|---|
In the early 1970s, Wolcott and Rallison (127) reported a novel recessive disorder in three siblings presenting with permanent congenital or infancy-onset diabetes mellitus, multiple epiphyseal dysplasia, and growth retardation. A decade later, a brother and a sister were reported with the same phenotype (128), supporting the hypothesis that the association of endocrine (NDM) and chondro-osseous (epiphyseal and spondylo-epiphyseal dysplasias) abnormalities could be the manifestation of a pleiotropic gene. After these two reports, other cases with additional clinical signs including learning difficulties, hepatic and renal dysfunction, cardiac abnormalities, and exocrine pancreatic dysfunction have been reported (129, 130, 131). The heterogeneity of this disorder was underscored by the observation at autopsy of a Wolcott-Rallison syndrome (WRS) case with severe pancreatic hypoplasia and congenital diabetes, abnormal bone histology, cardiomegaly, mental retardation and cerebellar cortical dysplasia, and hepatic and renal dysfunction (130, 131). This patient had a mosaic deletion of part of chromosome 15 (15q11–12) in 65% of examined karyotypes. Islet architecture was disorganized, with few insulin-positive cells and a preponderance of glucagon-positive cells.
Using linkage analysis, Delepine et al. (132) identified eukaryotic translation initiation factor 2 kinase 3 (EIF2AK3; also called PERK) as the WRS gene in two consanguineous families of Tunisian and Pakistani descent. EIF2AK3/PERK is highly expressed in pancreatic islets (133), plays an important role in the regulation of protein translation and the unfolded protein response (UPR) (134, 135), and maps to chromosome 2p12, a previously identified WRS locus. EIF2AK3/PERK is a single pass transmembrane protein with a lumenal domain that binds unfolded proteins in the endoplasmic reticulum (ER) and a cytoplasmic kinase domain. In professional secretory cells like β-cells, EIF2AK3/PERK plays a pivotal role in the UPR, a homeostatic signaling pathway that adjusts the protein-folding capacity of the ER in response to demand. EIF2AK3/PERK is one of three types of sensors that recognize unfolded proteins in the ER and activate the transcription of multiple genes to increase the ER-folding capacity or, if the stress is sufficient, to initiate apoptosis (reviewed in Refs. 136, 137, 138, 139). The binding of unfolded protein to EIF2AK3/PERK results in phosphorylation of eIF2, a factor critical for initiation of translation. Phosphorylation of eIF2 reduces the rate of synthesis of the majority of proteins (140, 141) and thus decreases protein load and stress on the ER. The loss of PERK activity secondary to mutation abolishes this feedback, and the increased stress on the ER can initiate apoptosis. Cells with the greatest secretory load are those likely to be at greatest risk, and impaired EIF2AK3/PERK function will affect many tissues.
Two inactivating mutations have been identified that segregate with the disorder. 1103insT, identified in three of four probands, introduces a termination codon that truncates the EIF2AK3/PERK catalytic domain. A second missense mutation, G1832A, changes a highly conserved residue in the catalytic domain. Recent reports (142, 143, 144, 145, 146) have identified 17 novel mutations in the EIF2AK3 gene with variable expressivity, including three cases with developmental regression secondary to hepatic failure in consanguineous families of different racial and ethnic background (Table 3). Durocher et al. (144) reported on two, possibly related, patients of French Canadian origin with the same mutation, E331X, that exhibit different phenotypes possibly as a result of modifier genes and/or environmental and epigenetic interactions. Although none of the mutations that result in a truncated protein have been characterized, the absence of the catalytic and most of the regulatory domain suggests they will be inactive.
Two independently derived Eif2ak3 knockout mouse lines have been generated (134, 147) that display a phenotype similar to that seen in WRS patients, i.e., normal birth weight, slow growth, hyperglycemia, low insulin levels, and bone abnormalities including osteoporosis, deficient mineralization, and developmental abnormalities. During the first 2 wk of life, these mice are euglycemic with normal pancreatic histology. By 4–6 wk, the animals become hyperglycemic secondary to β-cell death. There is a parallel increase in -cells and a severe reduction in digestive enzymes; massive apoptosis was observed histologically in the pancreas. Electron microscopy of endocrine and exocrine tissue revealed abnormalities in the ER lumen consistent with the accumulation of misfolded proteins (134, 147). Zhang et al. (148) generated a series of tissue- and cell-specific PERK transgenes and showed that PERK is required during embryogenesis to maintain postnatal β-cell function and glucose homeostasis but is not required to maintain normoglycemia in adult mice.
