Goldberg–Shprintzen syndrome is determined by the absence, or reduced expression levels, of KIFBP

Abstract Goldberg–Shprintzen syndrome (GOSHS) is caused by loss of function variants in the kinesin binding protein gene (KIFBP). However, the phenotypic range of this syndrome is wide, indicating that other factors may play a role. To date, 37 patients with GOSHS have been reported. Here, we document nine new patients with variants in KIFBP: seven with nonsense variants and two with missense variants. To our knowledge, this is the first time that missense variants have been reported in GOSHS. We functionally investigated the effect of the variants identified, in an attempt to find a genotype–phenotype correlation. We also determined whether common Hirschsprung disease (HSCR)‐associated single nucleotide polymorphisms (SNPs), could explain the presence of HSCR in GOSHS. Our results showed that the missense variants led to reduced expression of KIFBP, while the truncating variants resulted in lack of protein. However, no correlation was found between the severity of GOSHS and the location of the variants. We were also unable to find a correlation between common HSCR‐associated SNPs, and HSCR development in GOSHS. In conclusion, we show that reduced, as well as lack of KIFBP expression can lead to GOSHS, and our results suggest that a threshold expression of KIFBP may modulate phenotypic variability of the disease.

Hirschsprung disease (HSCR) is reported in~70% of patients with GOSHS, but is considered to be a variable feature. HSCR is characterized by the absence of enteric ganglia in the distal colon and occurs in multiple defined syndromes (Amiel et al., 2008). The link between GOSHS and HSCR is poorly understood, especially considering the variability of its presence, even between family members sharing the same pathogenic variant. It is suspected that a balance of protective and/or predisposing factors in the (epi)genome influences HSCR development (Brosens et al., 2016;Chatterjee et al., 2016;Emison et al., 2005;Kapoor et al., 2015). Common variants in the Rearranged during transfection gene (RET), the Semaphorin 3A gene (SEMA3A), and the Neuregulin 1 gene (NRG1) are known to be associated with HSCR risk. Previous studies have already investigated the effect of common variants located in intron 1 of RET in a series of patients diagnosed with a Mendelian syndrome where HSCR is part of the phenotype (de Pontual et al., 2006(de Pontual et al., , 2007.
However, no such study has been performed yet for GOSHS.
Here, we provide an update of all KIFBP reported cases and add nine unpublished cases with six new KIFBP variants, three of which are missense. This is the first time that missense variants have been reported to play a role in GOSHS. Whether these missense variants also result in LOF is unknown. Therefore, we functionally tested the effect of these variants, on KIFBP expression levels and cellular localization.
We also investigated if the truncating variants described, result in loss of protein. Finally, we determined if the occurrence of HSCR in GOSHS can be explained by the presence of common modifier alleles.

| Patient inclusion
In this study, nine new patients were included (Table 1). These patients were seen routinely in hospitals in the UK, Ireland, Norway, Poland, MACKENZIE et al. | 1907 T A B L E 1 All published and unpublished patients with KIFBP variants and their clinical features (ENST00000361983.7)

| Sequencing
Sanger sequencing of KIFBP was performed for eight of the nine patients, as previously described (Brooks et al., 2005). A list of primers is available on request. Patient NL1 was the only one subjected to whole-exome sequencing, due to lack of phenotypic features characteristic of GOSHS. Sequencing data were analyzed using a filter for neuronal migration abnormalities, leading to the identification of two heterozygous missense variants in KIFBP. Both variants were confirmed by Sanger sequence. All new variants described were submitted to ClinVAr (http://www.ncbi.nlm.nih.gov/clinvar/).
All patients and parents were also Sanger sequenced for the presence of selected common HSCR associated polymorphisms in RET (Chatterjee et al., 2016), NRG1, and SEMA3A Kapoor et al., 2015). Primers used are listed in Table S1.

