The Simple Chordate Ciona intestinalis Has a Reduced Complement of Genes Associated with Fanconi Anemia

Abstract

Fanconi anemia (FA) is a human genetic disease characterized by congenital defects, bone marrow failure, and increased cancer risk. FA is associated with mutation in one of 24 genes. The protein products of these genes function cooperatively in the FA pathway to orchestrate the repair of DNA interstrand cross-links. Few model organisms exist for the study of FA. Seeking a model organism with a simpler version of the FA pathway, we searched the genome of the simple chordate Ciona intestinalis for homologs of the human FA-associated proteins. BLAST searches, sequence alignments, hydropathy comparisons, maximum likelihood phylogenetic analysis, and structural modeling were used to infer the likelihood of homology between C. intestinalis and human FA proteins. Our analysis indicates that C. intestinalis indeed has a simpler and potentially functional FA pathway. The C. intestinalis genome was searched for candidates for homology to 24 human FA and FA-associated proteins. Support was found for the existence of homologs for 13 of these 24 human genes in C. intestinalis. Members of each of the three commonly recognized FA gene functional groups were found. In group I, we identified homologs of FANCE, FANCL, FANCM, and UBE2T/FANCT. Both members of group II, FANCD2 and FANCI, have homologs in C. intestinalis. In group III, we found evidence for homologs of FANCJ, FANCO, FANCQ/ERCC4, FANCR/RAD51, and FANCS/BRCA1, as well as the FA-associated proteins ERCC1 and FAN1. Evidence was very weak for the existence of homologs in C. intestinalis for any other recognized FA genes. This work supports the notion that C. intestinalis, as a close relative of vertebrates, but having a much reduced complement of FA genes, offers a means of studying the function of certain FA proteins in a simpler pathway than that of vertebrate cells.

Introduction

DNA repair mechanisms are a major way by which organisms avoid mutations that can lead to disease, especially cancer. However, the complexity of DNA repair pathways has hindered progress in fully understanding how they work. We have examined the genome of the simple chordate animal, Ciona intestinalis, which is the closest invertebrate relative of vertebrates, for genes associated with the repair of DNA interstrand cross-links (ICL repair), to see if it might possess a simplified version of this DNA repair mechanism. Fanconi anemia (FA) is clinically characterized by congenital abnormalities, pediatric bone marrow failure, and increased cancer risk during early adulthood. FA is caused by mutation of one of the 19 genes linked in a complex pathway. The proteins encoded by the FA genes function together in the process of ICL repair and in the maintenance of genome stability.^1–3 ICLs are highly toxic lesions that covalently link DNA strands, thereby imposing a direct physical block to DNA replication and RNA transcription. The FA protein interaction network is extensive and includes numerous other proteins that function in ICL repair, which have not been genetically linked to FA.⁴

The FA pathway proteins have been categorized into three distinct groups³: group I represents the FA core complex and comprises FANCA, FANCB, FANCC, FANCE, FANCF, FANCG, FANCL, FANCM, and FANCT/UBE2T. The FA core complex catalyzes the site-specific monoubiquitination of the FANCD2 and FANCI (group II) proteins.^5–7 FANCL is a RING domain containing E3 ubiquitin ligase,^8,9 while UBE2T is an E2 ubiquitin-conjugating enzyme.¹⁰ FANCM is a large (230 kDa) scaffold protein that possesses DNA binding and ATPase/translocase activities.^11,12 The functions of the remaining group I proteins remain poorly understood. The group II proteins FANCD2 and FANCI, when monoubiquitinated, facilitate the recruitment of several key DNA repair proteins, including FAN1, FANCP/SLX4, and CtIP, to the ICL.^13–18 The group III FA proteins comprise FANCD1/BRCA2, FANCJ/BRIP1, FANCN/PALB2, FANCO/RAD51C, FANCP/SLX4, FANCQ/ERCC4, FANCR/RAD51, and FANCS/BRCA1 and function downstream of FANCD2 and FANCI monoubiquitination. These proteins function primarily in the homologous recombination (HR) step of ICL repair. For example, FANCD1/BRCA2, FANCN/PALB2, and FANCO/RAD51C regulate the localization and activity of FANCR/RAD51, a well established and key HR protein.^19–25 Several of the FA proteins are ubiquitous among the eukaryotes.²⁶ Almost every organism surveyed possesses both of the group II proteins, as well as FANCL, FANCM, and an associated ubiquitin-conjugating (E2) enzyme (Fig. 1). There is no apparent evolutionary pattern associated with the presence or absence of the group I proteins outside of the vertebrates, as some are found in insects, while others are seen in plants and red algae before seemingly reappearing in Nematostella and then again in the vertebrates. Echinoderms, a sister group of the chordates, possess at least four of the group I proteins.

