Background
Williams syndrome (WS) is a rare autosomal dominant disorder arising from the hemizygotic deletion of approximately 27 genes on chromosome 7, at locus 7q11.23 [
1,
2]. The deletion occurs spontaneously during meiosis and is due to unequal crossing over at misaligned repeat segments [
3]. This typically results in a deletion spanning some 1.55 Mb (approximately 95% of cases) to around 1.84 Mb (approximately 5% of cases) of genomic DNA [
4‐
8]. However, a number of individuals with partial deletions within the WS critical region (WSCR) of chromosome 7 have also been identified [for examples, [
9,
10]. Such cases can provide important insights into the contribution of specific genes to the phenotypic outcome of WS.
Given the uneven cognitive profile characteristic of WS, particularly the contrast between poor non-verbal abilities relative to verbal cognition [for example, [
11], research into individuals with partial deletions has sought to identify candidate genes responsible for deficits in domains such as global intellectual difficulties [
10], social cognition [
12], and spatial cognition [
13‐
15].
In the case of individuals with the full WS deletion, spatial deficits have been well documented, with poor performance reported on visuospatial construction tasks, [
16,
17], mental imagery [
18,
19], and the use of spatial frames of reference [
20,
21]. More recently, deficits in large-scale spatial navigation have also been identified in WS [
22,
23]. But which genes contribute to these small-scale and large-scale spatial impairments remains a topic of debate. One study that examined two families with a partial WS phenotype, including supravalvular aortic stenosis (SVAS) and deficits in visuospatial construction, found that affected family members were hemizygous for the elastin (
ELN) and LIM-Kinase1 (
LIMK1) genes [
13], which lie within the WSCR. Given that
ELN is not expressed in the brain and mutations of which are not associated with spatial deficits but with cardiovascular abnormalities, it was concluded that it must be the other deleted gene,
LIMK1, that plays an important role in the phenotypic expression of impaired spatial cognition in WS. Indeed,
in vivo,
Limk1 knockout mice have impaired spatial learning performance when tested on reversal learning in the Morris water maze [
24]. They also present with abnormal synaptic structure and neuronal spine morphology, as well as altered hippocampal long-term potentiation.
The role of
LIMK1, however, has remained inconclusive, with other studies of patients with partial deletions that include
LIMK1 suggesting that hemizygosity for this gene does not in itself result in deficits in visuospatial cognition [for examples., [
9,
15]. Using a large battery of perceptual and visuospatial tasks, Gray and colleagues [
15] report a very detailed assessment of two patients with deletions of only
ELN and
LIMK1, compared with two adults with full WS matched on verbal ability. A profile of normal spatial performance emerged from the two partial deletion patients compared to those with the full deletion, suggesting that
LIMK1 alone did not explain spatial deficits in WS. In addition, the successful performance of these same two partial deletion patients on a large-scale spatial task indicated that the hemizygotic deletion of
LIMK1 was also insufficient to result in the poor large-scale search strategies identified in individuals with full WS on the same task [
14].
These findings concur to indicate that the sole deletion of LIMK1 is not sufficient to result in deficits in any form of spatial cognition. Instead, the authors suggest that LIMK1 may play a role in the spatial cognitive profile in WS only when deleted alongside other genes, particularly those at the telomeric end of the WSCR. It is these latter genes that have been the focus of recent studies.
Hirota
et al. [
25] present a detailed analysis of the relationships between partial deletions in three patients and their performance on standardised psychometric tests. The authors suggest that the general transcription factors
GTF2I and
GTF2IRD1, at the telomeric end of the WSCR, are likely to play a disproportionate and crucial role in the development of neural pathways involved in visuospatial cognition. Dai
et al. [
26] also sought to delineate the role of these transcription factors in the WS phenotype, finding that an individual with an atypical deletion that included
GTF2IRD1, but not
GTF2I, presented with poor performance on a number of spatial subtests from the Wechsler Preschool and Primary Scale of Intelligence- Revised (WPPSI-R) [
27], including ‘Block Design’, ‘Object Assembly’, and ‘Mazes’. Preservation of the normal copy number of
GTF2I in this individual was argued to contribute to the relative strengths found in those non-verbal cognitive measures that did not require visual-motor integration (‘Picture Completion’ and ‘Geometric Design Recognition’), and in verbal cognition. It appears, then, that the general transcription factor genes at the telomeric end of the WSCR are likely to make a significant contribution to the WS visuospatial phenotype.