| Expression vectors
The pcDNA-HA-hKIFBP vector was described before (Alves et al., 2010). When produced from this vector, the KIFBP protein contains an N-terminal HA-tag. The three missense variants identified were generated by site-directed mutagenesis on pcDNA-HA-hKIFBP, according to the manufacturer's instructions (QuickChange II Site-directed Mutagenesis Kit, Agilent Technologies). We used the same vector and procedure to introduce the two frameshifts variants, as well as the two deletions identified. Sanger sequencing confirmed the presence of all the

| Cell lysates and western blot analysis
Twenty-four to 48 h after transfection, cells were washed with PBS and lysed as described before (Alves et al., 2010). Protein quantification was determined using the Pierce Bicinchoninic Acid Protein Assay kit (Thermo Fisher Scientific), and 40 μg of cell lysates were stored in loading buffer at −80°C before they were processed further. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by western blot analysis was performed using an in-house anti-HA antibody, and a GAPDH antibody (Millipore), both at 1:5000 dilution. Secondary antibodies used were the IRDye 680RD goat antimouse and the IRDye 800CW goat antimouse (Li-Cor), at 1:10,000 dilution.

| Statistical analysis
All results are presented as the mean ± standard error of the mean (SEM). All data were analyzed using a two-tailed Student t test. p < .05 was considered to be statistically significant.

| RESULTS
3.1 | Three novel homozygous truncating variants in KIFBP were identified in five patients with GOSHS Seven previously unreported patients with GOSHS, three sibling pairs and an isolated patient, were sequenced for KIFBP (Table 1).
Large exon deletions, as well as new frameshift variants in KIFBP, were identified in these patients ( Figure 1 and NO2. This deletion is homozygous and has been previously reported (Dafsari et al., 2015).
According to the guidelines established by the American College of Medical Genetics (ACMG), all truncating variants identified in our GOSHS patients are classified as pathogenic (Table S2). These variants are also predicted to result in a total loss of KIFBP expression (Table 2)

| Three missense KIFBP variants were identified in two patients
The first patient (NL1)   Mut1 and 0.0000318 for Mut2 on gnomAD browser). Based on the ACMG guidelines, both variants were predicted to be likely benign (- Table S2). However, according to different prediction tools, Polyphen-2, SIFT, and conservation scores there seems to be evidence to support a deleterious effect on the gene (  Figure 1; Table 1), which was inherited from the parents. This variant has not been reported before (gnomAD browser), and based on the ACMG guidelines is classified as pathogenic (Table S2). Prediction tools, predict this variant as benign (Table 2). However, its CADD score is 17.95, indicating a potentially pathogenic effect.

| KIFBP expression levels were reduced by the missense variants
To evaluate the effect of the missense variants, expression levels of KIFBP were determined after the transfection of HEK293T cells with constructs expressing the WT and mutant KIFBP cDNA. q-PCR results showed significantly decreased RNA levels for all mutants when compared with the wild-type (Figure 3a). All mutants also showed a significantly decreased protein expression of KIFBP, when compared with WT. But, as expected based on the RNA levels, Mut3 showed the lowest expression (Figure 3b,c).

| Cellular localization of KIFBP is unaffected by the missense variants
Tagged WT and mutant KIFBP constructs were overexpressed in HEK293 cells, to determine any effect of the variants in the organization or localization of KIFBP within the cell. The WT protein is seen to have high cytoplasmic expression, as previously described (Alves et al., 2010). For the mutant proteins, no effect on KIFBP localization was observed (Figure 4).
3.5 | Previously associated common SNPs in RET, NRG1, and SEMA3A do not affect HSCR development in GOSHS It is known that phenotypic variability exists in patients with GOSHS and that HSCR is a variable feature, even within families with the same KIFBP truncating variant (Table 1). Here, we investigated whether the presence of previously common SNPs associated with HSCR, would be the determinant factor for the presence of this disorder in GOSHS. The SNPs we decided to investigate are located in intron 1 of RET, SEMA3A, and NRG1. Although, all these SNPS have been described to increase the risk for HSCR (de Pontual et al., 2006(de Pontual et al., , 2007, we were unable to find a significant correlation between them and the occurrence of HSCR in this subset of GOSHS patients and unaffected family members (t test, p = .526; Tables 3 and 4).