Figure 1.

Presence/absence of FA gene orthologs in selected eukaryotes, as determined by this study. Filled boxes denote estimated presence of a gene in that taxon. Outside of Ciona and humans, presence/absence was determined only by a Delta-BLAST search of the NCBI database using the human gene as query. The dendrogram at the top of the figure denotes the relationships between organisms.

C. intestinalis is a tunicate, the group thought to be the closest invertebrate relative of the vertebrates.²⁷ C. intestinalis has a number of characteristics that make it a promising model for human diseases. Its genome is very compact, at only 115 Mb, fully sequenced, most of which has been mapped to chromosomes. The current genebuild on Ensembl has 16,671 coding genes, as compared with 20,313 in humans.²⁸ Homologs of almost all human gene families are represented, but Ciona does not have the duplicate genes created by the genome duplications that occurred in vertebrates.²⁹ There are curated databases with abundant gene expression data,^30,31 as well as a proteome database.³² While in many cases Ciona has lost genes reflecting adaptation to its sessile lifestyle,³³ it can still be used to model simplified pathways,^34–36 as it possesses a simplified version of the vertebrate body plan, most notably as a larva.³⁷

A previous study focusing on zebrafish³⁸ looked into the Ciona FA pathway and was unable to find most of the genes. The genes that were found were concentrated in groups II and III, making it plausible that Ciona could at the very least be used as a model for the latter two-thirds of the pathway. A subset of the vertebrate group I proteins do appear to be present in Ciona, according to our study, suggesting that it may possess a minimal FA pathway.

In order to better assess the total complement of FA-associated genes in C. intestinalis, we have analyzed the protein structure, hydrophobicity, and phylogenetic relationships of candidates for each of the FA genes of vertebrates. These analyses indicate that C. intestinalis has both of the group II genes from vertebrates, as expected, but only one-third of the group I and two-thirds of the group III genes. In comparison with other animals, and even the plant Arabidopsis, C. intestinalis appears to have an extremely depauperate FA pathway. These data suggest that C. intestinalis may be a good model organism to study a simplified FA pathway and gain important insight into the poorly understood molecular basis of the developmental defects of FA patients.

Materials and Methods

Obtaining sequences

First, a Reciprocal Best BLAST (RBB)³⁹ search on 24 gene products was performed, searching the human genes of the FA pathway (Table 1) against the Ciona proteome, taking the closest match, and then searching the Ciona protein back against the human database to see if the same protein was returned as the closest result. This step was augmented with a search by the reciprocal smallest distance (RSD) method,⁴⁰ which in all but three cases returned the same protein as RBB. In these three cases the RSD candidate had a higher percentage of positive matches, so those proteins were the ones listed in Table 1.

Table 1.

BLAST (RBB/RSD) results. Refer to Supplemental Table S1 for additional accession numbers of sequences used in phylogenetic and other analyses.

