The Full Spectrum of Holoprosencephaly-Associated Mutations within the ZIC2 Gene in Humans Predicts Loss-of-Function as the Predominant Disease Mechanism

Abstract

Mutations of the ZIC2 transcription factor gene are among the most common heterozygous variations detected in holoprosencephaly (HPE) patients, a patient group who lack critical midline forebrain specification due to defective embryonic signaling during development. Recent studies indicate that complete deficiency of the related murine Zic2 transcription factor can also be a contributing factor to variable midline deficiencies, presenting during mid-gastrulation, that could explain similar forebrain anomalies in this model system. Here we collect and summarize all available mutations in the human ZIC2 gene detected in HPE patients (21 published and 62 novel). Our analysis corroborates this mechanism proposed in mice by predicting loss-of-function as the likely pathogenetic mechanism common to most, if not all, of these mutations in HPE.

Introduction

ZIC2 (MIM# 603073; HPE5 609637) accounts for at least 3% of detectable holoprosencephaly (HPE) associated genetic variation and is one of four genes, including SHH (Roessler et al., 1996; MIM# 600725; HPE3 142945), SIX3 (Wallis et al., 1999; MIM# 603714; HPE2 157170) and TGIF (Gripp et al., 2000; MIM# 602630; HPE4 142946), commonly screened in clinical laboratories as part of the molecular evaluation of new sporadic or familial cases (reviewed in Muenke and Beachy, 2001; Cohen, 2006; Dubourg et al., 2007). This transcription factor gene was originally implicated in causing this forebrain malformation disorder by mutational screening of candidate genes mapping within a minimal critical region on chromosome 13q32 determined to be causally responsible for HPE cases involving 13q- deletions (Brown et al, 1995; 1998). The original descriptions of the types of mutations observed noted a high prevalence of frameshifts or trunctions indicative of haploinsufficiency of this transcription factor as the consequence of either major genetic alterations of the coding region, or alternatively gross cytogenetic or recently described microdeletions of this gene eliminating normal gene function of one allele (Brown et al., 1998, 2001, 2002, 2005; Bendavid et al., 2005a, b; Paulussen et al., submitted). A second category of mutation prevalent among HPE cases involves an expansion of a poly-Alanine tract in the COOH terminus of the protein that can be shown to interfere with DNA binding and transcriptional activation and likely occurs by somatic recombination (Brown et al., 2001, 2005).

Recent studies of the developmental roles of the Zic2 ortholog in the mouse have elucidated several distinct requirements for this transcription factor during normal development: an early role in the development of the axial midline revealed by a Zic2 null allele (Warr et al., 2008) and a later role in dorsal telencephalic development manifesting as the anatomically distinct middle-hemispheric variant of HPE (Cheng et al., 2006; reviewed in Monuki 2007) provisionally considered characteristic of a hypomorphic allele. Here we tabulate the nature of the detected mutations seen in HPE patients in a classical structure/function analysis and relate these findings to the proposed mechanisms of forebrain pathogenesis suggested by these model systems.

Materials and Methods

Study population

At the NIH, we analyzed approximately 600 HPE patients (collectively comprising the entire spectrum of HPE brain malformations and prospectively collected over 17 years) for potential sequence variations in the ZIC2 gene under our NHGRI approved brain research protocol and newly established CLIA laboratory. In addition, we also studied 125 unrelated individual normal controls obtained as anonymous samples from the Coriell Institute for Medical Research that matched the predominant Northern European ethnicity of our HPE cases. In the cases extracted from literature reports the nature of the mutation was known to us only through these published sources (cited in Table 1). Similarly, in Rennes, 500 HPE patients were analyzed, prospectively collected over 12 years at the Laboratoire de Génétique Moléculaire (Rennes, France); we also include anonymous instances of mutations in the ZIC2 gene shared with us from prospective studies performed under CLIA standards by GeneDx (Gaithersburg, MD), or from investigators in Maastricht, the Netherlands as well as Göttingen, Germany.

Table 1. Summary of human ZIC2 mutations.