| DISCUSSION
In this manuscript, we report nine new patients with variants in

KIFBP.
A common feature in all these patients is the presence of intellectual disability and developmental delays. However, the phenotypic spectrum is broad, with distinct facial morphology,

F I G U R E 2 Expression of KIFBP is lost in the presence of the frameshifts and nonsense variants identified. (a) q-PCR results
showing relative normalized expression of KIFBP following transfection with wild type (WT) or mutant constructs. All mutant constructs showed a decrease in KIFBP expression compared to WT levels. (b) Western blot of KIFBP expression following transfection of either WT or mutant constructs. No KIFBP expression was detected in any of the mutants. q-PCR, quantitative polymerase chain reaction; UT, untransfected microcephaly, and other CNS malformations. Interestingly, this wide phenotypic range is even found in siblings carrying the same variant.
As can be seen in Table 1, the incidence of HSCR in GOSHS is ∼70% (24/34). Seven out of 10 patients without HSCR, have a family member carrying the same KIFBP variant, that does have this disease, suggesting the presence of modifying factors, or the absence of protective factors in these patients than can tilt the balance in favor of HSCR. Here, we hypothesized that selected common HSCR modifier variants in RET (Chatterjee et al., 2016), NRG1, and SEMA3A (Kapoor et al., 2015) may work as these modifying factors, as it has been shown for other syndromes (Chatterjee et al., 2016;De Pontual et al., 2007;Tang et al., 2016). However, our results did not show any correlation between the incidence of HSCR and the presence of these common polymorphisms in the cohort analyzed (p = .526, Table 3). As HSCR is a complex genetic disease, multiple factors are known to play a role in its development in addition to genetic risk factors, such as epigenetic changes (Tang et al., 2013), protective pathways (Griseri et al., 2007), threshold numbers of cells (Barlow, Wallace, Thapar, & Burns, 2008), or stochastic chance (Cheeseman, Zhang, Binder, Newgreen, & Landman, 2014). Moreover, since HSCR is such a common feature in GOSHS, one might argue that its occurrence is coupled to the expression levels of KIFBP, or to changes in the signaling network regulated by this protein. Further research is therefore required, to investigate which of these hypotheses can explain the variability of such features.
LOF variants in KIFBP are known to cause GOSHS. However, there seems to be no correlation between the location of the variant, and the severity of syndromic characteristics (Figure 1), as they all seem to result in total loss of protein (Brooks et al., 2005;Drévillon et al., 2013). In seven of the nine patients reported here, nonsense variants or frameshifts were identified, resulting in loss of expression of KIFBP (Figure 2  otherwise, as a reduction of KIFBP expression was detected for the three variants tested (Figure 3a-c), suggesting that a threshold expression of this protein may be required for regulation of developmental functions. While patient NL1 shows a mild reduction of KIFBP expression that seems to be, to some extent, tolerable, in patient CYP3 this threshold was not reached, leading to the typical GOSHS phenotype. Therefore, we conclude that missense variants in KIFBP can be as damaging as truncating variants, depending on its effect on protein expression levels.
It has been previously noted that the diagnosis of GOSHS should rely on molecular and genetic findings in place of phenotypic recognition only, due to its similarity with other syndromes (Salehpour et al., 2017). Based on our genetic findings, patient NL1 would be considered to have GOSHS due to the fact that no likely pathogenic variant has been identified in any other gene. However, this patient has no hallmark features of GOSHS. Since locus heterogeneity is lacking in this syndrome, and all patients with the typical features have KIFBP variants, we believe that the accurate classification of GOSHS based on phenotype by a clinical geneticist may be more useful for the family to appropriately meet the needs of the patient, as well as for advising clinical treatment.
T A B L E 4 Statistical analysis shows no correlation between the presence/absence of common SNPs in RET, NRG1, and SEMA3A and HSCR development in patients with GOSHS Abbreviations: ANOVA, analysis of variance; GOSHS, Goldberg-Shprintzen syndrome; HSCR, Hirschsprung disease; SNPs, single nucleotide polymorphisms.