Background
Mucopolysaccharidosis IV-A or Morquio A disease (MIM #253000] is an autosomal recessive lysosomal storage disease caused by the deficiency of
N-acetylgalactosamine-6-sulfatase (GALNS), the lysosomal enzyme responsible for the hydrolytic degradation of keratan sulfate and chondroitin-6-sulfate [
1]. GALNS is encoded by the
GALNS gene (NM_000512.4) which is located on chromosome 16q24.3; the gene has a length of about 50 kb and is organized into 14 exons [
2]. The
GALNS gene is alternatively spliced, with two other reported protein coding transcripts in the RefSeq database (NM_001323543.1 and NM_001323544.1) [
3]. In Ensembl, 13 transcripts are reported of which four are protein coding [
4].
In excess of 330 different mutations have been reported in the
GALNS gene causing Morquio A disease (Human Gene Mutation Database;
http://www.hgmd.org). Among them, only one solitary deep intronic mutation which created a cryptic donor splice site was previously reported [
5]. Empirically, standard sequencing procedures, covering all
GALNS exons and intron/exon boundaries, have failed to identify ~ 16% of mutant alleles in patients affected by the disease [
6,
7]. However, this percentage falls to 5% when gross DNA rearrangements are also screened for [
8].
Morquio A disease affects multiple organ systems but its principal features are the cartilage defects, caused by keratan sulfate accumulation, that are responsible for the typical skeletal complications including coxa valga, scoliosis, short trunk dwarfism and cervical instability [
1,
9]. The diagnosis tends to be particularly challenging in attenuated Morquio A patients, with consequent increased risk for missing the correct diagnosis [
1].
Several therapeutic approaches including hematopoietic stem cell transplantation, gene therapy, substrate reduction therapy, and enzyme replacement therapy (ERT) have been developed in order to try to ameliorate the disease [
10‐
14].
Newborn screening, whether by tandem mass spectrometry on dried blood spots and/or by fluorimetric assays, has proved to be both reliable and effective in identifying most mucopolysaccharidoses, including Morquio A disease [
15,
16]. It is likely that the widespread adoption of this methodology will have a significant impact on the diagnosis of Morquio A disease [
9,
17,
18].
Here we delineate the criteria that we have found efficacious in making an early diagnosis of Morquio A disease, together with a novel screening strategy that we have devised in order to optimize the probability of obtaining a molecular diagnosis in each case. Adoption of this strategy allowed the identification of novel splicing defects in three individuals in whom only one GALNS coding region mutation had originally been found.
Discussion
The spectrum of pathological mutations and benign polymorphisms in the
GALNS gene displays considerable allelic heterogeneity [HGMD Professional (Stenson et al., 2017), Exome Variant Server (
http://evs.gs.washington.edu/EVS/), GALNS Mutation Database (
http://galns.mutdb.org/database) etc.]. However, in approximately 16% of patients, the anticipated second disease-causing
GALNS mutation cannot be unequivocally identified within the gene coding region or at the exon-intron boundaries [
6,
7].
Establishing a diagnostic plan, including the requisite genetic analyses, is essential to distinguish between bona fide Morquio A disease patients and individuals with other disorders presenting with similar clinical and radiological findings [
6,
9,
12]. In addition, early diagnosis is crucial for the prompt deployment of available therapies before permanent systemic lesions occur. Here, we have integrated a series of clinical and analytical tools to provide a diagnostic flow chart for Morquio A disease (Fig.
1). In the algorithm we propose, next generation sequencing (NGS) procedures may be employed in two distinct analytical steps (Fig.
1). In the first, the exonic sequences and exon-intron boundaries of the gene in question are sequenced by NGS methodology (whole exome); alternatively, whole genome sequencing can be employed to sequence the entire gene region (~ 50 kb). In a second step, high-throughput sequencing-based methods can be used to perform transcriptome analysis (RNA-Seq) [
35], including of course, in our case,
GALNS transcripts.
We have presented here the cases of three Morquio A patients in whom the second disease-causing GALNS mutation was not initially identifiable either by standard sequencing procedures or by the analysis of gross DNA rearrangements and instead had to be determined by means of RT-PCR. Aberrant GALNS mRNA splicing products were noted in all three patients. In Pt1, the deep intronic mutation c.899–167 A > G was unequivocally identified as the lesion responsible for the aberrant mRNA (including an elongated exon) and a prematurely truncated GALNS protein. Hence, this aberrant splicing event can be directly and unambiguously related to the severe clinical phenotype observed in this patient.