GROUP	HUMAN GENE NAME	HUMAN ACC. NO. BLASTED	BEST CIONA GENE MATCH	E-VALUE HS→CI	CIONA ACCESSION NO.	RECIPROCAL HUMAN GENE MATCH	E-VALUE CI→HS	RECIPROCAL HUMAN ACC. NO.
I	FANCA	NP_000126	Trafficking protein particle complex subunit 10	1 × 10−122	XP_009858877	Trafficking protein particle complex subunit 10	4 × 10−174	NP_003265
	FANCB	NP_001018123	Lysine demethylase/histidyl hydroxylase MINA	0.076	XP_002131324	Lysine demethylase/histidyl hydroxylase MINA	6 × 10−73	Q8IUF8
	FANCC	NP_000127	Stabilin 2	5.2	XP_002122507	Stabilin 2	0.0	CAC82105
	FANCE	NP_068741	FA Group E protein C-term. domain, LOC100186252	6 × 10−16	XP_002129936	Fanconi Anemia Group E protein	7 × 10−65	NP_068741.1
	FANCF	NP_073562	Peroxidasin-like	4 × 10−146	XP_009859106	Peroxidasin	<1.7 × 10−308	NP_036425
	FANCG	NP_004620	Stress induced phosphoprotein 1	3 × 10−78	XP_002128875	Stress induced phosphoprotein 1	3 × 10−77	NP_006810
	FAAP20	NP_001139782	Polyamine-mod. fact. 1–1 (CiPMF1–1)	12	XP_002127003	Serine/Threonine Protein Kinase MRCK	2 × 10−17	NP_003598
	FAAP24	NP_689479	DNA polymerase β	1 × 10−145	XP_002128462	DNA polymerase β	4 × 10−145	NP_002681
	FAAP100	NP_079437	L-fucose Kinase	0.0002	XP_002122353	L-fucose Kinase	<1.7 × 10−308	NP_659496
	FANCL	AAH09042	Ubiquitin ligase Fanconi Anemia group L	2 × 10−74	XP_009861960	Ubiquitin ligase Fanconi Anemia group L	2 × 10−81	Q9 NW38.2
	FANCM	NP_065988	FANCM protein	5 × 10−156	XP_009862357	FANCM protein	<1.7 × 10−308	AAI44512.1
	UBE2T	NP_054895	UBE2–17kDa	2 × 10−71	XP_002129339	UBE2D4	2 × 10−77	XP_006715797
II	FANCD2	NP_149075	Fanconi Anemia Complement. Grp D2	<1.7 × 10−308	XP_002130241	Fanconi Anemia Complementation Group D2	<1.7 × 10−308	AAL05980.1
	FANCI	NP_060663	Fanconi Anemia Complement. Grp I	<1.7 × 10−308	XP_009858757	Fanconi Anemia Complementation Grp. I	0.0	ABQ63084.1
III	FANCD1/BRCA2	P51587	Unchar. prot. LOC100185089	6 × 10−12	XP_002129592	BRCA2	6 × 10−13	AAB07223
	FANCJ	NP_114432	CiTFIIH-X	5 × 10−157	XP_002126055a	CiTFIIH-X	0.0	NP_000391
	FANCN	Q86YC2	WD repeat containing protein-5 like	2 × 10−94	XP_002127700	WD repeat containing protein-5 like	7 × 10−116	P61964
	FANCO/RAD51C	AAC39604	RAD51 homolog 1 isoform 1	5 × 10−118	XP_002126934b	RAD51 homolog 1 isoform 1	2 × 10−93	NP_002866.2
	FANCP	NP_115820	Kelch-like protein 10	<1 × 10−27	XP_002122519	Kelch-like protein 20	0.0	NP_055273.2
	FANCQ/XPF	Q92889	DNA repair protein XPF	3 × 10−72	XP_009859502	DNA repair protein XPF	2 × 10−117	AAB50174
	FANCR/RAD51	CAG38796	RAD51 homolog 1 isoform 1	7 × 10−140	XP_002130341	RAD51 homolog 1 isoform 1	<1.7 × 10−308	NP_002866.2
	FANCS/BRCA1	NP_009225	BRCA1 putative homolog	1 × 10−33	KH2012: KH.C9.487c	breast cancer 1, early onset, isoform CRA_g	8 × 10−29	EAW60929
	ERCC1	P07992	ERCC1-like	9 × 10−41	XP_009861832	ERCC1	2 × 10−70	NP_001974
	FAN1	NP_055782	Fanconi Associated Nuclease 1 C-terminal domain	9 × 10−41	XP_004227197	Fanconi Associated Nuclease 1 C-terminal domain	1 × 10−101	NP_055782

Notes:

^aSubsequent analysis led to the identification of another sequence, Ciona Fanconi anemia group J protein homolog, NCBI acc. no. XP_002120239 as a better proposed ortholog of human FANCJ (see text).

^bSubsequent analysis led to the identification of another sequence, C. intestinalis DNA repair protein RAD51 homolog 3-like, NCBI acc. no. XP_002130341 as a better proposed ortholog of human FANCO (see text).

^cA longer gene model was found in the ANISEED database, to which this accession number corresponds.

BLAT⁴¹ in the JGI genome portal⁴² as well as OrthoDB⁴³ was used to look for synteny between human and Ciona FA genes, but none was detected for any of the candidates.

Protein information

Using ClustalX and ClustalΩ,⁴⁴ each Ciona FA protein sequence was aligned against the human and Xenopus laevis sequence. The sequences were imported into Jalview,⁴⁵ and the most closely aligned regions were isolated. Hydrophobicity plots of each sequence were created using Biopython and code built and modified from Dalke Scientific.⁴⁶ To determine whether the results were significant, the Pearson coefficients were evaluated for the Ciona amino acid (aa) sequence against the human and Xenopus sequences (again using Python), a beta distribution derived for each sequence,47 and a comparison of the critical values to a P < 0.002 level of significance was made. As a standard, P < 0.05 level of significance with 24 tests gives about a 30% chance of a false positive (Type I error), so a more thorough bound of significance was required. The Sidak test,⁴⁸ a familywise error correction method used to reduce type I errors, suggests a P-value of 1 – (1 − 0.05)^1/24, or about 0.0021, where 0.05 is the original level of significance and 24 is the number of comparison tests performed. This assumes that the genes and their products are independent – there does not appear to be any evidence that a mutation in one FA protein leads to the absence of any of the other FA proteins.