Mutation number	Base-pair alteration	Coding region alteration	Functional activity	Comments	Reference for report
1	-24C>T	N/A	N/A	5′ non-coding region (immediately preceding the ATG) of unknown importance	GeneDx, this report
2	c.21delG	p.Gln8SerfsX33	Predicted null	lacks homeodomain	Brown 2001 (#1)
3	c.81_86delGGCGGC insTCGGT	p.Ala28ArgfsX13	Predicted null	lacks homeodomain	NIH, this report CLIA; GeneDx confirmed
4	c.107A>C	p.Gln36Pro	170%	NH2-terminus	Dubourg 2004; Brown 2005
5	c.109G>A	p.Asp37Asn	unknown	NH2-terminus. Also c.1059C>T (p.H353H)	NIH, this report CLIA; GeneDx confirmed
6	c.129_184dup56	p.Leu62Argfs175	Predicted null	NH2-terminus truncation	NIH, this report
7	c.136C>T	p.Gln46X	Predicted null	NH2-terminus truncation	Dubourg 2004; and GeneDx confirmed
8	c.172G>T	p.Gly58X	Predicted null	NH2-terminus truncation	Dubourg 2004
9	c.177ins56	p.Phe60GlnfsX176	Predicted null	lacks homeodomain	Brown 1998 (#a)
10	c.191dupC	p.Ala66ArgfsX301	Predicted null	lacks homeodomain	U. Rennes, this report
11	c.217C>T	p.Gln73X	Predicted null	lacks homeodomain	U. Rennes, this report
12	c.217delC	p.Gln73Argfs145	Predicted null	lacks homeodomain	NIH, this report
13	c.367delA	p.Ser123AlafsX95	Predicted null	lacks homeodomain	U. Rennes, this report
14	c.382G>A	p.Asp128Asn	unknown	Mis-sense	NIH, this report
15	c.386_392delCGGCGCC	p.Ser129TrpfsX87	Predicted null	lacks homeodomain	NIH, this report
16	c.392_398del7	p.Gly133SerfsX83	Predicted null	lacks homeodomain	GeneDx, this report
17	c.454_455delinsTT	p.Asp152Phe	60%	Mis-sense NH2-terminus	Brown 2001 (#3); Dubourg 2004
18	c.479delC	p.Pro160ArgfsX58	Predicted null	lacks homeodomain	Maastricht
19	c.490G>T	p.Glu164X	Predicted null	lacks homeodomain	NIH, this report; confirmed GeneDx
20	c.557_572dup16	p.Glu192GlyfsX180	Predicted null	lacks homeodomain	GeneDx, this report
21	c.577delC	p.Gln193AsnfsX25	Predicted null	lacks homeodomain	GeneDx, this report
22	c.582C>A	p.Tyr194X	Predicted null	lacks homeodomain	U. Rennes, this report
23	c.612delC	p.Tyr205ThrfsX13	Predicted null	lacks homeodomain	NIH, this report CLIA
24	c.659delA	p.Asn220ThrfsX4	Predicted null	lacks homeodomain	U. Göttingen, this report
25	c.665_676dup12	p.Gly222_Met225dup	unknown	unknown	U. Rennes, this report
26	c.710_718dupCACCACCAC	H12 variant	unknown	Observed length variation of 8 up to 12 H's	Brown 2001; and NIH, this report CLIA
27	c.716_718del	H8 variant	unknown	Observed length variation in controls	Dubourg 2004
28	c.716_718dupACC	H10 variant	unknown	Observed length variation in controls	Brown 2001; Orioli 2001; Dubourg 2004 GeneDx, and NIH this report CLIA
29	c.748C>T	p.Gln250X	Predicted null	Truncation in ZF-NC; lacks homeodomain	NIH, this report
30	c.779G>A	p.Trp260X	Predicted null	Truncation in ZF#1	U. Göttingen, this report
31	c.793C>T	p.Gln265X	Predicted null	lacks homeodomain	GeneDx, this report
32	c.808_809ins17	p.K270ThrfsX2	Predicted null	Truncation in ZF#1	Dubourg 2004
33	c.815G>A	p.Ser272Asn	Predicted null	Mis-sense in ZF#1	U. Rennes, this report
34	c.825_826delAA	p.Lys275AsnfsX91	Predicted null	Fs in ZF#1; lacks homeodomain	GeneDx, this report
35	c.829_830dupTT	p.Thr279AlafsX7	Predicted null	FS in ZF#1; lacks homeodomain	GeneDx, this report
36	c.862_863delTC	p.Ser288GlyfsX78	Predicted null	Fs in ZF#1; lacks homeodomain	Orioli 2001
37	c.857A>T	p.His286Leu	Predicted null	Invariant C2H2 structure ZF#1	GeneDx, this report
38	c.858C>G also c.1059C>T	p.His286Gln also p.His353His	Predicted null also Common variant	Invariant C2H2 structure ZF#1;	NIH, this report; GeneDx confirmed
39	c.856C>T	p.His286Tyr	Predicted null	Invariant C2H2 structure ZF#1	U. Rennes, this report
40	c.871C>T	p.His291Tyr	Predicted null	Invariant C2H2 structure ZF#1	U. Rennes, this report
41	c.910T>A	p.Trp304Arg	Predicted null	ZF#2	GeneDx, this report
42	c.912G>A	p.Trp304X	Predicted null	Truncation in ZF#2	GeneDx and U. Rennes, this report
43	c.928G>T	p.Glu310X	Predicted null	ZF#2 truncation	NIH and U. Rennes, this report
44	c.932delG	p.Gly311AlafsX102	Predicted null	ZF#2 frameshift	Brown 2001 (#4)
45	c.941T>G	p.Phe314Cys	Predicted null	Conserved ZF#2	Rennes, this report