The general transcription factor (GTF) genes are also thought to have widespread effects on the expression of other genes [
28], and are differentially expressed in the developing brain compared to the adult brain [
1]. The impact of mutations of these telomeric genes, in particular on the expression of other genes, may therefore be diverse and have varying cascading effects throughout development. As such, to gain a more thorough understanding of the role, combinatorial effects, and penetrance of all 28 transcripts within the WSCR (particularly the GTF genes) on the phenotypic profile of WS, research must examine the different effects of specific genetic mutations across partial deletion patients with differing genomic makeup.
It is not only at the level of the genotype that more in-depth research has been necessary; the phenotypic outcome also calls for more subtle analyses rather than solely relying on psychometric spatial tasks. Indeed, recent research has sought to elucidate whether there are dissociable deficits within the visuospatial domain in individuals with the full WS deletion. In particular, the different cognitive demands associated with understanding the location of the self (‘egocentric’ spatial representations) and object-based spatial relationships (‘allocentric’ spatial representations) have been examined through the use of both small-scale table-top tasks [for example, [
20] and large-scale navigation tasks [for examples, [
14,
23]. The findings of such studies have identified difficulties in the use of both egocentric and allocentric spatial representations in WS. However, little can as yet be concluded regarding the genetic contributions to the specific deficits in the use of these different spatial frames of reference. Given the importance of these aspects of spatial representation to human navigational abilities, and hence to every-day living, attempts to examine the genotypic correlations with these specific spatial deficits must address the use of these different cognitive processes both in individuals with the full WS deletion and in individuals with partial deletions within the WSCR. We therefore argue that it is critical to test, not only performance on psychometric spatial tasks, but particularly performance on novel, hypothesis-driven small-scale tasks and navigational large-scale search tasks that tap into egocentric and allocentric spatial cognitive demands.
Here, we present case studies of spatial cognition in two individuals (HR and JB) neither of whom meets both genetic and phenotypic criteria for a typical diagnosis of WS, but present with contrasting partial genetic deletions within the WSCR. Previous comparisons of the socio-cognitive profiles of these two patients highlighted the different levels of social impairment that result from such contrasting deletions [
12]. The current study focussed on visuospatial cognition and examined the impact of these differing genetic deletions in the WSCR by investigating performance of HR and JB, using a range of table-top psychometric tasks, and small- and large-scale spatial tasks. Visuospatial abilities of both individuals were also compared to performance on the same tasks previously reported both in typical development and in individuals with the full WS genotype.
Discussion
The uneven cognitive profile in WS has provided insights, but also controversies, into the genetic contributions to human visuospatial cognition. Indeed, the elucidation of which of the 28 WSCR genes play a role in the visuospatial phenotype in WS is complex. Initial studies had implicated
LIMK1 as a major contributor to the visuospatial deficits in WS, on the basis of human partial deletion patients and mouse models [
13,
24]. However, subsequent work on other partial deletion patients showed that if
LIMK1 played a role, it had to be in combination with other genes at the telomeric end of the WSCR [
9,
14,
15]. The two genetically contrasting case studies presented in this paper provide further insight into the possible combinatorial effects of genes within the WSCR, including the role of the general transcription factors on
some aspects of visuospatial cognition. Moreover, whereas previous studies had compared small-scale, table-top spatial deficits in WS with large-scale navigational deficits in the mouse, which place very different cognitive demands on each species, the current study examined both small- and large-scale visuospatial abilities in the same participants. This more in-depth analysis of visuospatial cognition is critical if we are to understand genotype/phenotype relations in Williams syndrome.
