The T2T-CHM13 reference assembly uncovers essential WASH1 and GPRIN2 paralogues

Abstract

Summary

The recently published T2T-CHM13 reference assembly completed the annotation of the final 8% of the human genome. It introduced 1956 genes, close to 100 of which are predicted to be coding because they have a protein coding parent gene. Here, we confirm the coding status and functional relevance of two of these genes, paralogues of WASHC1 and GPRIN2. We find that LOC124908094, one of four novel subtelomeric WASH1 genes uncovered in the new assembly, produces the WASH1 protein that forms part of the vital actin-regulatory WASH complex. Its coding status is supported by abundant proteomics, conservation, and cDNA evidence. It was previously assumed that gene WASHC1 produced the functional WASH1 protein, but new evidence shows that WASHC1 is a human-derived duplication and likely to be one of 12 WASH1 pseudogenes in the human gene set. We also find that the T2T-CHM13 assembly has added a functionally important copy of GPRIN2 to the human gene set. We demonstrate that uniquely mapping peptides from proteomics databases support the novel LOC124900631 rather than the GRCh38 assembly GPRIN2 gene. These new additions to the set of human coding genes underlines the importance of the new T2T-CHM13 assembly.

Availability and implementation

None.

1 Introduction

The publication of the T2T-CHM13 genome assembly by the Telomere-to-Telomere consortium (Nurk et al. 2022) addressed the 8% of the human genome that was missing in the GRCh38 reference. The new gapless assembly added 1956 gene predictions, 140 of which were similar to known coding genes. It is not clear how many of these 140 predicted coding genes can code for proteins. Most are recent duplications, and the consortium did not assess coding potential. Many recent gene duplications are likely to be either pseudogenes or in the process of pseudogenization (Lynch and Conery 2000, Wagner 2001).

The T2T-CHM13 genome assembly uncovered four novel WASHC1 paralogues in newly annotated subtelomeric regions on chromosomes 3, 11, and 20. Nine WASHC1 paralogues were already annotated in the GRCh38 human reference set: two pseudogenes on the p arm of chromosome 1, paralogues on chromosomes 2, 9, 12, 15, 16, and 19, and the pseudoautosomal regions of X and Y. Eight are found in subtelomeric regions, while the ninth, WASH2P, is present in the ancestral telomere-telomere fusion site on chromosome 2 (IJdo et al. 1991).

WASHC1 is thought to be the gene that produces the WASH1 protein, one of the components of the WASH (Wiskott-Aldrich syndrome protein and SCAR Homologue) complex. The WASH complex is one of several that activate the Arp2/3 complex at different subcellular locations (Alekhina et al. 2017). The activated Arp2/3 complex has a crucial role in the generation of branched actin filaments and is important in a range of processes including endocytosis, intracellular trafficking and membrane remodelling. The WASH complex plays a key role in endosomal sorting by recruiting the Arp2/3 complex to induce actin polymerization (Derivery et al. 2009, Gomez and Billadeau 2009). The WASH complex has been shown to be essential in both Drosophila (Linardopoulou et al. 2007) and mice (Xia et al. 2014), and defects in its components lead to inherited developmental and neurological disorders (Valdmanis et al. 2007, Courtland et al. 2021).

The WASH complex is made up of proteins from five genes, WASHC1 (previously WASH1), WASHC2A (FAM21A), WASHC3 (CCDC53), WASHC4 (SWIP), and WASHC5 (KIAA0196). The exact composition of the WASH complex in great apes is complicated by the presence of two highly similar FAM21 genes, WASHC2A and WASHC2C. The two FAM21 genes were generated by a duplication and translocation on chromosome 10 between q11.22 and q11.23, apparently in the ancestor of humans and chimpanzees. In humans, both WASHC2A and WASHC2C have clear peptide support.

In addition, humans, chimpanzees and gorillas have multiple WASH1 genes. In great apes, WASH1 genes are found in rearrangement-prone subtelomeric regions near the ends of chromosomes (Linardopoulou et al. 2007), and gene duplications from interchromosomal duplication are common in subtelomeric regions.

Most of the nine WASH1 genes currently annotated in the GRCh38 assembly are either not full length or have premature stop codons or frameshifts that will likely render them inactive (Linardopoulou et al. 2007). Just one gene, WASHC1 on chromosome 9, is full length and intact. The other eight genes have names that mark them out as pseudogenes, from WASH2P (on chromosome 2) to WASH9P (chromosome 1).

There is disagreement between the three main reference databases as to which WASH1 genes code for proteins. RefSeq (Sayers et al. 2023a) annotates WASHC1 alone as coding, but Ensembl/GENCODE (Frankish et al. 2023, Martin et al. 2023) predicts that both WASHC1 and WASH6P (chromosomes X/Y) are coding. Finally, UniProtKB (The UniProtKB Consortium 2023) lists isoforms for WASHC1, WASH2P, WASH3P (chromosome 15), WASH4P (chromosome 16), and WASH6P. WASH4P and WASH6P lost their 5ʹ ends when duplicated, so do not produce the whole WASH1 protein, WASH2P contains two premature stop codons, and WASH3P has a 4-base frame-shifting deletion and a premature stop codon.