46	c.973C>A	p.Arg325Ser	Predicted null	Conserved ZF#2	NIH, this report; GeneDx confirmed
47	c.974G>T	p.Arg325Leu	Predicted null	Conserved ZF#2	U. Rennes, this report
48	c.979C>T	p.His327Tyr	Predicted null	ZF#2 invariant C2H2 motif	NIH, this report
49	c.1004G>T	p.Cys335Phe	Predicted null	ZF#3 invariant C2H2 motif	NIH, this report
50	c.1025_1026delAA	p.Lys342SerfsX24	Predicted null	ZF#3 frameshift	Dubourg 2004
51	c.1031_1032delTC	p.Phe344CysfsX22	Predicted null	ZF#3 frameshift	Brown 2001 (#5)
52	c.1040_1046del	p.Glu348SerfsX63	Predicted null	ZF#3 frameshift	Brown 1998 (#d) and NIH, this report
53	c.1051A>T	p.Lys351X	Predicted null	ZF#3 truncation	NIH, this report
54	c.1052_1053insAAGGTT CACACAGAACCTCAA	p.Lys351_Ile352ins7	Predicted null	ZF#3 deletion/insertion	U. Rennes, this report
55	c.1075+2T>A (IVS1+2T>A)	Gly359fsX62	Predicted null	Eliminates ZF#4 and 5	U. Rennes, this report
56	c.1076-1G>A (IVS1-1G>A)	alternative splicing	Predicted null	ZF#1-3 intact; likely eliminates ZF#4-5	NIH, this report; GeneDx confirmed
	c.1076-1G>A (IVS1-1G>A)	alternative splicing	Predicted null	ZF#1-3 intact; likely eliminates ZF#4-5	NIH, this report
	c.1076-1G>A (IVS1-1G>A)	alternative splicing	Predicted null	ZF#1-3 intact; likely eliminates ZF#4-5	NIH, this report
57	c.1239+1G>C (IVS2+1G>C)	Possible inclusion of intron 2 codons or alternative splicing	Predicted null	Disrupts ZF#5	GeneDx, this report
58	c.1240-2A>G (IVS2-2A>G)	alternative splicing	Predicted null	ZF#1-3 intact; likely eliminates ZF#4-5	NIH, this report
59	c.1090C>T	p.Gln364X	Predicted null	Eliminates all of ZF#4-5	Maastricht
60	c.1091_1092delAG	p.Gln364LeufsX2	Predicted null	Eliminates all of ZF#4-5	Dubourg 2004
61	c.1095_1096delTG	p.Cys365X	Predicted null	Eliminates all of ZF#4-5	NIH, this report (case #1)
	c.1095_1096delTG	p.Cys365X	Predicted null	Eliminates all of ZF#4-5	NIH, this report (apparently unrelated case #2)
62	c.1097_1098delAG	p.Glu366Valfs2	Predicted null	ZF#4 truncation	NIH, this report; GeneDx confirmed; Brown 2001 (#7)
63	c.1119_1120delCT	p.Phe374ArgfsX17	Predicted null	ZF#4 frameshift	NIH, this report
64	c.1118G>C	p.Arg373Pro	Predicted null	ZF#4 conserved	GeneDx, this report
65	c.1206C>G	p.Tyr402X	Predicated null	ZF #5 truncation	NIH, this report CLIA
66	c.1204T>A	p.Tyr402Asn	Predicted null	ZF#5 core sequence	NIH, this report; GeneDx confirmed
67	c. 1208C>A	p.Thr403Lys	Predicted null	ZF#5 core sequence	NIH, this report; GeneDx confirmed
68	c.1211A>G	p.His404Arg	Predicted null	ZF#5 core sequence	GeneDx, this report
69	c.1225C>T	p.Arg409Trp	Predicted null	ZF#5 core sequence	NIH, this report
70	c.1245T>G	p.His415Gln	Predicted null	Invariant C2H2 structure ZF#5	U. Rennes, this report
71	c.1277delC	p.Pro426ArgfsX129	Predicted null	Intact ZF#1-5	NIH, this report; GeneDx confirmed
72	c.1313dupC	p.Leu440AlafsX90	Predicted null	Intact ZF#1-5	Brown 1998 (#b)
73	c.1323dupG	p.Ser442Valfs88	2%	Intact ZF#1-5	NIH, this report; Brown 2001 (#8) and 2005
74	c.1330_1365del	p.444_455del In frame deletion	60%	Intact ZF#1-5 and intact polyalanine tract	Brown 2001 (#9) and 2005
75	c.1377_1406dup30	p.Ala461_470dup In frame duplication	5% with reduced DNA binding	25 alanine variant	Brown 1998 (#c) and 2001 (#10-16) and 2005; NIH, this report CLIA; Maastricht, this report (one case maternal and one de novo).
76	c.1392_1406del	In frame deletion	Predicted null	10 alanine variant	NIH, this report; GeneDx confirmed
77	c.1377_1406del	In frame deletion	Predicted null	5 alanine variant	U. Rennes, this report
78	c.[1420-1427dup; 1428-1433delinsCG]	p.Gly477CysfsX54	Predicted null	Novel COOH terminus	U. Rennes, this report
79	c.1445_1461del17	p.Ser482ArgfsX42	Predicted null	Novel COOH-terminus	NIH, this report; GeneDx confirmed
80	c.1452_1456delCGCGG	p.Ala485ArgfsX43	Predicted null	Novel COOH terminus	U. Rennes, this report
81	c.1455_1461delinsCG	p.Gly487LeufsX41	Predicted null	Novel COOH terminus	U. Rennes, this report
82	c.1508_1520delGCGGCGG GGGCGG	p.Gly503AlafsX48	Predicted null	Novel COOH terminus	Maastricht
83	c.1559delA	p.His520ProfsX35	Predicted null	Novel COOH terminus	U. Rennes, this report