Protein structural models (Figs. 2E, F, J, and K and 4F and G) were constructed using Discovery Studio v. 3.1 (BIO-VIA), based on pdb files in the RCSB Protein Data Bank, using 50 iterations with loop refinement. The protein motif diagrams were based on the information in Pfam 29.0.⁴⁹

Figure 2.

Analysis of FANCE (A–F) and FANCL (G–K) putative homologs in C. intestinalis. (A) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCE. (B) Best ML tree for alignment of FANCE and putative homologs in C. intestinalis and other eukaryotes. CiUP1 (LOC100186252) has 93% bootstrap support for membership in the clade with vertebrate FANCE proteins. (C) Forcing CiUP1 into the vertebrate FANCE clade does not result in a statistically worse tree, whereas if the locations of the two best C. intestinalis BLAST matches to FANCE are switched in the ML tree (D), the tree is worse at the P < 0.01 level, giving further support to LOC100186252 as the homolog of FANCE. (E,F) Structural modeling of human FANCE and C. intestinalis LOC100186252, showing extreme similarity of overall structures. (G) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCL. (H) Best ML tree for alignment of putative FANCL homologs, showing 87% bootstrap support for Cifancl clustering with vertebrate and other FANCL proteins. (I) Diagrammatic comparison of human and C. intestinalis FANCL inferred protein motifs. (J,K) Modeling of D. melanogaster and C. intestinalis FANCL protein structures.

Figure 4.

Analysis of FANCD2 (A–G) and FANCI (H–K) putative homologs in C. intestinalis. (A) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCD2. (B) Best ML tree for alignment of FANCE and putative homologs in C. intestinalis and other eukaryotes. Cifancd2 groups closely with FANCD2, but with low bootstrap support. (C) If Cifancd2 is moved out of the FANCD2 clade, the tree is statistically worse at the P < 0.02 level, supporting the case for Cifancd2 as a true homolog of FANCD2. (D) Alignment of human, mouse, and C. intestinalis FANCD2 protein sequences showing conservation of L215, P216, L234, and L235, critical residues of the CUE domain (red boxes).⁶³ (E) Alignment showing partial conservation of critical residues around human aa 525 (arrows, and box), as well as K561, the site of monoubiquitination⁵ (red arrowhead) in C. intestinalis. (F, G) Modeling of mouse and C. intestinalis FANCD2 homolog protein structures, respectively. (H) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCI. (I) Best ML tree for alignment of putative FANCI homologs, showing weak bootstrap support for Cifancl being more closely related to FANCI than the next most similar C. intestinalis protein. (J) Alignment of human, mouse, and C. intestinalis FANCI protein sequences showing the conservation of K523 and K715, the site of FANCI monoubiquitination,⁶,⁷ and, the site of FANCI SUMOylation, respectively (arrowheads). (K) SQ/TQ phosphosite clusters (red boxes) shown to be critical for FANCI function.⁶⁸ Dashed boxes denote possible additional functional SQ/TQ phosphosite clusters in C. intestinalis sequence not found in humans and mouse.

Phylogenetic analysis

Full protein sequences (see Supple mentary Table S1 for accession numbers) were aligned using MAFFT with default settings.⁵⁰ Poorly aligned regions were excised using TrimAI v. 1.3 using the Gappyout setting on the Phylemon 2.0 web server.⁵¹ RAxML v. 8.0.0⁵² was used to construct a maximum likelihood (ML) tree with bootstrap number determined with the FC bootstrapping criterion and PROTGAMMABLOSUM62 substitution model. User supplied trees with candidate genes rearranged were statistically evaluated using the Shimodaira–Hasegawa (SH) log likelihood test in RAxML.

Results

Ciona has orthologs of vertebrate FA genes from each functional group

Our analysis revealed that Ciona has highly conserved orthologs of genes from each of the three FA protein groups (Fig. 1). Like all the other multicellular organisms examined, Ciona has both members of group II: FANCD2 and FANCI. However, only 4 of 9 members of group I and 5 of 8 members of group III were found, as well as only 2 of several “FA associated” proteins. In fact, Ciona appears to have as few or fewer members of the FA pathway of any multicellular organism examined, including plants, slime mold, and the primitive metazoan Nematostella.

Below we present evidence for or against orthology in C. intestinalis of each of the members of the FA pathway. The first analyses described are for those genes that we estimate are present in Ciona, organized by the functional group. We then list those that do not have orthologs in Ciona according to our methods. The order of the genes in the text is similar to the vertical order in Figure 1.