An interesting pattern of strengths and weaknesses within the spatio-cognitive domain emerged in both participants. In particular, HR showed poor performance on psychometric measures of spatial intellect (BAS-II) and mental-rotation, alongside a relative strength in non-verbal reasoning and VPT. HR’s performance on our battery of tasks was, despite her deletion of over 24 genes in the WSCR, therefore not reflective of the cognitive profile of individuals with the full WS deletion. Indeed, HR performed at a level significantly above that observed in individuals with full WS on the VPT task, but below an age-appropriate level, and in line with individuals with WS on the MR task.
Neither was the pattern of performance across the small- and large-scale experimental tasks in HR entirely typical. Relatively poor performance by HR both on the psychometric measure of spatial intellect and the mental rotation task, alongside proficient performance on VPT, large-scale navigation and route learning, reflects the multi-faceted nature of spatial cognition. In typical individuals, moderate correlations are found between performance on psychometric and small-scale spatial measures and large-scale spatial abilities [
50,
51]. However, in typical development, although performance on tests of visuospatial memory correlates highly with route learning ability, this is largely moderated by executive control [
52]. As such, although somewhat overlapping, small and large-scale spatial abilities are partly dissociable and it can be inferred that visuospatial abilities may be differentially affected by divergent genetic deletions.
What about the relationship between VPT and navigation strategies? Moderate correlations are also found in typical adults between VPT and the ability to use an allocentric navigation strategy in large-scale space [
53]. Accordingly, HR’s high level of performance on the VPT and navigation tasks (in contrast to individuals with the full WS deletion, who show substantial deficits on both of these tasks) suggests that she may be able to use this ability to spatially update the location of the self and to navigate successfully following a change in position in a large-scale familiar environment. That said, little is known about the extent to which performance on VPT tasks account for the variance in navigational abilities across development. Indeed, a high level of performance in JB on the VPT task contrasted with his poor navigational ability and low overall cognitive functioning (discussed later) suggests that they are not wholly associated, and may be differentially affected by genotypic variation.
In summary, in spite of HR’s deletion of over 24 genes on the WSCR, including haploinsufficiency for GTF2IRD1, only some atypical spatio-cognitive functioning was observed that resembles that of individuals with full WS. These findings suggest that the retention of the more telomeric 7q11.23 genes contribute to the relatively good large-scale visuospatial performance observed in HR, particularly in the face of her relatively low level of cognitive functioning as measured on Verbal and Spatial psychometric scales. However, it remains inconclusive from this whether each of these telomeric genes - GTF2I, NCF1 and GTF2IRD2- play an equal role in large-scale spatial cognition.
If
LIMK1 (deleted in HR) plays a role in spatial cognition, then it may be that the preservation of the most telomeric genes within the WSCR allowed for the development of compensatory spatial strategies in HR, reflected in part by her good performance, albeit with some atypical features, on the large-scale navigation and route learning tasks. That said, it is difficult to disentangle between whether such compensation is due to a genetic mechanism, or to the application of an alternative strategy that transpired through specific training. Further consideration of the modulatory role of both
LIMK1 and GTFs on various visual processes within the WS phenotype, particularly in regards to their expression in different neural tissues [for example, [
54], is therefore imperative. As such, more comprehensive phenotypic studies at different levels of the nervous system would highlight more specifically the profile of visuospatial strengths and weaknesses in relation to the genetic underpinnings.
The profile of HR was presented alongside that of JB, an individual with a contrasting hemizygous deletion, which extends telomerically from within
GTF2I to beyond the WSCR. Two quite differing cognitive profiles emerged from HR and JB. At the cognitive level, JB presented with profound impairments across the Verbal, Non-verbal and Spatial domains, as measured using the BAS-II. Despite his preservation of the majority of the genes on the WSCR, this profile of deficits is expected, given the probable role of the telomeric genes on the expression of other genes [
1] as well as the role of
GTF2I on general intellectual ability [
10]. It should be noted however, that JB’s deletion includes haploinsufficiency for up to 21 genes, many of which have unknown function. As such, it remains unclear what contribution they make to his profile. For example, haploinsufficiency for
HIP1 (deleted in JB) has been reported to be associated with neurological and neuropsychological deficits including epilepsy and autistic traits in other individuals with atypical deletions flanking the WSCR [
55]. It is therefore important to take into account that JB has a large number of genes deleted outside of the WSCR, and is likely to have altered expression of
GTF2I and is haploinsufficient for
HIP1, deletions of which are known to contribute to lower cognitive functioning. Conclusions regarding comparisons of the two cases are therefore tentative. That said, the inclusion of these two cases together provides insight into the differential effects of deleted genes within 7q11.23 on visuospatial abilities at different spatial scales, in the face of differing overall levels of intellectual ability.