Mutations located within deep intronic regions, that appear capable of promoting the use of alternative natural or non-natural splicing sites, were identified by GALNS whole gene sequencing analyses of Pt2 and Pt3 samples. After following the variant prioritisation protocol described above, specific GALNS variants in Pt2 (c.1002 + 307G > C) and Pt3 (c.759-67G > A) were predicted to make a contribution to the clinical phenotype in these individuals by impacting mRNA splicing. These variants exhibited a very low Minor allele frequency (MAF), and were predicted to modulate splicing, particularly with respect to potential ESE and ESS sequences.
Since both of the aberrantly spliced products detected in Pt2 and Pt3 disrupt exons, the mechanism responsible for these splicing alterations cannot be precisely ascertained. Indeed, it remains possible that a particular combination of variants could have been responsible for the observed splicing defects, rather than one variant on its own. Consequently, it may be that any of the other variants detected in the patients, including the putatively non-pathogenic GALNS variants (143 variants in Pt2 and 69 variants in Pt3, post-prioritization), may have contributed to the generation of the non-physiological splicing events detected e.g. c.423-862C > T identified in Pt3 and his mother. It should also be appreciated that variants such as c.423-862C > T may disrupt canonical splice junction sequences, i.e. cryptic acceptor and donor splice sites. Thus, it may well be that it is the combination of altered canonical and non-canonical splice sites in both patients that gives rise to these unique splicing alterations.
The GALNS gene is known to be alternatively spliced, with at least three known protein coding transcripts currently annotated (NM_00512.4, NM_001323544.1 and NM_00132354.1). It is possible that these alternative transcripts were differentially expressed between Pt3 and his mother; if so, this might have led to changes in splicing factor supply and demand, which could in turn account for the differences in the observed mRNA splicing phenotype between the patient and his mother.
Owing to the difficulties inherent in interpreting this atypical type of splicing event and our veritable ignorance of splicing regulatory regions, our approach did not unequivocally identify disease-causing variants at the DNA level for Pt2 and Pt3 even after intensive DNA sequence analysis of the
GALNS gene region. Additional functional studies, including the construction of expression systems that variously combine the identified candidate mis-spliced variants, would be necessary to formally confirm our hypotheses. However, for diagnostic purposes, the pathogenicity of the observed mRNA splicing defects is evidenced by: a) the absence of RT-PCR amplification, corresponding to the aberrant mRNA transcripts detected in Pt2 and Pt3 samples, in a pool of 10 normal controls both treated and untreated with cycloheximide; b) the splicing products detected in Pt2 and Pt3 differing from the 13 known
GALNS physiological mRNAs, collected in the Ensembl Human Genome browser (
http://www.ensembl.org/index.html). These considerations imply that the alternative
GALNS mRNA splicing products detected in Pt2 and Pt3 are non-physiological, and are therefore likely to be consequent to the (hitherto unidentified) second disease-causing
GALNS alleles in these Morquio A patients.
Conclusions
Morquio A disease is particularly prone to delayed diagnoses and/or misdiagnoses, owing to the difficulties inherent in the differential diagnosis of this rheumatic disease that requires specialist metabolic expertise. The addition of mRNA analysis and whole GALNS gene sequencing to this flowchart promises to help to identify those molecular causes of Morquio A disease which until now have been refractory to analysis. These analyses are likely to be particularly important for Morquio A screening programs in which the drawing up of a general diagnostic molecular plan is key to distinguishing between newborns who are carrying mutations associated with severe forms of the disease and those who are carrying mutations that are likely to give rise to milder or asymptomatic forms.
Our sequence analysis of the GALNS gene, involving gene level, genomic and RT-PCR analyses, suggests that although deep intronic mutations may be individually infrequent, they may be largely responsible for our occasional failure to identify GALNS disease alleles in Morquio A disease; it follows that marked improvements in our knowledge of the splicing machinery will be required before any diagnostic workflow can be regarded as being 100% effective.
Acknowledgements
We gratefully thank the AMMeC (Associazione Malattie Metaboliche e Congenite, Italia) for technical assistance. We also thank Dr. Marzia Rossato, Functional Genomic Laboratory, Department of Biotechnology, University of Verona, for the whole genome sequencing service.