Group I orthologs found

FANCE

FANCE is part of the FA core complex with an unknown function. RBB returns an uncharacterized C. intestinalis protein LOC100186252 (XP_002129936). The Ciona candidate protein aligns well with the last 250–300 aa of vertebrate FANCE proteins (R² = 0.202), but on the whole, the correlation is only 0.08 (and the region outside the C-terminal registering at only 0.05; Fig. 2A). The Ciona candidate is about 400 aa in length, while vertebrate FANCE proteins are all between 550 and 600 aa. Moderate alignment is seen between the two globular domains in the Ciona candidate and the two C-terminal globular regions in the human protein, though no other shared secondary structure is found in the ELM analysis (data not shown). The ML best tree (Fig. 2B) groups the Ciona candidate LOC100186252 (“CiUP1”) in a sister group to the vertebrate FANCA proteins, more closely related to the plant and fungal candidates. However, if LOC100186252 is forced to group with the vertebrate FANCE proteins (Fig. 2C), the tree is not significantly worse, while moving LOC100186252 more distant from the FANCE clade is statistically worse (Fig. 2D; P < 0.01), consistent with the orthology of FANCE. In addition, a crystal structure exists for human FANCE,⁵³ allowing us to perform structural homology modeling between the human protein and the inferred C. intestinalis protein (Fig. 2E and F). The 3D models indicate that the structure of LOC100186252 is potentially very similar to human FANCE. Taken as a whole, these data provide support for LOC100186252 being the homolog of FANCE in C. intestinalis.

FANCL

FANCL is an E3 ubiquitin ligase and a component of the FA core complex, which serves to ubiquit-inate FANCD2 and FANCI.⁹ RBB returns a putative Ciona FANCL protein with an E-value of 2 × 10⁻⁷⁴ (Table 1). The Ciona candidate hydrophobicity plot shows close correspondence to the vertebrate proteins (Fig. 2G). SMART and Pfam primary sequence-based prediction analyses both detect three amino-terminal WD40 repeats and a carboxy-terminal RING domain in Ciona fancl (Fig. 2I), similar to that originally described for human FANCL.^9,54 Subsequent structural analyses of Drosophila and human FANCL have revealed that FANCL encompasses three distinct domains: an amino- terminal E2-like fold, a central double RWD-like domain, and a carboxy-terminal RING domain.^55,56 Structural homology modeling of Ciona fancl, based on the 3.2 Å Drosophila melanogaster FANCL structure (PDB ID 3K1L),⁵⁵ indicates the existence of close structural similarity (Fig. 2J and K). In addition, Clustal Omega multiple sequence alignment (MSA) analyses of human, mouse, and Ciona FANCL indicate that K22, a predicted site of autoubiquitination, is conserved in all three species (data not shown). The ML best tree (Fig. 2H) agrees with this finding, showing that the C. intestinalis candidate falls in a clade with the vertebrate FANCL proteins to the exclusion of the second most similar Ciona and human proteins. However, moving the C. intestinalis candidate further from the vertebrate FANCL clade, or as a sister taxon to the vertebrate FANCL genes, does not make for a statistically worse tree (data not shown). This ambiguity indicates that the phylogenetic evidence for orthology is weak. However, based on the structural similarities, there is reasonably strong support for the C. intestinalis gene to be a true ortholog of human FANCL.

FANCM

FANCM is also a component of the FA core complex and plays a key role in DNA replication fork remodeling and the chromatin recruitment of the group I proteins during ICL repair.^11,57–61 RBB returns a putative Ciona FANCM protein as the closest match. Secondary structure analysis shows that both the human and Ciona candidate proteins possess a DEAH-box helicase/DNA-stimulated ATPase domain (Fig. 3B). The human FANCM protein also possesses a degenerate XPF/ERCC4 endonuclease domain that the Ciona protein lacks.¹² The hydropho-bicity plot shows high levels of correlation, especially toward the amino-terminus (Fig. 3A). In the ML tree, the Ciona FANCM candidate clusters with the vertebrate FANCM proteins in a clade with 94% bootstrap support (Fig. 3C). These data indicate strong support for the orthology of the C. intestinalis candidate.

Figure 3.

Analysis of FANCM (A–C) and FANCT/UBE2T (D–F) putative homologs in C. intestinalis. (A) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCM. (B) Diagrammatic comparison of human and C. intestinalis FANCM inferred protein motifs. (C) Best ML tree for alignment of FANCM and putative homologs in C. intestinalis and other eukaryotes. Cifancm has 100% bootstrap support for membership in the clade with vertebrate FANCM proteins, to the exclusion of the next most similar C. intestinalis protein. (D) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for UBE2T. (E) Diagrammatic comparison of human and C. intestinalis UBE2T inferred protein motifs. (F) Best ML tree for alignment of putative UBE2T homologs, showing that the best BLAST match to UBE2T, Ciube2–17, may not be as closely related to UBE2T as Ciube2D3l (see text). (G) Switching the relationships of Ciube2 J1l with Ciube2D3l does not make a significantly worse tree. (H) However, switching the best BLAST hit, Ciube2–17, with Ciube2D3l does result in a significantly worse tree (**P < 0.02).

