Data on evaluation of TCF4 variants.

The evaluation of variants identified in TCF4 should rely, in the first instance, on the use of the ACMG guidelines. When dealing with patients with PTHS, the criteria that are usually considered are the following:
a) PVS1 (very strong evidence): predicted null variant in a gene where LOF is known mechanism of disease;
b) PM2 (moderate evidence): the variant is absent in population database;
c) PS2 (strong evidence) or PM6 (moderate): de novo variant (paternity and maternity confirmed for PS2);
d) PS1 (strong): same amino acid change as an established pathogenic variant;
e) PP4 (supporting evidence): patient’s phenotype highly specific for the gene.

More specifically for missense variants, the following criteria are also taken into account:
a) PM1 (moderate): mutational hot spot or well‐studied functional domain without benign variation;
b) PM4 (moderate): novel missense change at an amino acid residue where a different pathogenic missense change has been seen before;
c) PP2 (supporting): missense in gene with low rate of benign missense variants and pathogenic missenses common;
d) PP3 (supporting): multiple lines of computational evidence support a deleterious effect on the gene /gene product.

When a TCF4 truncating variant (PVS1), absent in population databases (PM2), is detected in a subject with the PTHS phenotype (PP4), it has to be considered pathogenic with even no need for testing parents.
However, considering the very important role that can be played by the PVS1 parameter in the evaluation of pathogenicity, some caution is in order, because different kinds of variants may have diverse effects on the multiple TCF4 transcripts.

The following aspects should be considered, with specific attention to variant categories:
1) Truncating variants, including both nonsense substitutions and frameshift insertions or deletions of few nucleotides. To the best of our knowledge, no truncating point variant upstream exon 7 has been reported yet in PTHS patients, thus, at least for the time being, should such a variant be found, the PVS1 parameter could not be assigned, but the other criteria should be evaluated. This consideration should be applied to variants in internal exons from 2 to 6, and, moreover, to all 5’ exons that are exclusive for individual TCF4 transcripts.
Variants in exons 7 and 8 have also been rarely reported, including two recurring nonsense and one frameshift variant. While they are usually associated with a phenotype that shows an at least partial overlap with PTHS, in some other cases they may also be responsible for a milder phenotype that can hardly be defined as PTHS (Hamdan et al. 2013). Consequently, even if a PVS1 evidence can be assigned to truncating variants in exons 7-8, caution should be used in defining the resulting phenotype as PTHS.
Truncating variants in exons 9-18 are almost invariably associated with PTHS.
No nonsense variant has been reported yet at the extreme 3’ end of the gene, specifically in exon 19 downstream the region coding for the bHLH domain, thus no evidence can be drawn at the moment.
Frameshift variants resulting in the elongation of the protein are possible and can be classified in 2 types: 1) deletions of 1 bp or insertions of 2 bps (plus, possibly, a number of nucleotides that is a multiple of three) downstream the codon coding for lysine 638 in exon 19, resulting in a protein that is 10 amino acids longer than usual; 2) deletions of 2 bps or insertions of 1 bp downstream the codon 602, coding for aspartic acid in exon 18, which give rise to a 37 amino acid longer protein.
Two instances of the second type have been described to date in PTHS patients (Zweier et al. 2008, Steinbusch et al. 2012), both in exon 19 of TCF4 and both responsible for the typical PTHS phenotype. We might expect that downstream variants of this kind, possibly involving part of the bHLH domain in exon 18, would be even more detrimental. Consequently, we may reasonably expect that this type of elongating variants are loss of function. This hypothesis is further supported by functional studies performed by Sepp and colleagues (Sepp et al. 2012).

A single patient has been reported with an elongating variant of the first category (a deletion of 1 bp) (Goodspeed et al. 2018), and he was defined as “the highest functioning patient described to date”, but the provided clinical description is not sufficient to classify him as PTHS. For this reason, there is not enough evidence to expect that such kind of variants are always pathogenic.
In TableS3 it is indicated when the PVS1 parameter can be used for a truncating variant by considering the involved exon.

2) Splice variants. Few variants altering normal TCF4 splicing have been reported in literature, all affecting intronic regions adjacent to exons 9-18. To obtain a higher level of evidence when assessing their pathogenicity, their consequences at mRNA level should be investigated whenever it is possible. If a certain variant is demonstrated to cause a disruption in the reading frame of resulting transcripts, it can be considered as a frameshift variant, with all the related considerations listed in the previous paragraph. Conversely, milder or no phenotypical consequences could be expected in case of single splice variants resulting in an in-frame inclusion of an amino acid, as shown in a patient by Redin et al. (2014), who presented with a nonspecific form of mild intellectual disability. In this regard, it is worth of note that some known functional transcripts of the TCF4 gene result from an alternative splicing of certain exons (specifically, 7, 9, 14, and 17), and it should be taken into account when evaluating potential splice variants in these alternatively spliced regions.
3) Partial gene deletions (and duplications) and balanced translocations. An in-depth characterization of the observed rearrangement is usually required, since it is important to specifically define the exons involved and the structural consequences at level of the genic region. If a disruption of the coding sequence of exons 7-19 is predicted, a severe phenotype likely resembling PTHS can be reasonably expected, while milder forms of ID have been observed in association with partial deletions of the first six exons or balanced translocations
with breakpoint upstream exon 6. An analysis of the resulting mRNAs constitutes a very useful tool in further defining the functional consequences of this category of variants, which have been observed in an increasing number of subjects with different and peculiar phenotypes (Bedeschi et al. 2017, Goodspeed et al. 2018).
4) Missense variants. They constitute a common cause of PTHS, accounting for about 20 % of reported patients. In most cases, recurrent variants in bHLH domain are found, with a quite straightforward evaluation of their pathogenicity, if both PS1 and PS2 criteria are satisfied. Missense variant in other genic regions have been more rarely detected in PTHS patients: two different de novo variants in exons 14 and 17 were reported (Zweier et al. 2008; De Pontual et al. 2008), and some functional studies were also performed (Sepp et al. 2012). Caution must be used in case other missense variants are identified: even if TCF4 is a gene with a low rate of benign missense variants (PP2 parameter), some are still observed, with low allele frequencies, in general population. Missense variants in the first six exons are less likely to be pathogenic, but also variants in the bHLH domain, different from those already reported, should be dealt with care.