Since many pathogenic sequence variations have been identified in the ARSB gene, a genotype-phenotype correlation is slowly emerging [14–19] – a picture to which this description of our twelve patients and their genotypes now contributes. Five of our patients presented with rapidly progressive disease; progression in the other seven was slower. In total, we identified 9 different pathogenic mutations, 4 of which had not been published previously. Together they compose 8 different pathogenic genotypes. MPS VI is indeed very heterogeneous .
Genotypes and phenotypes
Like others, we found that the level of residual ARSB activity in cultured fibroblasts using rather non-specific artificial substrates is a poor means of differentiating between patients with different clinical phenotypes. The radiolabeled three-saccharide substrates perform better [20, 21]. By contrast, urinary GAG content seems to help indicating the degree of disease severity [12, 22]. In our patient group too, a high urinary GAG content (mainly dermatan sulphate) of 300 μgram/mg creatinine or higher was associated with a severe phenotype.
With regard to our patients’ genotypes, 4 patients were homozygous for missense mutations and one was homozygous for a splice-site mutation. The advantage of homozygous genotypes over compound heterozygous genotypes is that the impact of the genotype does not need to be gathered from the additive effects of the two individual mutant alleles. Karageorgos et al. were able to correlate the genotype, urinary GAG excretion and phenotype of several patients with homozygous missense mutations .
Three of our patients were homozygous for p.R152W. The fact that they were the least severely affected patients in our study group is consistent with the literature indicating that patients with this mutation are not amongst the most severely affected. One publication about a 38-year-old woman describes this mutation as being associated primarily with a ‘cardiac phenotype’ . While our three patients with this mutation do not have prominent cardiac signs, the oldest –who is currently 21 years old– does have mild aortic regurgitation. By contrast, four of the other missense mutations that we identified (p.N301K, p.V332G, p.G324V, and p.A237D) are severely pathogenic; this is suggested by the rapidly progressive phenotypes of the homozygotes in which they were found. Only p.G324V has been described previously, but in a compound heterozygote, meaning that its effect was still obscure . Its great severity is indicated by the fact that our patient who is homozygous for it had macrocephalia, hepatosplenomegaly, severe joint deformities and eye problems at a very young age (2y old). As patient 2, who is homozygous for 1142 + 2 T > C, also had a rapidly progressive phenotype we conclude that this newly discovered mutation must be severe too. Moreover, this mutation affects the invariant splice donor sequence and is therefore expected to completely disrupt splicing . Homozygosity for c.1142 + 1G > T in a patient with a very high GAG content in urine –like our patient– also resulted in total loss of ARSB protein . Among the remaining patients, we identified four compound heterozygotes carrying the common p.Y210C mutation in combination with either p.P313A (patient 8) or p.R327X (patients 9, 10 and 11). p.Y210C is known to be associated with an intermediate MPS VI phenotype, which was also the case in our patients with this mutation . Compound heterozygosity for p.Y210C and p.P313A was earlier described by Karakeorgos et al. in association with a slowly progressive phenotype .
The effect of 8 mutations was investigated by site directed mutagenesis and transient expression studies. Introduction of the non-sense mutation p.R327X in the wild type ARSB cDNA construct led to total absence of ARSB protein. The seven missense mutations had variable effects on the biosynthesis of ARSB; some resulted in (much) less than the normal amount of 66 kD precursor while others did not seem to have a dramatic effect on the amount of 66 kD ARSB as judged from several independent experiments. The low amount of 66 kD precursor in the respective cases might point to aberrantly folded proteins that are partly degraded by endoplasmic reticulum-associated degradation (ERAD) [12, 26, 27]. The expression of none of the 7 missense mutations in COS-7 cells led to the formation of both the 66 kD precursor as well as the 43 kD processed form of ARSB and none resulted in clearly measurable ARSB activity. This outcome agreed with the very low ARSB activity in fibroblasts of the patients. The lack of 43 kD ARSB by expression of p.Y210C was unexpected since Brooks et al. demonstrated in a pulse-chase experiment that 67% of p.Y210C-substituted ARSB is processed to mature 43 kD enzyme. However, the apparent discrepancy can relate to the fact that different methods and different antibodies were used (steady-state detection as opposed to pulse-chase labelling) .
Antibody response to galsulfase
Within 26 weeks from the start of galsulfase infusions all patients showed seroconversion, which is consistent with observations during the clinical trials . The ELISA titers differed between patients, varying during ERT from relatively low in some patients (1,250) to relatively high (1:156,250) in others.
Patient 2 consistently had the highest antibody titers and was the only patient who experienced infusion associated reactions (IARs). Notably, due to homozygosity for c.1142 + 2 T > C, the patient is not expected to have any level of endogenous ARSB expression, and is therefore immunologically naive to ARSB. This might underlie the strong immune response. Her high antibody titer as measured by ELISA was confirmed by immuno-precipitation. While the ELISA titer remained constantly high after 26 weeks of treatment, the titer obtained by immuno-precipiation declined. This would imply that the pool of antibodies directed against immobilized galsulfase (ELISA) remained constant while the pool of antibodies against native galsulfase decreased. The assay in which the pool of antibody-bound galsulfase is measured during enzyme infusion probably mimics the actual situation best; from it, we inferred that approximately 50% of galsulfase administered to patient 2 was captured by antibodies at the time that she had an ELISA titer of 1:31,250. In patient 6, whose ELISA titer was 1:6,250, the percentage of antibody-bound galsulfase was much lower.
With regard to the effect of antibody binding, we demonstrated for patient 2 that it can inhibit the uptake of galsulfase by cultured fibroblasts. It is plausible that a similar effect occurs in vivo and can limit the effect of ERT. In such a case, the adverse effect of antibody formation can possibly be overcome by higher dosing, unless higher dosing would lead to higher antibody titers or severe infusion-associated reactions.
We have no clinical indications that antibody formation in MPS VI has a very strong impact on the effect of ERT. Despite the much higher percentage of antibody-bound galsulfase during infusion in patient 2 compared to patient 6, both patients had similar increases in shoulder flexion, FVC and decrease in urinary GAG. It is known from other lysosomal storage diseases that immune response to ERT can reduce therapeutic outcome. In Pompe disease, for instance, antibody formation has been associated with loss of milestones and decreased pulmonary function . In MPS VI, however, clinical changes in response to ERT are more subtle and the adverse effect of antibodies may go unnoticed . We have investigated the effects of enzyme-replacement therapy in this patient group in a separate paper .