FANCT/UBE2T

FANCT/UBE2T is one of the many E2 ubiquitin-conjugating enzymes found in the human proteome and is the specific one implicated in the monoubiquitination of FANCD2 and FANCI.¹⁰ In humans, UBE2T interacts with FANCL to ubiquitinate FANCD2. Patient-derived mutations in the UBE2T gene have recently been discovered in two unrelated patients, leading to a call to denote UBE2T as FANCT.⁶²

The Delta-BLAST search returns Ciona ube2–17kd as the closest match to human UBE2T. However, the reciprocal BLAST against human proteins returns human UBE2D4 with an E-value of 2 × 10⁻⁷⁷ (Table 1). The RSD method returns Ciona ube2 J1l with an E-value of 9 × 10⁻⁷⁶. Apparently, these very similar E2 ubiquitin-conjugating enzymes cannot be reliably distinguished by BLAST searches (Fig. 3E). The hydropathy and phylogenetic analyses (Fig. 3D and F) do not help to resolve the exact relationship. In the hydropathy plot, it is apparent that both the Xenopus and Ciona proteins roughly follow the pattern of human UBE2T, but neither closely matches with the hydropathy of the human protein. Curiously, in the ML phylogenetic analysis, the best tree shows human UBE2T clustering with another Ciona ube2 protein, Ciona ube2D3-like, but not the Ciona ube2–17, or ube2J1l proteins that are the best hits in the RBB and RSD analyses (Fig. 3F). If Ciona ube2J1l is grouped with human UBE2T, the tree is not statistically worse (Fig. 3G), but if Ciona ube2D3l is swapped with Ciona ube2–17, the tree does become significantly worse (Fig. 3H). In short, there are multiple ube2 proteins in Ciona that have such high similarity to the human UBE2T that they alternately appear as putative homologs in different analytic methods. We suggest that it is likely that one of these performs the same E2 ubiquitin conjugation function as UBE2T does in the human FA pathway.

Both group II genes have orthologs in Ciona

FANCD2

FANCD2 is one of the proteins monoubiquitinated by FANCL and FANCT/UBE2T during ICL repair.^5,9,10 Both RBB and RSD returned a putative FA complementation group D2 protein in C. intestinalis as the closest match for this protein in humans, with the BLAST search returning 25% identity, a 44% match on positives, and an E-value of less than 1.7 × 10⁻³⁰⁸, indicating extremely strong similarity (Table 1). The Ciona fancd2 protein contains 1394 aa, while the most common isoform in humans is 1451 aa long.

When the sequences are aligned and gaps removed, the smoothed hydrophobicity plots show multiple similarities (Fig. 4A). The proteins have highly similar (R² $ 0.71) regions at around aligned Ciona aa 100–125, 240–280, 510–540, 660–760, 1010–1045, and 1130–1170. Both the human and Ciona proteins show five globular domains with moderate alignment. The phylogenetic analysis groups the C. intestinalis fancd2 candidate with vertebrate, fly, urchin, and amphioxus putative orthologs, although at low bootstrap support (Fig. 4B). Forcing the C. intestinalis candidate out of the FANCD2 clade makes the tree significantly worse at the P < 0.02 level (Fig. 4C).

In addition, Clustal Omega MSA analyses of human, mouse, and Ciona FANCD2 revealed a strong conservation of the CUE ubiquitin-binding domain,⁶³ the PCNA- interaction motif,⁶⁴ and the site of FANCD2 monoubiquitination K561 (Fig. 4D and E).⁵ Furthermore, structural homology modeling of Ciona fancd2, based on the 3.4 Å Mus musculus Fancd2-Fanci heterodimer structure (PDB ID 3S4W),⁶⁵ reveals a largely favorable structural similarity (Fig. 4F and G). Taken together, we consider that these data provide good support for the presence of a C. intestinalis fancd2 gene.