¹Sequence variant numbering is based on ZIC2 cDNA sequence (GenBank NM_007129.2). Nucleotide numbering uses the A of the ATG translation initiation site as nucleotide +1 and the 5′ untranslated variant is numbered on this basis, rather than by its genomic position.

Mutation screening, PCR amplification and DNA sequencing

A strategy for screening the ZIC2 gene has previously been described (Brown et al., 1998; Dubourg et al., 2004). Descriptions of mutations (Table 1) are based on the NM_007129.2 reference sequence utilizing current nomenclature recommendations www.hgvs.org/mutnomen. Nucleotide numbering reflects cDNA numbering with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence. The initiation codon is codon 1. Our current approach for CLIA testing for ZIC2 differs slightly from previous reports and is available upon request. Amplification of human genomic DNA is performed in a 30 μl reaction volume, using 60-100 ng DNA template, 50 μM each of deoxynucleotide triphosphate, 0.25 μM of each primer, 3 μl of 10X PCR Amplification buffer (Invitrogen), 1.5 μl 10X Enhancer buffer (Invitrogen) and 0.3μl Taq polymerase. All reactions were performed using a PTC-255 thermocycler (MJ Research, MA). Typical PCR cycling parameters were 95°C for 4 minutes followed by 30 cycles at 95°C, annealing at 62°C, extension at 72°C for 1 minute, and a final extension step of 72°C for 5 minutes. One half of the PCR product was used for denaturing high-pressure liquid chromatography (dHPLC) analysis (Trangenomics WAVE™) and the remainder was retained for direct DNA sequencing. Amplicons displaying heterozygous profiles were purified using a high pure PCR purification kit (Roche, IN) and bi-directionally sequenced using the BigDye™ version 3.1 terminator cycle sequencing kit according to the manufacturer's protocol (Applied Biosystems, CA) on an ABI 3100 automated sequencer.