Part of the reason for the disagreements between reference sets is that much of the peptide evidence from proteomics experiments maps to genes other than WASHC1, suggesting that multiple WASH1 paralogues exist. In addition, the WASH1 cDNA (BC048328.1) produced by the Mammalian Gene Collection Program (Strausberg et al. 2002) is quite distinct from the cDNA expected from WASHC1. It is notable that experiments investigating the function of the WASH complex and WASH1 have tended not to use human WASHC1. For example, the initial experiments that discovered the function of the WASH complex used murine WASH1 (Derivery et al. 2009), or the BC048328.1 cDNA product (Linardopoulou et al. 2007, Gomez and Billadeau 2009).

The ancestral WASH1 gene in apes is predicted to be located on the p arm of chromosome 12; probes used in distinct primate species detected WASH1 genes here in macaque, orangutan, gorilla, and chimpanzee, as well as human (Linardopoulou et al. 2007). Multiple WASH1 paralogues have been detected in subtelomeric regions in the gorilla, chimpanzee and human genomes (Linardopoulou et al. 2007). Along with the ancestral gene on chromosome 12 (WASH8P in the GRCh38 human genome assembly), a second WASH1 gene on the p arm of chromosome 20 was also detected in these three species (Linardopoulou et al. 2007). This gene is not present in the GRCh38 assembly.

The T2T-CHM13 genome assembly also uncovered a novel GPRIN2 gene, LOC124900631. By way of contrast to WASHC1, less is known about gene GPRIN2 (G protein-regulated inducer of neurite outgrowth 2). GPRIN2 was found to interact with G-proteins from GNAO1 (Chen et al. 1999) and induced morphological changes to neurites. GPRIN2 has been highlighted as an illustrative example of copy number variation (CNV) in several recent publications (Handsaker et al. 2015, Vollger et al. 2022, Liao et al. 2023). In these papers GPRIN2 was found to be present in two copies in most individuals.

Here, we show that one of the two novel full length WASH1 genes annotated on chromosome 20 as part of the new T2T-CHM13 reference assembly is likely to produce the WASH complex protein, and that the newly annotated GPRIN2 gene is likely to be the protein involved in the regulation of neurite growth.

2 Methods

2.1 Sequence analysis

Predicted protein sequences for WASHC1, WASH2P, WASH3P, WASH4P, and WASH6P were obtained from UniProtKB. The predicted protein sequences of XP_047302880.1 and XP_047302896.1 from WASH1 genes LOC124908094 and LOC124908102 were downloaded from RefSeq.

For the GPRIN2 analyses, the sequence of XP_047301623.1 from LOC124900631 was downloaded from RefSeq, and the principal isoforms from GPRIN2, SYT15, SYT15B, NPY4R, and NPY4R2 were downloaded from APPRIS (Rodriguez et al. 2022).

We downloaded the protein sequence generated from clones BC048328.1 (for WASHC1) and BAA25440.2 (GPRIN2) from GenBank (Sayers et al. 2023b).

The alignment of vertebrate WASHC1 proteins was carried out using UniProtKB WASHC1 protein sequences from 27 vertebrate species plus the human WASH1 protein sequences from in UniProtKB and XP_047302880.1 from LOC124908094. The only requirement for inclusion in the alignment was that the all sequences had to have the ancestral N- and C-terminal sequences and no insertions or deletions of >7 amino acid residues relative to the human WASHC1 sequences. We included sequences from multiple eutherian, therian, bird, reptile, amphibian, and fish species as well as WASHC1 sequences from ghost shark and European lancelet.

2.2 Peptide mapping

We downloaded the peptides detected for all WASH1 and GPRIN2 proteins in the current version (2023–01) of the PeptideAtlas database (Kusebauch et al. 2014). These genes that produce these proteins include WASHC1 and GPRIN2, but also WASH2P, WASH3P, WASH4P, and WASH6P because UniProtKB predicts proteins for these genes. PeptideAtlas also maps peptides to protein sequences from the T2T-CHM13 assembly with RefSeq coding sequences (LOC124908094, LOC124908102, and LOC124900631) and includes peptides that are single amino acid variations of the peptides that map to these proteins.

PeptideAtlas carries out a reanalysis of spectra from multiple large- and small-scale proteomics experiments using state-of-the-art search engines before post-processing with the in-house trans-proteomic pipeline (Deutsch et al. 2023). Each peptide may be detected multiple times over the distinct experiments. PeptideAtlas records the number of distinct “observations” in which a peptide is detected. We used these observations as a rough measure of the expression level of each peptide.

To be sure that we were working with higher quality peptides, we required that all peptides had at least two observations (we removed all singletons), and that each peptide was fully tryptic and had no more than two missed lysine or arginine cleavages.

For the WASH1 analysis, we remapped the peptides recorded in PeptideAtlas to the five WASH1 proteins annotated in UniProtKB, and to the predicted sequence of XP_047302880.1 from gene LOC124908094. This gene is located on chromosome 20 at the terminus of the p arm of chromosome 20. The predicted protein XP_047302880.1 is referred to throughout the paper as WASH1-20p13.

For the GPRIN2 analysis, we mapped the peptides recorded in PeptideAtlas to the GPRIN2 principal isoform ENST00000374314.4 and to XP_047301623.1.