FANCI

Like FANCD2, FANCI is monoubiquitinated by FANCL and FANCT/UBE2T during ICL repair. Both RBB and RSD returned a C. intestinalis candidate fanci as the closest match to the human FANCI protein, with an E-value of 0. The hydrophobicity plots return an R² value of 0.33, but several areas, notably a 150 amino acid stretch toward the carboxy-terminal end of the protein, have much higher correlations (Fig. 4H). Both proteins show multiple globular domains with moderate alignment and no recognizable secondary motifs. Clustal Omega MSA analyses of human, mouse, and Ciona FANCI indicate the conservation of K523 and K715, the sites of FANCI monoubiquitination and SUMOylation, respectively (Fig. 4J).^6,7,66 In addition, Ciona fanci contains multiple conserved SQ/TQ ATM/ATR kinase phosphorylation motifs proximal to the putative monoubiquitination site (Fig. 4K). In vertebrates, these sites have been demonstrated to be critical for FANCI regulation and function.^67,68 On the other hand, the ML phylogenetic analysis is inconclusive with respect to the orthology of the C. intestinalis candidate and FANCI. The best ML tree places the Ciona candidate as a sister taxon to a clade of deuterostome plus cnidarian FANCI proteins (Fig. 4I). However, forcing the Ciona candidate into the vertebrate FANCI clade results in a statistically worse tree, while forcing the Ciona candidate to group with the next most similar Ciona protein is not significantly different from the best ML tree (data not shown). In spite of the lack of support from the phylogenetic analysis, the sequence motif and structural data strongly suggest that Ciona fanci is a true FANCI ortholog.

Seven group III orthologs were found

FANCJ/BRIP1

In humans, FANCJ is a 5′–3′ DNA helicase that interacts directly with BRCA1.^69,70 RBB returns the ERCC2 nucleotide excision repair protein, but RSD returns human FANCJ. There is good alignment between the globular domains in human FANCJ and the Ciona candidate, and the hydrophobicity plot shows high correlation (Fig. 5A). The human protein is of a similar size to the Ciona protein, and they both possess a DEAH-box helicase domain (Fig. 5B). The ML tree groups C. intestinalis fancj in the vertebrate FANCJ clade at 100% bootstrap support, and moving the C. intestinalis candidate out of that clade makes the tree significantly worse (Fig. 5C and D). Given these data, the C. intestinalis fancj candidate is a clear ortholog of human FANCJ.

Figure 5.

Analysis of FANCJ (A–D) and FAN1 (E–H) putative homologs in C. intestinalis. (A) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCJ. (B) Diagrammatic comparison of human and C. intestinalis FANCJ inferred protein motifs. (C) Best ML tree for alignment of FANCJ and putative homologs in C. intestinalis and other eukaryotes. Cifancj has 100% bootstrap support for membership in the clade with vertebrate FANCJ proteins, to the exclusion of the next most similar C. intestinalis protein. (D) If Cifancj is moved out of the FANCJ clade, the tree is statistically worse at the P = 0.01 level, supporting the case for Cifancj as a true homolog of FANCJ. (E) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FAN1. (F) Diagrammatic comparison of human and C. intestinalis FAN1 inferred protein motifs. (G) Best ML tree for alignment of putative FAN1 homologs, with Cifan1 falling in a clade with the vertebrate FAN1 homologs with 100% bootstrap support. Forcing Cifan1 out of that clade results in a statistically worse tree (P < 0.01).

FAN1

Fanconi-associated nuclease 1 is a DNA repair protein known to interact with monoubiquitinated FANCD2¹⁴ and FANCI.⁷¹ The RBB returns a protein annotated as Ciona fan1, with an E-value of 4 × 10⁻¹⁴⁵. The fan1 C-terminal region shows 41% identity and 63% positive matches. The human and Ciona proteins align extremely well in the hydropathy plot (Fig. 5E) and both contain a 110 aa VRR nuclease domain (Fig. 5F). The ML tree clusters the C. intestinalis candidate with the vertebrate FAN1 proteins (Fig. 5G) and is significantly worse when the C. intestinalis protein is taken out of that clade (Fig. 5H; P < 0.01). Taken together, the evidence is strongly in favor of Ciona fan1 being a homolog of FAN1.

FANCQ/ERCC4

The FANCQ gene product, also known as ERCC4 or XPF, forms a heterodimer with ERCC1 and functions as a DNA repair structure-specific endonuclease.⁷² Both search methods return a Ciona xpf as the most closely matching protein, with 50% identity, and 64% positive matches. The hydrophobicity plots show a high correlation, excepting one area corresponding to aa 390–430 in Ciona and aa 520–560 in humans (Fig. 6A). Both proteins possess an ERCC4 endonuclease domain of the same size approximately the same distance from the carboxy-terminal end of the protein (Fig. 6C). The ML analysis clusters the C. intestinalis xpf in the FANCQ clade (Fig. 6B), although moving the C. intestinalis protein out of that clade does not make the tree likelihood significantly worse (data not shown). Taken together, we conclude that C. intestinalis does have a FANCQ ortholog.

Figure 6.