Surprisingly, JB performed at a very high level on the small-scale mental rotation and VPT tasks, with scores significantly above those observed in individuals with full WS. By contrast, on both of the large-scale navigation tasks, and like individuals with WS, JB took a long time to learn the route, with performance below TD 8 and 10 year-olds. Allocentric spatial coding was also somewhat compromised, although not significantly different from any TD or WS groups (possibly due to the stringent nature of the modified
t-test, given that differences are previously reported across TD and WS groups on allocentric score [
23]). Despite these impairments on large-scale spatial tasks, his performance reflected an overall profile unlike that observed in individuals with the full WS deletion. Indeed, the uneven profile of JB’s relative strengths and weaknesses across different spatial scales in the visuospatial domain was also not comparable to that observed in TD individuals. Furthermore, given that JB performed at a high level on mental rotation and VPT tasks, his relatively poorer performance on large-scale navigation cannot purely be considered a reflection of low general cognitive ability. Instead, these findings strongly indicate the role of other genes at 7q11.23 in mental rotation ability other than
GTF2I or
GTF2IRD2 (for which JB is haploinsufficient). Similar robust conclusions cannot be made regarding the role of WSCR genes on VPT ability, given that both JB and HR performed at a high level on this task, and the vastly differing intellectual profiles of the two cases. However, there may be combinatorial effects of the GTFs and other more centromeric WSCR genes on VPT, although this can only be tentatively inferred given the difficulties in drawing direct comparisons from the two cases presented here.
Across the tasks in the current study, neither HR nor JB presented with a clear WS spatio-cognitive profile, and both performed outside of the typically-observed variations in performance by individuals with full WS. A spatial advantage, particularly for mental rotation [
56], is usually attributed to males, which could, at first blush, be considered to explain the higher level of performance on this task by JB than HR. However, no gender differences were apparent in the TD or WS participants across the battery of spatial tasks employed here, including in mental rotation, suggesting that differences in the two cases presented here are not likely related to gender. Furthermore, on other tasks HR typically performed at a higher level than JB, which is more likely a reflection of their differences in overall intellectual functioning, than of gender.
These contrasting profiles pose an interesting question as to the combinatorial effects of genes at locus 7q11.23 on the WS visuospatial phenotype, particularly those at the telomeric end. As mentioned, HR’s deletion includes that of
LIMK1, a gene that had gained much attention in the search for mapping from genotype to spatial phenotype. Although
limk1 plays a critical role in long-term potentiation in the mouse hippocampus [
24], research in humans with atypical deletions in the WSCR has challenged the independent contribution of
LIMK1 to the visuospatial deficits seen in WS [for examples, 9,14,15]. More recently, the chromosomal region telomeric to RFC2, including
CYLN2,
GTF2IRD1, and
GTF2I, has become a focus of interest as a possible contributor to the spatial cognitive profile in WS [
6,
57].
CYLN2 (also known as
CLIP2), for example, encodes CLIP-115, which is expressed in dendrites and cell bodies in a number of brain regions, and has been found to effect hippocampal memory processes [
57]. HR is not deleted for
GTF2I or other more telomeric genes, but her deletion does include a reduced expression of (and thus haploinsufficiency for)
GTF2IRD1. Given HR’s difficulties in mentally rotating objects, these results support previous findings that haploinsufficiency for
GTF2IRD1 in combination with other 7q11.23 genes such as
CYLN2 or
LIMK1 may play a role in some of the (small-scale) visuospatial cognitive deficits observed in individuals with WS [
14,
58]. This is supported by JB who is not deleted for either of these genes and performed well on the mental rotation task. Nonetheless, we cannot rule out that the deletion of the other more telomeric general transcription factors has impacted the expression of
CYLN2 and other 7q11.23 genes in JB [for example, see [
58]. It also remains unclear whether the deletion of other genes beyond the WSCR plays a role in the expression of intact genes within the WSCR.