2.3 Phylogenetic analysis

For the WASH1 phylogenetic tree we included the full length WASH1 genes annotated in human, chimpanzee, pygmy chimpanzee (bonobo), gorilla, orangutan, and gibbon. These included six of the nine WASH1 genes annotated in the GRCh38 assembly of the human genome (WASH4P, WASH5P and WASH6P were left out of the analysis because they are missing exons) and the four WASH1 genes added in the T2T-CHM13 assembly. Both chimpanzee (LOC741303, LOC456647) and bonobo (LOC117978042, LOC117975334) are annotated with two WASH1 genes; in both species the transcripts are full length and located on chromosomes 12 and 20. Gorilla is annotated with multiple WASH1 genes, but only one is full length and without stop codons (LOC109025132). Orangutan (WASHC1) and gibbon (WASHC1) are annotated with a single WASH1 gene each.

The DNA sequences of novel WASH1 transcripts XR_007073267.1, XM_047445807.1, XM_047446924.1, and XM_047446940.1 from LOC124906207, LOC124906931, LOC124908094 and LOC124908102 were downloaded from RefSeq along with transcripts with equivalent exon combinations for WASHC1, WASH2P, WASH3P, WASH7P, WASH8P and WASH9P. We added chimpanzee (XM_024353717.2, XM_024347602.2), bonobo (XM_055104758.1, XM_034935049.2), gorilla (XM_031011818.2), orangutan (XM_009237661.3), and gibbon (XM_032159216.2) WASHC1 transcripts. We created the phylogenetic tree of the WASH1 genes from the aligned coding exons and introns of the 17 human and primate WASH1 genes.

The phylogenetic tree was generated with Phyml 3.0 with 100 bootstrap replicates (Guindon et al. 2010) using the TN93+I substitution model according to SMS (Lefort et al. 2017) and running 1000 non-parametric bootstrap replicas. The gibbon WASHC1 sequence was forced to root the tree in the analysis.

3 Results

3.1 Novel WASH1 paralogues are found in subtelomeric regions

All four genes newly annotated in the T2T-CHM13 assembly are located at the subtelomeric regions at the terminal end of chromosomes (Fig. 1A). Subtelomeric regions are prone to rearrangement and interchromosomal duplications affect gene content (Linardopoulou et al. 2005). Two of the novel WASH1 genes would produce full length proteins, both are on chromosome 20.

Figure 1.

Annotated human WASH1 genes and WASH1 gene neighbourhoods. (A) The 13 WASH1 genes in the human genome and their approximate position in the chromosome. Chromosomes are represented by a single chromatid and are not to scale. The approximate position and the direction of the labelled WASH1 gene is indicated with an arrow. Genes represented with red arrows have premature stops and/or frameshift mutations, genes represented by pink arrows cannot produce the full WASH1 protein, genes represented by blue arrows are full length and without high impact mutations. Genes are positioned on the p arm (p), the q arm (q), or the ancestral fusion site (FS). The p arm of chromosome 1 has two WASH1 genes on the same strand. (B) The gene neighbours of WASHC1 on chromosome 9. Exons (blocks) and introns not to scale. Coding exons are wider than non-coding exons. Direction of the strand is shown by the arrows in the intronic regions. The DDX11-like pseudogene is on the left. The FAM138 gene on the right. This block spans the chromosome 9 region from approximately base 12 000 to base 37 500. Most human WASH1 genes have this gene neighbourhood. (C) The gene neighbours of WASH4P on chromosome 16. WASH4P is missing its 5ʹ end and is predicted to translate from an upstream start site in an L2 transposon region. The DDX11-like pseudogene is on the left. The WASIR gene on the right. (D) The gene neighbours of LOC124908094 on chromosome 20. Thje DDX11-like pseudogene is on the left, lncRNA exons on the right.

In total just three WASH1 genes could produce full length proteins. The remainder either have frameshifts or premature stop codons that would truncate the protein (WASH2P, WASH3P, WASH7P, WASH8P, WASH9P, LOC124906207, LOC124906931) or would not produce a complete WASH1 protein (WASH4P, WASH5P, WASH6P). WASH4P and WASH6P have lost the 5ʹ coding exon (this exon codes for 50 amino acids). WASH5P is missing most of its coding exons in the GRCh38 assembly. This region has been completed in the T2T-CHM13 assembly, though exons are still unannotated. The WASH1 genes analysed in this work are shown in Table 1 and in Supplementary Table S1.

Table 1.

Annotated and novel WASH1 and GPRIN2 genes and pseudogenes.^a

Gene symbol	Chr	Assembly	Alias	Predicted
WASHC1	9	GRCh38		Pseudogene
WASH2P	2	GRCh38		Pseudogene
WASH3P	15	GRCh38		Pseudogene
WASH4P	16	GRCh38		Pseudogene
WASH5P	19	GRCh38		Pseudogene
WASH6P	X	GRCh38		Pseudogene
WASH7P	1	GRCh38		Pseudogene
WASH8P	12	GRCh38		Pseudogene
WASH9P	1	GRCh38		Pseudogene
LOC124908094	20	CHM13	WASH1-20p13	Coding
LOC124908102	20	CHM13	WASH1-20q33	Pseudogene
LOC124906207	3	CHM13		Pseudogene
LOC124906931	11	CHM13		Pseudogene
GPRIN2	10	GRCh38	.	Coding
LOC124900631	10	CHM13	GPRIN2L	Coding

^aA list of the 10 GRCh38 and 5 T2T-CHM13 assembly genes mentioned in this paper, along with any aliases used and whether we predict these genes are pseudogenes or coding genes. There is an expanded version of this paper in the supplementary material.