Results

In order to derive insight into the important functions and domains of the ZIC2 protein inferred from our mutational analysis, we have compiled all 83 of the known variants detected exclusively among HPE cases and not in controls into Table 1. Several important findings emerge from this compilation (broken down for convenience by mutation type in Figure 1A-D). Regardless of mutation type these categories of variation have either been directly shown to reduce biological activity (Brown et al., 2005) or would be predicted to lose activity by analysis of the effects on key structural motifs of virtually translated protein products.

Figure 1. A schematic representation of all of the mutations affecting the ZIC2 coding region sorted by their position and category.

A: Mis-sense variations (red triangles, 25% overall) cluster within the zinc-finger motifs (17/21 = 81%). The ZIC2 transcription factor is illustrated as a line of aminoacids (from left to right, NH₂ to COOH) with key motifs (Aruga et al., 2004, 2006) depicted by rectangular boxes: ZOC homology block (blue), His-9 variable region (grey, range 8-12), homology block ZF-NC (green) preceding the first of five zinc-fingers (increasing shades of blue to indicate that each finger has a unique structure) and Ala-15 variable region (orange, range 5-25). B: 32/83 = 38.5% of the mutations (black diamond) would predict a significant alteration of the reading frame and variable truncations primarily affecting DNA binding. C: Note that all non-sense truncations (black arrows, 17% overall) occur prior to zinc-fingers 4 and 5 primarily implicated in DNA binding by Krüppel/GLI-related transcription factors (Kinzler et al., 1990; Pavletich and Pabo, 1993). D: The typical observed range of poly-Alanine and poly-Histidine segments are shown (Brown et al., 1998, 2001, 2002; Dubourg et al., 2004). All four invariant splice sites including both donor (arrow, no fill) and both acceptor (yellow arrow) sites are subject to mutation. Note that the failure to initiate a splice from IVS2 donor site nevertheless maintains the reading frame predicting a protein with 71 additional aminoacids and a distortion of the zinc-finger 5 (inverted hatched triangle, see also Figure 2). Duplications that retain or resume frame are shown as diamonds with no fill while the inverted open triangle illustrates an in frame deletion. Complex insertion/deletion sequence variants causing a shift of the reading frame are the black inverted triangles. The non-coding 5′ UTR variation is not indicated by any symbol.

First, except for the poly-Alanine tract expansion/contraction cases, that continue to represent a prevalent mutation type (Fig. 1D, see below), we observe that most mutations are detected in only a single family (i.e. family-specific alterations) and at least 50% can be demonstrated to be de novo within the HPE proband's DNA (Dubourg, this study). There is a strikingly similar molecular pathology elucidated for the zinc-finger mutations in the X-linked ZIC3 gene and their proven effects on diminished DNA binding and reporter gene activation (Ware et al., 2004; Zhu et al., 2008). In this respect, heterozygous ZIC2 mutations differ from the hemizygocity of the X-linked ZIC3 gene or the homozygous hypomorph or null models of HPE in mice (Nagai et al., 2000; Warr et al., 2008).

Second, our findings clearly illustrate that there is a common feature of predicted loss-of-function attributable to virtually all of the HPE-specific variants, especially evident in our continued detection of a high frequency of frameshifts and truncations (>55%) also noted in the original studies (Brown et al. 1998, 2001; Dubourg 2004). None of the truncation mutations when hypothetically translated are consistent with a complete set of zinc-fingers; this is also true of the majority of the predicted frameshift variants (compare Fig. 1B to C). However, a significant number of frameshifts (Fig. 1B) or insertion/deletions (Fig. 1D) occur in the extreme COOH terminus of the protein indicating that simple variations affecting the zinc-finger domain are not the only mechanism affecting ZIC2 function and confirming previous interpretations (Brown et al. 2005).