Scheuner et al. (149) generated another model in which the PERK phosphorylation site on eIF2, Ser51, was replaced with an alanine. This model emphasized the importance of PERK in the regulation of glucose homeostasis via the liver; the animals died of hypoglycemia within 18 h due to impaired gluconeogenesis. Animals could be rescued by giving glucose every 8 h but displayed severe growth retardation. Islet studies showed that β-cell insulin content and mass were reduced by approximately 50%.
|
X. Insulin
|
|---|
Støy et al. (150) have reported that mutations in the insulin (INS) gene and its precursors are a novel cause of PNDM. Ten recessive de novo mutations were identified in 16 cases, but this initial report stimulated additional screening (151, 152) that identified further variants (Table 4). The mutations are localized to amino acid residues that could potentially affect cleavage and/or folding of pre-proinsulin and proinsulin. Some of these properties are summarized in Table 4. An accumulation of improperly folded insulin precursors would induce prolonged ER stress, the UPR, and the initiation of β-cell apoptosis. Unlike the EIF2AK3/PERK mutations that affect multiple tissues, increased apoptosis and cell death due to increased UPR should be restricted to pancreatic β-cells.
The initial missense mutations were identified in a cohort of patients with NDM. Although there is heterogeneity in the age of presentation, most cases were diagnosed in the first 6 months of life, with three cases diagnosed between 6 and 12 months. The father of one patient was diagnosed at 30 yr of age with a mild form of the disease. Most of the infants presented with severe hyperglycemia (681 mg/dl, median plasma value) or ketoacidosis and low or undetectable values of C-peptide. Interestingly, although the birth weights were reduced, the median [2846 g (20th centile)] was above that found in cases of NDM resulting from other causes. The authors note that in three cases the mothers developed gestational diabetes, which may have increased infant birth weight.
It is worth pointing out that the C96Y mutation was previously identified in the Akita mouse (153) and shown to impair folding and processing because a disulfide bond between the two insulin chains is not formed and the protein is partially retained in the ER. Akita mice are born normoglycemic with normal-sized islets but become hyperglycemic with age secondary to the loss of β-cells. Overexpression of mutant C96Y insulin in MIN6 cells resulted in apoptosis (154). A second mouse, the Munich Ins(C95S) model (155), was identified in a mouse mutagenesis screen using N-ethyl-N-nitrosourea. This mutation, which will disrupt the A6-A11 disulfide bond, produces hyperglycemia by 1 month of age secondary to reduced insulin release. Electron microscopy of male mouse β-cells revealed a lack of insulin secretory granules, enlarged ER, and swollen mitochondria consistent with UPR. The available data are consistent with loss of β-cells during the fetal period secondary to activation of the UPR by an excess of abnormally folded mutant insulin; however, this is a recent discovery, and additional molecular mechanisms may be uncovered.
|
XI. Immunodysregulation, Polyendocrinopathy, and Enteropathy, X-Linked (IPEX), FOXP3 (OMIM 304790/304930)
|
|---|
Initially described by Powell et al. (156) and later by Bennett et al. (157), IPEX (immunodysregulation, polyendocrinopathy, and enteropathy, X-linked) includes a rare, heterogeneous group of disorders that are almost always fatal. The disease can affect multiple tissue types, is reported under various names, and includes congenital diabetes, colitis and diarrhea, hypothyroidism, and frequent infections among others. Chatila et al. (158) and Wildin et al. (159) showed that the FOXP3 gene encoding scurfin (160) was altered in IPEX patients and in the scurfy mouse, which has an IPEX-like phenotype (reviewed in Ref. 161). Scurfin is a member of the forkhead/winged helix domain family of DNA-binding proteins, both transcriptional activators and repressors, that function in the control of lineage commitment and developmental differentiation (reviewed in Ref. 162). The mutations identified in IPEX patients (158, 159, 163, 164) affect protein dimerization or alter the forkhead/winged helix domain, thus impairing scurfin-DNA interactions. Scurfin binds to and represses IL-2 promoter activity and has been suggested to be required for the development of CD4+/CD25+ regulatory T cells, a naturally occurring population of regulatory T cells, that can inhibit harmful immunopathological responses against foreign or self antigens (se
Top
Abstract
I. Introduction