3.2 Annotated WASH1 genes have a similar gene neighbourhood

The 13 WASH1 paralogues are a result of inter-chromosomal duplications in the subtelomeric regions, duplications that involve various genes in a single block. As a result, most WASH1 genes have similar gene neighbours. These duplicated blocks seem to be of different sizes, but even those genes where only part of the WASH1 gene was duplicated (WASH4P, WASH6P) have retained their gene neighbours on one side at least.

Most WASH1 genes are preceded upstream on the same strand by a pseudogene from the FAM138 family (Fig. 1B). FAM138 genes are non-coding in the human gene set but annotated as coding in some primate species. Just downstream of the WASH1 genes, and on the opposite strand, is a DDX11-like gene, a probable pseudogene made up of a short fragment of a dead-box helicase gene.

Eight of the 13 WASH1 genes have the WASH1C gene neighbourhood. The exceptions include WASH4P and WASH6P, which were duplicated without their 5ʹ exons and without the FAM138 family gene beyond the 5ʹ exons. Instead, WASH4P and WASH6P have upstream WASIR genes (Fig. 1C). WASH5P is missing the DDX11-like gene in GRCh38 because of a gap in the assembly, and WASH9P is missing the FAM138 gene neighbour for the same reason. In the T2T-CHM13 assembly, WASH5P does have a neighbouring DDX11-like gene, but WASH5P is not annotated with all its exons. Neither WASH9P nor its neighbours are yet annotated in the T2T-CHM13 assembly. Finally, LOC124908094, on the p arm of chromosome 20, is missing the FAM138 family gene (Fig. 1D).

At present, the reference sets of bonobo and chimpanzee are annotated with two copies of WASH1 genes each. In chimpanzee and bonobo, WASH1 genes are annotated on both chromosome 12 and chromosome 20. In both species, the chromosome 12 WASH1 gene neighbourhood is bounded by a DDX11-like pseudogene and a FAM138 family gene, just like the equivalent human WASH1 gene on chromosome 12 (Fig. 1B). The chimpanzee WASH1 paralogue (LOC741303) on chromosome 20 is adjacent to just the DDX11-like pseudogene. As with LOC124908094 on human chromosome 20, there is no upstream FAM138 family gene (Fig. 1C). However, both LOC741303 and LOC117978042 (the bonobo WASH1 gene on chromosome 20) do have a DEFB125 coding gene a little further upstream, just like the human chromosome 20 paralogue, LOC124908094.

3.3 The WASH1 paralogue on chromosome 12 is the ancestral WASHC1 gene

Curiously, the association between the WASHC1 gene and DDX11-like helicases has ancient roots. Zebrafish has a single wash1 gene, and the full length ddx11 gene is downstream and on the opposite strand. When the ancestral WASHC1 gene ended up in the subtelomeric regions, it seems it only retained part of the DDX11 gene.

The ancestral WASHC1 gene in human is clearly the chromosome 12 WASH1 gene (WASH8P). WASH8P has an antisense IQSEC3 coding gene just upstream, beyond which are SLC6A12, SLC6A13, and KDM5A coding genes on the same strand as WASHC1. This is also true of the chromosome 12 WASHC1 genes in chimpanzee, bonobo, gorilla and orangutan. In fact, the FKBP4-DDX11-WASHC1-IQSEC3-SLC6A12/SLC6A13-KDM5A gene block is contiguous across tetrapods, while IQSEC3 is a neighbour of WASHC1 even in elephant shark.

It appears that a block of genes that included DDX11 and WASH1C was duplicated at some point in the ancestor of the Euarchontoglires clade and this allowed species to rearrange the order of these genes. In primates, a block of 16 coding genes (including WASHC1, IQSEC3, SLC6A12, SLC6A13, and KDM5A) at the end of the chromosome has been reversed with respect to its ancestral direction. The block ended in middle of DDX11, leaving WASHC1 and the DDX11 fragment at the border of the reversed region. WASHC1 and the DDX11 fragment are part of or close to the subtelomeric region in almost all old-world monkeys. In humans the DDX11 coding gene is still in the p arm of chromosome 12 but is now close to the centromere.

3.4 Cross-species conservation supports the importance of LOC124908094 protein WASH1-20p13

UniProtKB annotates five WASH1 proteins. Two of these predicted proteins (from WASH2P and WASH3P) would be truncated because they have premature stop codons and frameshifts relative to the ancestral WASHC1 gene. Two more (from WASH4P and WASH6P) are predicted to initiate their translation from different frames of the same non-conserved L2 transposon region (Fig. 1C) and would be very different from ancestral WASH1 protein in the N-terminal region. In addition, alignments with WASH1 proteins from vertebrate species show that all five UniProtKB WASH1 proteins have multiple single amino acid differences in residues that are conserved across mammals and tetrapods (Supplementary Fig. S1).