Thirdly, almost all of the mis-sense mutations occur in residues known to be either extensively conserved among the zinc-finger regions of all Zic-related genes (Fig. 1A; see also Aruga et al., 2006) or intrinsic to their ability to fold and interact with DNA. Specifically, each of the five zinc fingers is defined by a Cystidine2-Histidine2 (C2H2 motif) comprising four bonds that coordinate with a zinc-atom into a unique three-dimensional configuration (Redemann et al., 1988). We observe that 5 of these 20 key residues are either directly mutated through codon changes or displaced by insertions (Fig. 2). In addition, we demonstrate that by compiling additional cases there are now numerous examples of codons or genetic elements important to the function of the protein that are targets of independent mutational damaging events in different families. For example, p.286His is not only an invariant residue within Zinc-finger #1 (ZF #1 in Table 1, Fig. 1A and Fig. 2) but it is also found mutated to different codons in three different HPE subjects, strengthening the interpretation that the zinc-fingers and their integrity are key to the normal function of the ZIC2 protein.

Figure 2. Detailed structure of the zinc-finger region.

The aminoacid sequence (single aminoacid code) of the region from approximately His-9 to Ala-15 that contains the individually numbered Zinc-finger (ZF domains) and the most mutations with individual motifs colored as in Figure 1. The C2H2 components of each zinc-finger are bold and underlined and residues invariably present in the ZIC-family alignment are identified with an asterix. Note the non-random clustering of the mis-sense mutations (red) within these conserved aminoacid residues and particularly those that define the zinc-binding motif of the C2H2 zinc fingers (Redemann et al., 1988). Similarly, the addition of 7 novel aminoacids within the ZF #3 likely results in distorted alignment (inverted black triangle). The c.1239+1G>C (or IVS2+1G>C) allows for the incorporation in toto of the IVS2 into the reading frame of the transcript and adding an additional 71 aminoacids. This insertion also removes the ValHis sequence from its normal context within the C2H2 motif of ZF #5 and distorting its structure; this same key C2H2 Histidine is a target of mutation in another HPE individual.

Of the four mis-sense alterations detected outside of the zinc-finger domain there is evidence for apparent clustering. The c.107A>C Gln36Pro variant has previously been shown to retain 170% activity in cell-based luciferase assays (Brown et al., 2005). Interestingly, we now report that the adjacent residue c.109G>A Asp37Asn is altered in a novel HPE proband potentially supporting a role for this region of the NH2-terminus in ZIC2 function. Similarly, c.454_455delGAinsTT Asp152Phe has been shown to deleteriously affect function, while the nearby mutation c.382G>A Asp128Asn is presently untested. It is also worth noting that although the ZOC (Zic/Opa [Odd-paired] Conserved) domain is highly conserved among ZIC proteins (Aruga et al., 2006) it has not been implicated in our mutational analysis thus far.

Defects in splicing are predicted to follow from independent mutations in both donor and both acceptor sites (Fig. 1D); the common consequence of these changes is to deprive the transcript from the ability to encode an intact set of zinc-fingers. Note that three apparently unrelated families have the same alteration in the splice acceptor site of the first intervening sequence (c.1076-1G>A or IVS1-1G>A). Even in-frame insertions tend to occur in key structural elements, such as ZF #3 and ZF #5 predicting altered DNA binding or its absence as the most common component of the molecular pathology.

One of the principal conclusions of the most rigorous functional analysis to date (Brown et al., 2005) was that beyond the obvious importance of the zinc-finger domain (see above) there remained a significant fraction of all cases demonstrating length variations of either the poly-Histidine tract or especially the poly-Alanine tract (Brown et al. 2001, 2002, 2005). While the former His length variation is currently postulated to be a simple population variant (stretches of 8-12 residues have been observed in both patients and controls) there is overwhelming evidence that an expansion of this poly-Alanine tract to 25 affects both DNA binding and reporter gene activation. We now show that length reductions to 5 or 10 aminoacids are also detected. While these two length reduction variations will require additional study, it is intriguing that the spacing between the zinc-fingers and the poly-Alanine tract is proven to be important for ZIC2 function, since the in-frame deletion of 12 aminoacids, variant #74 c.1330_1365del36, between the end of ZF #5 and the beginning of the poly-Alanine tract clearly reduces biological activity (Brown et al., 1998, 2005).

