In this issue, Raca et al
1
present two cases of childhood apraxia of speech (CAS) arising from microdeletions
of chromosome 16p11.2. They propose that comprehensive phenotypic profiling may assist
in the delineation and classification of such cases. To complement this study, we
would like to report on a third, unrelated, child who presents with CAS and a chromosome
16p11.2 heterozygous deletion. We use genetic data from this child and his family
to illustrate how comprehensive genetic profiling may also assist in the characterisation
of 16p11.2 microdeletion syndrome.
A number of chromosome 16p11.2 aberrations have been reported in the recent literature,
including gross rearrangements and sub-microscopic (<1 Mb) deletions and duplications
with incomplete penetrance and variable expressivity and in a heterozygous form.
2
In general, microdeletions appear to be more penetrant than their respective duplications.
3
The ‘typical' 16p11.2 deletion encompasses 539 kb (from chromosome position 29.5–30.1 Mb,
GRCh37/hg19) and 24 genes, but a smaller adjacent distal or ‘atypical' deletion (between
chromosome positions 28.7 and 28.95 Mb, GRCh37/hg19) has also been reported, as have
novel anomalies outside of these specified regions.
2, 4, 5
Individuals have been described with deletions spanning both these regions
4, 6
and families have been observed to carry both rearrangement types.
7
It has been suggested that proximal rearrangements may be associated with developmental
impairments and distal variations correlated with altered body mass index,
2
although developmental delays and speech and language impairments appear to be a common
feature of individuals with various anomalies across this chromosome band.
8, 9, 10
Screens of clinical cohorts indicate that chromosome 16p11.2 rearrangements are observed
at a frequency of 0.3–0.7% in patients with various developmental impairments including
autistic disorder (eg, Weiss et al
11
), developmental delay (eg, Shinawi et al
3
), epilepsy
12
and schizophrenia (eg, McCarthy et al
13
). Deletions are also observed in apparently healthy individuals, at a similar frequency
to clinical cohorts,
5
and while the majority of cases appear to be de novo, inherited imbalances of this
region are not uncommon.
Recently there has been a drive to define a core clinical phenotype of the 16p11.2
microdeletion syndrome. In a retrospective screen of 9773 individuals referred for
microarray testing, Rosenfeld et al
8
found that 77 carried chromosome 16p11.2 anomalies (45 deletions, 32 duplications,
0.78%). A detailed review of 18 patients found that the most consistent clinical manifestations
in these individuals were intellectual impairment and speech and language delays.
8
These findings were supported by a similar study that included 7400 patients who had
undergone array comparative genomic hybridisation (array-CGH) testing in a clinical
context, 45 of whom carried 16p11.2 anomalies (27 deletions, 18 duplications, 0.6%).
3
Phenotypic characterisation of 27 individuals also found that all had speech and language
delays and cognitive impairment.
3
Other predominant features of 16p11.2 syndrome include dysmorphism, macrocephaly and
autistic disorders.
3, 4, 8, 14
However, all of these features have been disputed and it is likely that ascertainment
bias will affect the conclusions of many studies, particularly those that focus upon
single cases. Thus, the characterisation of the relationships between genetic aberration
and clinical presentation is ongoing and will require further, more refined, studies
with detailed investigations of this chromosome region and consistent phenotyping
of affected individuals.
The child described here was originally assessed for the presence of FOXP2 (OMIM #605317)
mutations and rearrangements, as part of an earlier screening project,
15
as disruptions of this gene have been implicated in rare cases of severe speech and
language disorder.
16
Although no FOXP2 mutations were identified in the child, we discovered a deletion
of chromosome 16p11.2. The child is a second-born male child of unrelated and healthy
parents of European (Caucasian) descent. His early development was normal until the
age of 1.5 years, when there was evidence of a social withdrawal. He was referred
following concerns regarding his speech and language development and received a diagnosis
of developmental verbal dyspraxia, also known as CAS, and pervasive developmental
disorder not otherwise specified (PDD-NOS) according to DSM-IV.
17
On assessment with the Vineland Adaptive Behaviour Scales,
18
he showed a remarkable impairment in language and communication. At age 14, he scored
at an age equivalent of 3 years and 3 months in the communication domain, with a major
impairment in the expressive subdomain. He also displayed stereotypic movements and
behavioural disturbances with self-aggressive episodes. He did not have hearing or
ocular problems and had normal height and weight. The patient showed macrocephaly
and peculiar facial features, such as heavy eyebrows with mild synophris, down-slanting
palpebral fissures, hypertelorism, short philtrum, carp shaped mouth and full lips.
He had brachydactyly and single crease bilaterally. His parents were clinically normal
with no family history of speech delay, autistic disorders or mental retardation.