However, one WASH1 protein has largely maintained the conserved ancestral WASH1 sequence, the predicted full length WASH1 protein from LOC124908094 on the p arm of chromosome 20, XP_047302880.1. This protein is referred to as “WASH1-20p13” in the remainder of the text. Not only is WASH1-20p13 the only WASH1 isoform that maintains amino acids conserved across vertebrates, but it is also the only isoform without any deletions (Fig. 2A). All WASH1 proteins other than WASH1-20p13 have a three-residue deletion with respect to vertebrate WASH1 sequences.

Figure 2.

WASH1 variants and phylogenetic analysis. (A) The number of non-conserved amino acids (single amino acid variations, SAAVs) and deleted regions in the five full length WASH1 protein isoforms that differ from amino acids that are conserved across primates, mammals and tetrapods. WASH1-20q33 is the predicted protein on the q arm of chromosome 20. WASH1-20p13 is left blank because it only differs in non-conserved regions. (B) Amino acid differences between the WASHC1 protein and WASH1-20p13 in regions conserved across vertebrates. The WASHC1 protein is different in all conserved positions. (C) Phylogenetic tree of great ape and human genes. Genes newly annotated in T2T-CHM13 assembly are labelled with their RefSeq gene names, the likely WASH1 gene, LOC124908094, is highlighted. Great ape WASH1 genes are labelled with the chromosome number in which they are annotated.

Most of the amino acid substitutions in annotated UniProtKB WASH1 proteins are not conservative. For example, none of the 10 single amino acid differences between the WASHC1 principal isoform and WASH1-20p13 that fall in conserved regions are conservative (Taylor 1986). The ten differences between the WASHC1 principal isoform and WASH1-20p13 can be seen in Fig. 2B.

Several amino acid substitutions in WASHC1 might even be considered radical (Taylor 1986). One example is the tyrosine to aspartate swap at residue 141. Tyrosine residue Y141 is phosphorylated by the Src kinase LCK and mutating this residue to phenylalanine has been shown to interfere with WASH-mediated NK cell cytotoxicity (Huang et al. 2016). Since aspartate is a phosphomimetic, the effect of a Y141D mutation may well be permanent phosphorylation, leading to sustained intensification of NK cell cytotoxicity.

We carried out a phylogenetic analysis with ten of the WASH1 paralogues and annotated WASH1 genes in great apes. The analysis confirms that WASHC1, WASH2P, WASH3P, WASH7P, WASH8P, WASH9P, LOC124906207, LOC124906931, and LOC124908102 all derived from a common ancestor because they form a separate outgroup (Fig. 2C). WASH4P and WASH6P were not included in the analysis because of the large 5ʹ region differences, but their common variants and indels suggest that they almost certainly derived from the same common ancestor. WASH5P has too little sequence to include in the analysis. LOC124908094 from chromosome 20 grouped with the primate WASH1 genes.

3.5 Peptide evidence does not support the currently annotated WASHC1 coding gene

To determine which WASH1 genes had translation evidence, we mapped all peptides recorded in PeptideAtlas to the five WASH1 isoforms annotated in UniProtKB, and to the full length WASH1 protein from LOC124908094, WASH1-20p13.

In total, 52 peptides with a total of 23 813 observations in PeptideAtlas map to at least one of the six WASH1 isoforms. The peptides, coloured by the number of observations, are mapped onto the predicted sequences of the six proteins in Fig. 3. None of the WASH1 isoforms captures all 52 of the PeptideAtlas peptides, at first glance suggesting that more than one WASH1 gene might be coding.

Figure 3.

Mapping peptides to WASH1-20p13 and WASH1 sequences in UniProtKB. Peptides are mapped to alignments between the five UniProtKB annotated WASH1 sequences and the newly annotated WASH1-20p13 protein. Residues that peptides map to are colour coded by the number of observations detected for that protein. Dark red, red, and orange fonts indicate the most observed peptides. Residues with a yellow background indicate the position of stop codons and frameshifts in WASH2P and WASH3P. Grey sequence indicates the parts of the predicted sequence of WASH2P and WASH3P that cannot be translated because of these stop codons and frameshifts. Peptides with a light green background and text in bold show those peptides that map uniquely to one of the six sequences. Amino acids columns with a blue background indicate the position of single amino acid differences between the predicted proteins. Deletions relative to ancestral sequences are marked with a pink background.

WASHC1, the only gene annotated as coding in all three main human reference databases, captures surprisingly few peptides. Just half of the 52 PeptideAtlas peptides map to the WASHC1 isoform. Ensembl/GENCODE annotates WASH6P and WASHC1 as coding, but WASH6P and WASHC1 between them capture only 34 PeptideAtlas peptides. This leaves more than a third of PeptideAtlas WASH1 peptides unaccounted for.

Extending the mapping to predicted UniProtKB proteins from WASH2P, WASH3P, and WASH4P would cover most of the 52 WASH1 peptides in PeptideAtlas, if the predicted sequences of WASH2P and WASH3P in UniProtKB were correct. Unfortunately, WASH3P has a frameshift that would eliminate the final two thirds of the protein, and WASH2P has two premature stop codons, the first of which would leave a rump of just 25 amino acid residues. These two genes cannot produce the predicted proteins.