Initially, the examination of visuospatial performance by HR and JB seems indicative of the additive effect of deleting each of the
GTF2I family genes on the severity of cognitive impairment, and is in line with other findings in individuals with extended deletions (approximately 1.8 Mb) that encompass
GTF2IRD2, who present with significantly greater neurological impairments than individuals with shorter deletions typical of WS [
59]. However, although we cannot dismiss the effects of other deleted genes in JB that extend beyond the WSCR on cognitive functioning, JB’s high level of performance on small-scale mental rotation and VPT rules out the conclusion of a general deficit, and excludes the role of genes telomeric to
GTF2I in these specific small-scale spatial abilities. Indeed, the preservation of most other genes within the WSCR in JB suggest that spatial skills may be differentially affected when GTF genes are deleted in combination with other more centromeric genes on the WSCR. Of note here, is that information regarding whether the deleted region in any participants with WS in the study also included
NCF1 and
GTF2IRD2 (an approximate 1.8 Mb deletion that occurs in around 5% of cases) was not obtained. This is because, for the majority of individuals with WS, the genetic contributor to diagnosis is via a Fluorescent
in situ Hybridisation (FISH) test, which does not provide deletion size information. We assume that the majority of our sample had a standard 1.55 Mb deletion in line with 95% of the WS population. Inclusion of individuals with greater deletions would have resulted in an underestimation of the ability of the WS group. That said, JB’s deletion did include these genes and was found to perform at a higher level than WS on some tasks and in line with WS on others, findings that would not have transpired had WS group data been compromised due to the inclusion of such individuals.
While the current study has highlighted candidate genotype/phenotype relations in the presentation of two contrasting case studies, it is clear that future studies need more in-depth genetics and phenotypics, measured over developmental time [
60]. Indeed, with reference to genetics, it would be preferable to perform DNA and RNA sequencing to delineate the precise genomic boundaries of the deleted regions, and to examine gene expression patterns in the WSCR region and throughout the genome [
4]. However, it is also critical that such detailed genetic studies be accompanied by in-depth phenotypic studies, particularly at the cognitive level. The phenotypic analysis requires a broad spectrum of tasks, both psychometric and hypothesis-driven, like in the current study, [see also, [
15] as examining performance only on psychometric tasks [for example., [
4] does not yield a detailed account of phenotypic expression, particularly regarding the multi-faceted nature of spatial cognition. The fact that genome function is modified over developmental time as a function of the epigenome requires a longitudinal assessment from infancy onwards of the details of the changing phenotype. Furthermore, case studies do not make it possible to ascertain whether gender influences gene expression in the WSCR nor whether, in our particular cases, JB or HR have other genetic mutations elsewhere in their genome, outside the WSCR, that may affect general intellectual outcome.
Genetic and phenotypic examination of other family members would be critical to complete the picture of such case studies. Thus, numerous factors that include changes in gene expression across development, but also environmental influences, education and other individual differences may contribute to the complex phenotypic outcomes in HR and JB. For this reason, conclusions regarding genotype-phenotype associations from individuals with partial deletions must always be considered with some caution and supplemented by appropriate animal models. Moreover, such genotype-phenotype correlations should take into account the possible role of candidate genes in the development of other neural tissues. For example, Castelo-Branco and colleagues [
54] examined the contribution of the general transcription factors to the neural retinal phenotype in WS, finding patterns of visual impairment that were separate from the known cortical dorsal-stream phenotype. This highlights the important nature of in-depth phenotypic analyses in order to draw more robust conclusions as to the contributions of specific genes to cognitive phenotypes.
Authors’ contributions
HB, EKF, and AK-S conceived of the study, participated in the design, data collection and analysis, and manuscript presentation; EC contributed to the conception, ran control participants, and design of the study and manuscript preparation; KM, MT, FS and PT contributed to the genetic analysis and participant recruitment and participated in manuscript preparation; EM contributed to interpretation of genetic analyses, preparation of the manuscript and Figure. All authors read and approved the final manuscript.