Peripheral blood samples were collected from the proband and his parents, and DNA
extracted according to standard procedures.
19
To identify genomic imbalances, DNA samples were hybridised to Agilent 244K and Agilent
4 × 44K arrays (Agilent Technologies Inc., Santa Clara, CA, USA) for the proband and
his parents, respectively. The array-CGH was completed as part of an assessment of
36 children with specific language impairment, PDD-NOS and autism spectrum disorders
(ASD). Image data were extracted using Agilent Feature Extraction software version
8.5 (Agilent Technologies Inc.) and analysed using Agilent CGH Analytics software
version 3.4 (z-score method setting) (Agilent Technologies Inc.). The reference genomic
DNA samples used throughout the study were from the same consented individuals, one
male and one female. We estimate that the mean resolution of the Agilent 244 K arrays
is ∼40 kb.
We identified a de novo chromosome 16p11.2 deletion in the proband's sample (see Figure
1). The minimal region affected by this deletion spans from chromosome position 29 652 999
to 30 199 351 (GRCh37/hg19) and encompasses 28 Refseq genes, thus coinciding with
the ‘proximal' type reported in the literature.
2
The presence of speech and language abnormalities, macrocephaly and PDD in this child
coincides with previously reported core phenotypes of chromosome 16p deletions.
1, 3, 4, 8, 14
The CAS diagnosis of this case provides further support to the findings of Raca et
al
1
and their theory as to the importance of proximal chromosome 16p11.2 abnormalities
in CAS. It would thus be of interest to fully assess the CAS in this child using the
Madison Speech Assessment Protocol and other relevant speech batteries suggested by
Raca et al.
20, 21
Using the array-CGH data, we catalogued all observed imbalances that spanned four
or more consecutive oligonucleotide probes with values outside the log10 Cy-dye threshold
ratios for the proband. We excluded any region that had been observed repeatedly either
in control data deposited in the Database of Genomic Variants (DGV)
22
or within our own sample sets, and small imbalances that mapped to regions without
noted reference genes or mRNAs. This approach allowed us to identify an additional
novel duplication of chromosome 6q22.31, which occurred both in the proband and in
his clinically normal mother. This duplication has a minimal region from chromosome
position 123 527 545 to 124 311 813 (GRCh37/hg19) and does not overlap significantly
with any known CNVs in the DGV. The duplication covers two genes: the entire coding
region of TRDN (OMIM #603283) (triadin), a ryanodine-sensitive calcium channel expressed
in cardiac and skeletal muscle,
23
and the first exon of NKAIN2 (OMIM #609758) (sodium/potassium-transporting ATPase
subunit beta-1-interacting protein 2 isoform 2), a transmembrane protein. Truncation
of NKAIN2 has been described in patients with developmental delay
24
and complex neurological impairment.
25
All other events found in the patient overlapped with those reported in the DGV. A
full list of events can be found in Supplementary Table 1.
Thus we hypothesise that the inherited chromosome 6q22.31 duplication may compound
the presence of the de novo 16p11.2 deletion, leading to the observed clinical phenotype
in this patient. As most researchers focus solely on the chromosome 16p abnormality,
or choose to exclude inherited CNVs, most cases of ‘dual CNV disorder' such as this
will have been missed in the literature. In an attempt to identify similar cases,
we performed a PubMed search for ‘16p11.2' that matched 130 articles (August 1990–February
2012). Fifty of these manuscripts described the characterisation of 16p11.2 anomalies,
of which only nine explicitly reported information regarding concurrent CNVs.
7, 10, 12, 14, 26, 27, 28, 29, 30
When limiting our search to cases with typical proximal (29.5–30.1 Mb) 16p11.2 anomalies,
we were unable to identify any 16p11.2 cases reported to co-occur with NKAIN2 CNVs.
Across the nine studies available, the only regions that were consistently reported
across multiple studies as secondary CNVs in 16p11.2 patients were 15q11.2 (Prader–Willi
syndrome region, found in 2 of 31 16p11.2 patients studied in Bachmann-Gagescu et
al
10
and 1 of 427 autistic individuals studied in Marshall et al
28
), 15q13.2 (found in 2 of 138 16p11.2 duplication carriers studied in Jacquemont et
al
27
and 1 of 427 autistic individuals studied in Marshall et al
28
) and 22q11.2 (DiGeorge syndrome critical region, found in 1 of 31 16p11.2 patients
studied inBachmann-Gagescu et al
10
and 1 of 36 autistic individuals studied in Davis et al
30
). Interestingly, these recurrent secondary CNVs align with regions known to be involved
in autism and developmental delays.
31, 32, 33
When we widened our search to include novel chromosome 16p11.2 CNVs outside of the
typical region (28.0–31.4 Mb), we did find one study that documented cases with co-occurring
chromosome 16p11.2 and 6q22.31 abnormalities.