Of the two full length WASH1 genes identified in the T2T-CHM13 assembly on chromosome 20, the protein from the WASH1 paralogue on the q arm has little support; just 18 of the 52 PeptideAtlas peptides mapped to this protein. By way of contrast, the LOC124908094 isoform, WASH1-20p13, captures 47 of the 52 PeptideAtlas peptides. In fact, 17 peptides map to WASH1-20p13 and to no other paralogue. Two of these peptides are among the most highly observed, ATSQGGDLMSDLFNK, with >2000 observations, and MQHSLAGQTYAVPLIQPDLR, with 1300 observations. The 17 peptides that map uniquely to WASH1-20p13 strongly suggest that LOC124908094 produces the WASH complex protein. In PeptideAtlas most observations are from a range of cell lines or cancer cells. However, peptides unique to WASH1-20p13 were detected in all major tissue types, as would be expected from a protein that is ubiquitously expressed (Uhlén et al. 2015).

3.6 G343S is a common variant in WASH1-20p13

Five of the 52 WASH1 peptides in PeptideAtlas do not map to WASH1-20p13. There are three possible explanations for these unmapped peptide spectrum matches (PSMs). Either WASH1-20p13 has one or more common amino acid variants, two (or more) WASH1 genes are translated, or the PSMs are false positive matches. No single WASH1 isoform captures all five of these peptides. Four of the 5 peptides have 4 or fewer observations, but one peptide, QDDSSSSASPSVQGAPR, was detected 1239 times in experiments across PeptideAtlas. A similar peptide, QDDGSSSASPSVQGAPR, which does map to WASH1-20p13, was observed 1999 times.

It is curious that peptide QDDSSSSASPSVQGAPR has so many observations when none of the other four peptides that do not map to WASH1-20p13 have >4. One possibility is that there is a common variant in WASH1-20p13 that changes glycine residue 343 into a serine. It is noticeable that the only difference between WASH1-20p13 and the protein produced by the BC048328.1 cDNA clone is the same glycine to serine substitution.

Here, the annotations from the Human Pangenome Reference Consortium (Liao et al. 2023) were particularly useful. The Human Pangenome Reference Consortium annotation is based on the T2T-CHM13 assembly, so individuals are annotated with the T2T-CHM13 WASH1 genes as well as those only present in the GRCh38 assembly. Individuals are not annotated with all 13 paralogues. We were able to retrieve 33 predicted protein sequences that were most similar to the WASH1-20p13 (those sequences with the ancestral “APP” insertion). Of these, 17 have a glycine at position 343 and 16 have a serine. G343S is quite clearly a common WASH1-20p13 variant. A second variant, S119P, that explains another of the undetected peptides is also present though less common in these 33 sequences. Between them, these two variants explain three of the peptides that do not map to WASH1-20p13. So, all but two of the 52 PeptideAtlas peptides, and 23 808 of the 23 813 peptide observations (99.98%), map to WASH1-20p13 and its common variants.

The remaining two peptides, with a total of five observations, map to WASH6P. Only one also maps to the previously predicted WASH1 gene, WASHC1. These peptides were detected in cancer cells or cell lines, so there is no evidence that WASH6P (if translated) is tissue specific. It seems unlikely that these peptides are from WASH6P though, because WASH6P has a large novel N-terminal for which we find no supporting peptides at all. Of the two peptides, one (with just two observations and unconvincing peptide spectrum matches) is likely to be a false positive match. The other, which has three observations, is a single amino acid variant of the WASH1-20p13 peptide ATSQGGDLMSDLFNK which has 2024 observations.

3.7 LOC124908094 is likely to be the only WASH1 coding gene

There are 13 WASH1 genes annotated in the T2T-CHM13 assembly. WASH8P, the ancestral copy of WASH1 on chromosome 12, is no longer functional because it would generate a transcript with a premature stop codon. This stop codon is present in another five human WASH1 pseudogenes, WASH3P, WASH9P, WASH7P, LOC124906931, the copy of WASH1 on chromosome 11, and LOC124906207, the copy of WASH1 on chromosome 3. In addition, WASH2P has multiple disabling mutations and WASH5P, as annotated, is just a fragment of the whole gene. These eight WASH1 paralogues are clearly pseudogenes.

There are clues that all paralogues apart from LOC124908094 on chromosome 20 derived from the WASH8P pseudogene. FAM138 pseudogenes are found upstream of the chromosome 12 WASH1 genes in both human and chimpanzee, but not upstream of the chromosome 20 WASH1 genes. Instead, chromosome 20 WASH1 genes have an upstream DEFB125 gene. Nine of the 12 human WASH1 paralogues of LOC124908094 have upstream FAM138 family pseudogenes. Two of the paralogues without a FAM138-like pseudogene at the 5ʹ end are WASH4P and WASH6P, but these two genes have also lost their 5ʹ coding exons. The other is WASH9P, where the position of the FAM138 gene coincides with a gap in the GRCh38 assembly.