Consistent with the fact that ZIC2 was initially described as a key gene within a minimal critical region of common HPE-associated cytogenetic re-arrangements, it follows that copy-number variations of the gene should be readily detectable and produce similar biological effects as a more typical loss-of-function allele through crucial coding region alterations. This is indeed the case, since the U. Rennes has identified four independent cases of ZIC2 loss through microdeletion (Bendavid and David, in preparation), the Maastricht team has detected one ZIC2 microdeletion (Paulussen et al., submitted), and our NIH studies have also identified one additional case (∼1.25% of tested probands, Lacbawan and Muenke, in preparation). As new knowledge is gained relating to non-coding regulatory elements essential for ZIC2 expression there are likely to be additional mutational targets identified.

Discussion

Holoprosencephaly is the most common structural malformation of the developing forebrain in humans and its causation is multifarious (including genetic and environmental components) and complex (Muenke and Beachy, 2001; Roessler et al., 2001; Cohen, 2006; Dubourg et al., 2007; Monuki et al., 2007). Its genetic heterogeneity is supported by the fact that nearly 75% of cases are wild-type in their DNA sequence of the coding regions of the nine currently known HPE genes in patients with normal chromosomes. Its characteristic variable expressivity adds additional support for a model of combinatorial interactions among multiple factors (Ming and Muenke, 2002). Although much more needs to be elucidated regarding the potential roles of additional genes, a concerted focus is also required to integrate the potential interactions among the presently known HPE genes and environmental stressors, such as ethanol exposure and maternal diabetes.

The vertebrate system that most closely mirrors genetic susceptibilities for HPE in humans is that of the mouse (reviewed in Krauss et al., 2007). The recent finding that the Kumba mutation (Elms et al., 2003) in murine Zic2 is likely a null suggests that complete absence of this transcription factor causes a mid-gastrulation failure of the axial midline development that can manifest itself as HPE to various degrees (Warr et al., 2008). By contrast, a homozygous hypomorphic Zic2 mutant line (estimated to have ∼20% activity) develops normally through gastrulation and instead displays dorsal forebrain malformations at later stages (Nagai et al., 2001). Mice heterozygous for Zic2 null alleles (50% activity) are phenotypically normal. Therefore, the model that emerges suggests that the cyclopia that follows the loss of both alleles in mice is a consequence of a variable degree of midline arrest, with secondary effects at a later stage on forebrain patterning activities such as Sonic hedgehog. Interestingly, these same investigators could not demonstrate any synergy in the genetic interactions between Zic2 and Shh compound mutants, suggesting that as reflected by these studies that these factors act in parallel, but not directly linked molecular pathways. This is a potentially important observation since the zinc-finger domains of the Zic and Gli factors bind to the identical target sequences (Kinzler et al. 1990; Pavletich and Pabo 1993; Mizugishi et al., 2001) and these facts had suggested that Zic2 was a likely downstream effector of hedgehog signals (Roessler et al., 2001). It now appears likely that the mid-gastrulation functions of Zic2 preceed those of Shh in the mouse (Warr et al., 2008) and that these factors can play important roles at more than one stage of embryonic development.

Our reprise of ZIC2 mutations brings this necessity for integrative gene/environment interactions directly into focus. Heterozygous mutations are the norm in humans. No instances of simultaneous loss of both copies of human ZIC2 have been described to date, despite our clear understanding that this would be a pre-requisite for producing such a phenotype in the mouse model (Warr et al., 2008). Furthermore, no clear pattern of co-morbid genetic changes is apparent among the patients summarized in this report. While ZIC2 screening has been performed for nearly a decade the number of cases of simultaneous genetic lesions involving ZIC2 with other known or candidate HPE genes are anecdotal and infrequent (Ming and Muenke, 2002). Hence, our certitude with respect to the importance and implications of a genetic variant in the ZIC2 gene (as one component of a multiple-hit process) become essential when the second key component of pathogenesis remains obscure. At the present time we can say that the proposed second-hit is not commonly in the SHH gene, or its functionally related pathway factors (Roessler et al., 2005), and may well be one of the poorly understood factors dependent on either direct ZIC2 function or those its downstream targets.

Our structural analysis of these mutations indicates that DNA binding, influenced by the structural integrity of the zinc-fingers and modulated by the COOH-terminus, are the most critical domains detected as mutations among our combined HPE cohorts. For the purposes of genetic counseling, it is important to convey that the expectation of detecting an abnormality in ZIC2 in a new HPE case is high and most of these changes are risk factors for HPE in these families. Undetected sequence variations at other genomic loci, as well as environmental factors and stochastic effects also factor into the extremely variable expression typical of the complex genetic underpinnings of this disorder.