29
This study, by Sanders et al,
29
investigated 1124 individuals with autism and their unaffected family members (2248
parents and 872 sibs) and identified several recurrent copy number events associated
with autism, including rearrangements of 16p11.2, both within and outside the ‘typical'
region. As part of their Supplementary data, the authors published full lists of all
high-confidence CNVs found in samples passing quality control. These included lists
of rare CNVs that did not overlap more than 50% with a CNV present at >1% frequency
in the DGV
29
(Supplementary Table S8). Using these Supplementary data, we were able to identify
seven autistic probands who carried concurrent 16p11.2 (five duplications and two
deletions) and NKAIN2 (six duplications and one deletion) anomalies, all inherited
from healthy parents (Table 1). However, all of these events were small (<50 kb) and
none overlapped with those observed in our patient. No 16p11.2 events were found to
be concurrent with TRDN CNVs. Furthermore, the chromosome 16p11.2 anomalies identified
by Sanders et al
29
in these concurrent cases were all outside of the typical region and, on review of
the DGV, we noted that those involving the SULT1A1 (OMIM #171150) gene overlapped
significantly with regions of common variation (Table 1). Similarly, the NKAIN2 anomalies
were intronic to the RefSeq NKAIN2 consensus sequence, or overlapped with common CNVs
in the DGV (Table 1). However, it is worth noting that we observed a mRNA, BC035062,
that is annotated to include an exon contained within the common ‘intronic' duplication,
suggesting these events may affect a splice variant.
The patient we describe in this letter carries the typical 16p11.2 loss co-occurring
with a further 6q22.31 duplication, both of which are distinct from those described
above. The latter does not overlap significantly with noted DGV variants. However,
further mining of the Sanders et al
29
Supplementary data identified three healthy individuals carrying apparently identical
6q22.31 duplications (a father and son and another father) (Supplementary Table S8).
Thus the 6q22.31 duplication in our case may represent a very rare CNV with little
independent effect, but we cannot rule out a modifying role in combination with the
16p11.2 loss, particularly in view of the gene content. The 6q22.31 duplicated region
in our patient encompasses all of the coding regions of TRDN and the first exon of
NKAIN2. TRDN codes for a muscle-specific protein, deletion of which leads to cardiac
arrhythmia.
34
Although primarily expressed in cardiac tissue, this gene is also expressed in skeletal
muscle, where it is involved in the regulation of resting calcium levels.
35
NKAIN2 is a transmembrane protein with four homologues (NKAIN1–4), all of which are
highly conserved and have brain-specific expression.
36
Interestingly, it is the only gene in common with the smaller 6q22.31 CNV regions
described above. The cellular functions of the NKAIN proteins are unknown but they
have been shown to localise and interact with the plasma membrane protein ATP1B1.
Drosophila dNKAIN mutants show decreased co-ordination and temperature-sensitive paralysis.
36
Microdeletions in NKAIN2 have previously been reported as rare events contributing
to the risk of schizophrenia
37
and Attention Deficit/ Hyperactivity Disorder (ADHD),
38
and variants within this gene have been associated with neuroticism.
39
Taking all of this information into consideration, we believe that the clinical presentation
of chromosome 16p11.2 deletion cases may be modulated by the presence of additional
genomic imbalances, such as the inherited duplication of chromosome 6q22.31 observed
in our case. Researchers of developmental disorders have proposed a dual CNV model
at other loci,
33, 40, 41, 42
as well as compound heterozygotes with a CNV-mediated deletion of one allele and non-synonymous
mutation of the other (mixed genomic disorders).
43, 44
The genetic background, of course, extends beyond CNVs and, as genetic technologies
advance, we predict that a whole-genome view will allow the elucidation of many combinatorial
factors. For example, a recent study extended the dual CNV model to incorporate rare
point mutations across common functional pathways, where an ASD proband was identified
with both a de novo mutation of FOXP1 (OMIM #605515) and an inherited mutation of
CNTNAP2 (OMIM #604569).
45
The validity of this model and the significance of concurrent CNVs can only be tested
by the consistent and detailed description of CNV cohorts in a whole-genome context.
This is especially true for studies such as ours, which involve only a single patient.
We would therefore urge researchers characterising chromosome abnormalities to consider,
and to explicitly report, the anomalies in the context of whole genome copy number
variation and genomic cataloguing. Advances in genetic technology mean that there
is no longer a need to consider genomic imbalances in isolation, particularly in case
reports. We suggest that the capture of complete genomic contexts, alongside detailed
phenotypic profiling, will allow us to develop a better understanding of the variability
of the chromosome 16p11.2 phenotype and may assist in the delineation of a core clinical
phenotype.