In order to generate proteins for the WASH1 paralogues without peptide support, we would have to ignore frameshifts and stop codons. However, if it were possible to produce proteins from these genes, 11 of the 12 would produce proteins with a 3 amino acid deletion at the start of the proline-rich region and have four single amino acid differences in common, R54Q, Y141D/E, V386M, and D410H. The odd one out, WASH5P, would produce a highly truncated protein so was not analysed. None of these five radical changes are present in WASH1 proteins of any other species, not even gorilla or chimpanzee, reinforcing the suggestion that these WASH1 paralogues arose in the human lineage.

Finally, the phylogenetic analysis of the ancestral WASH1 genes confirms that human WASH1 paralogues apart from LOC124908094 have a common ancestor because they form an outgroup separate from the primate WASH1 genes and LOC124908094. While the phylogenetic analysis does not wholly confirm the theory that these 12 WASH1 genes arose in the human lineage, the alternative possibility, that they appeared earlier and have either been lost from other primate species or have not yet been annotated, is highly unlikely.

One curious result from the analysis is that the bonobo (pygmy chimpanzee) and chimpanzee WASH1 genes from chromosomes 12 and 20 form their own outgroup. Alignments between the annotated WASH1 isoforms in chimpanzee and bonobo (Supplementary Fig. S2) reveal few amino acid residue differences. It is possible that chromosome 12 and 20 WASH1 paralogues arose independently in humans and chimpanzees, even though human and chimpanzee WASH1 gene neighbourhoods are similar for both genes. Although this muddies the water regarding WASH1 evolution in subtelomeric regions in primates, it does not affect the evolutionary history of the human WASH1 paralogues.

3.8 The T2T assembly uncovers a GPRIN2 paralogue

The T2T assembly has added >3600 predicted genes and 140 of these genes were found to be most similar to coding genes. Yet similarity to a coding gene is not a guarantee that the duplication itself is coding. These genes have yet to undergo manual curation, and most duplications tend to lose coding ability over time and become pseudogenes (Lynch and Conery 2000, Wagner 2001). No fewer than 57 of the 140 potential coding genes come from just 3 parents—TAF11L5, USP17L11, and USP17L27. We found that at least one of these potential coding genes is coding.

LOC124900631 is a duplication of gene GPRIN2 (G protein-regulated inducer of neurite outgrowth), a gene annotated on the reverse strand of region q11.22 on chromosome 10 (Deloukas et al. 2004). The nearest downstream coding gene is NPY4R (Neuropeptide Y receptor type 4), the nearest upstream gene is SYT15 (Synaptotagmin-15). This block is one of four blocks of genes in the proximal end of the q arm of chromosome 10 to have undergone recent duplication and translocation (Fig. 4A). A region of approximately 0.5MB at the end of q11.22 has been duplicated and inserted back into q11.22, but in the reverse direction. Curiously, the duplicated WASH2 genes are generated from one of the other three duplicated blocks (Fig. 4A). In GRCh38, this segmental duplication region was well characterized (Deloukas et al. 2004, Vollger et al. 2022) except for a short, gapped region between genes SYT15B and NPY4R2. The T2T-CHM13 assembly adds one gene in this missing region, LOC124900631. Unsurprisingly, LOC124900631 is a close paralogue of GPRIN2. This novel T2T-CHM13 GPRIN2 gene is referred to as GPRIN2L in this paper.

Figure 4.

Segmental duplications on chromosome 10 and GPRIN2 peptides. (A) Duplication and translocation in the q arm of chromosome 10. At least four gene blocks have duplicated in this region since the last common ancestor of humans and chimpanzees. Genes are shown as arrows indicating direction of strand, coding genes are orange arrows, pseudogenes pink. The GPRIN2L paralogue uncovered by the T2T-CHM13 assembly is shown in light blue. The four gene blocks that have duplicated are colour coded. The light green background blocks include the WASH2 genes, the light blue background the GPRIN2 paralogues. The AGAP-FAM25 block (orange background) appears to have duplicated three times with multiple genes in this region alone. Genes in the yellow background block are all pseudogenes. Several of the gene blocks are contiguous. There is no gap between the gene blocks when the blocks are contiguous. The approximate coordinates of the contiguous blocks are indicated above the blocks. (B) Alignment of the two human GPRIN2 proteins. The differences between the sequences are highlighted in orange. Peptides detected in PeptideAtlas are mapped onto the sequences. Red peptides have >100 observations, orange >20 observations, yellow >10, green >5, and blue >2. The newly annotated gene, GPRIN2L (LOC124900631), is supported by six unique peptides with a total of 43 observations.

Although GPRIN2 has been used as an illustrative example of copy number variation (CNV) in several recent publications (Handsaker et al. 2015, Vollger et al. 2022, Liao et al. 2023), this is an error caused by a gap in the GRCh38 assembly. The T2T-CHM13 assembly shows that this extra copy of the GPRIN2 gene had been missing from the GRCh38 assembly. The detection of a GPRIN2 paralogue (GPRIN2L) is all the more interesting because we find that the new gene is likely to be the functionally important paralogue, as was the case with WASH1 gene LOC124908094.

The protein sequences of the APPRIS principal isoforms (Rodriguez et al. 2022) of SYT15 and SYT15B, and NPY4R and NPY4R2, are identical. However, analysis of the APPRIS principal isoforms of GPRIN2 and GPRIN2L (ENSP00000363433.4 and XP_047301623.1) finds eight single amino acid differences between the two proteins as well as a single insertion of three amino acids in the GPRIN2L protein (Fig. 4B). From the sequence evidence alone, it is not clear which sequence is closer to the ancestral gene. ENSP00000363433.4 (GPRIN2) has four single amino acid differences with respect to conserved mammalian GPRIN2 sequences, while XP_047301623.1 (GPRIN2L) differs by three amino acids and a 3-residue insert.

However, the two genes are distinguished by the peptide evidence. ENSP00000363433.4 has no unique peptide support in PeptideAtlas, while the GPRIN2L protein is supported by six unique peptides with a total of 43 observations (Fig. 4B). The GPRIN2L protein, including the insertion, is also supported by a cDNA sequence (Nagase et al. 1998), a clone derived from a brain cDNA library. These results suggest that the T2T-CHM13 paralogue, GPRIN2L, has more functional relevance, although in this case there is nothing to suggest that GPRIN2 is not also protein coding.

4 Discussion

There are many novel duplicated genes in the T2T-CHM13 human genome reference, and more than 100 have a parent gene that it is coding. However, the functional relevance of most of these duplicated genes is still unclear. The majority appear to be recent duplications, and many are likely to be pseudogenes, or at least to be in the process of pseudogenization (Lynch and Conery 2000, Wagner 2001). Here, we find that at least two of the newly uncovered duplicated genes, paralogues of WASHC1 and GPRIN2, are clearly functionally important.

We have shown that one of the four novel WASH1 genes added as part of the T2T-CHM13 assembly, LOC124908094, produces the WASH1 protein that forms an integral part of the WASH complex, a vital complex involved in the regulation of branched actin in endosomes. Previously, the WASH1 protein was thought to be coded by WASHC1.

The protein produced by LOC124908094, WASH1-20p13, captures almost all the known peptide evidence. Seventeen of the peptides detected for WASH1 map exclusively to WASH1-20p13. WASH1-20p13 is also the human WASH1 protein that is most similar to the ancestral WASHC1 proteins, and only differs from the predicted protein sequence produced by the GenBank WASHC1 cDNA by one amino acid residue. We have shown that this change, G343S, is a common variant (Liao et al. 2023).

The other 12 WASH1 paralogues either have premature stop codons or produce a series of radical amino acid changes in conserved residues. Most are already annotated as pseudogenes. WASHC1 on chromosome 9 is annotated as coding and produces a full-length protein. However, even though WASHC1 does not have a disabling mutation, it would have multiple radical amino acid changes relative to WASH1-20p13 in regions that are highly conserved across vertebrates. The WASHC1 protein is not uniquely identified by any peptide in PeptideAtlas.

Phylogenetic analysis, cross species conservation and gene neighbourhoods all suggest that these 12 WASH1 paralogues appeared in the human lineage. Alignments suggest that the gene duplications occurred after the ancestral paralogue gene, WASH8P on chromosome 12 (Linardopoulou et al. 2007), had already undergone substantial changes. If 12 WASH1 paralogues derived from a gene that was in the process of becoming a pseudogene, they too are likely to be pseudogenes.

Currently, all functional information for the WASH1 protein links to WASHC1. It will be important to transfer all the functional information related to the WASH1 protein from WASHC1 to LOC124908094, especially since WASHC1 is likely to be a pseudogene. It is known that the WASH complex has a role in developmental and neurodegenerative disorders (Valdmanis et al. 2007, Courtland et al. 2021), so annotating the correct WASH1 coding gene may allow the discovery of previously hidden disease-relevant variants.

As part of our analysis, we were also able to show that a novel GPRIN2 paralogue, GPRIN2L, has uniquely mapped peptides in PeptideAtlas and is likely to have more functional relevance than its paralogue from the GRCh38 assembly, GPRIN2. Annotating the correct GPRIN2 may help elucidate its role in neurite development.

Uncovering both the WASH1 coding gene and an important novel GPRIN2 paralogue in the novel T2T-CHM13 assembly demonstrates the importance of annotating the missing regions in the human genome and suggests that there might be other interesting coding genes yet to be discovered in the novel regions annotated by the T2T consortium. The T2T-CHM13 assembly is a more complete and accurate reference for variant calling than the GRCh38 assembly (Aganezov et al. 2022). The discovery of functionally relevant coding genes in the T2T-CHM13 assembly is further confirmation of the importance of this reference set for evolutionary and biomedical research.

Author contributions

Daniel Cerdán-Vélez (Data curation [supporting], Resources [equal], Writing—review & editing [supporting]), and Michael Liam Tress (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Funding acquisition [lead], Investigation [lead], Methodology [lead], Project administration [lead], Resources [equal], Supervision [lead], Validation [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [lead])

Supplementary data

Supplementary data are available at Bioinformatics Advances online.

Conflict of interest

None declared.

Funding

This work was supported by the National Human Genome Research Institute of the National Institutes of Health [U41 HG007234].

Data availability

No new data or software were generated or analysed in support of this research.