Analysis of FANCQ/ERCC4 (A–C) and ERCC1 (D–H) putative homologs in C. intestinalis. (A) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCQ/ERCC4. The best C. intestinalis BLASTmatch to FANCQ is termed XPFin GenBank (Table 1). (B) Best MLtree for alignment of FANCQ and putative homologs in C. intestinalis and other eukaryotes. CiXPFhas 100% bootstrap support for membership in the clade with vertebrate FANCQ proteins, to the exclusion of the next most similar C. intestinalis protein. (C) Diagrammatic comparison of human FANCQ and C. intestinalis XPFinferred protein motifs. (D) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for ERCC1. (E) Diagrammatic comparison of human and C. intestinalis ERCC1 inferred protein motifs. (F) Best MLtree for alignment of putative ERCC1 homologs, with Ci-ercc1 shown closely related to ERCC1. However, the next closest C. intestinalis match, Ci-xpf, is in the same clade. However, forcing Ci-ercc1 out of the clade (G), or further from ERCC1 (H), both result in statistically worse trees (P < 0.02).

ERCC1

ERCC1 interacts directly with FANCQ/ERCC4. The Ciona candidate returned by RBB (XP_009861832) has an extremely similar hydropathy plot as the human and frog ERCC1 proteins, except at the N- terminal-most 50 residues (Fig. 6D), although the Ciona candidate appears to lack an intact HhH1 domain present in the human protein (Fig. 6E). The ML analysis groups the Ciona candidate within the vertebrate ERCC1 clade (Fig. 6F). Moving the Ciona protein outside that clade or grouping it with the next most similar human gene (FAAP24) makes the trees statistically worse at the P < 0.02 level (Fig. 6G and H). These data strongly support the orthology of the Ciona candidate.

FANCO/RAD51C

RAD51C is also required for the maintenance of chromosome stability by functioning in HR repair.⁷³ Ciona has five potential RAD51 family homologs if the proteins listed as lim15 and xrcc2 are included. RBB finds Ciona rad51 (XP_002126934) as the closest match to human FANCO. However, if the Ciona protein identified as rad51C in GenBank (XP_002130341) is used in the ML analysis with FANCO, Ciona rad51C robustly groups with FANCO to the exclusion of Ciona rad51 (Fig. 7C). Forcing Ciona rad51C out of the FANCO clade results in a statistically worse tree (Fig. 7D, P < 0.01). Structurally, the Ciona rad51C is more similar to FANCO than the higher BLAST match Ciona rad51 (Fig. 7A and B). Based on these analyses, we conclude that Ciona does have a FANCO homolog.

Figure 7.

Analysis of FANCO/RAD51C putative orthologs in C. intestinalis. (A) Hydropathy plot of human, Xenopus, and Ciona putative FANCO proteins. (B) Diagrammatic comparison of Homo sapiens FANCO and C. intestinalis RAD51C protein motifs. (C) ML analysis has CiRAD51C grouping with the vertebrate FANCO protein sequences at moderate (86%) bootstrap support. (D) If CiRAD51C is excluded from the clade with the vertebrate FANCO proteins, the tree is significantly worse (***P < 0.01).

FANCR/RAD51

In humans, RAD51, recently gaining the name FANCR, is the major DNA strand exchange protein and is critical for the HR DNA repair process.^74,75 De novo heterozygous RAD51 mutations have recently been reported in two unrelated individuals with an FA-like syndrome.⁷⁶ RAD51 is known to interact with both FANCS/BRCA1 and FANCD1/BRCA2 in the cellular DNA damage response.⁷⁷ Both search methods return a Ciona rad51 as the most likely counterpart to the human protein. RAD51 appears to be the most highly conserved protein in the entire FA pathway. The protein possesses 82% identity between human and Ciona as well as a 92% level of positive matches, far outstripping any other gene product tested. The Ciona product is 338 aa in length, while the human product is 339 aa (Fig. 8B). Both RAD51 and the Ciona rad51 candidate possess a 20 amino acid helix-hairpin-helix domain starting at about amino acid 60, as well as a 187 aa AAA-ATPase domain ending 33 aa before the C-terminus. The hydrophobicity plots show extreme similarity, returning a Pearson coefficient of 0.92 (Fig. 8A). The ML analysis shows the Ciona rad51 candidate grouping with other deuterostome RAD51 proteins (Fig. 8C), while excluding Ciona rad51 from that clade results in a statistically worse tree (Fig. 8D, P < 0.01). It is highly likely that Ciona rad51 is a true ortholog of human FANCR/RAD51.

Figure 8.

Analysis of FANCR/RAD51 (A–D) and FANCS/BRCA1 (E–G) putative homologs in C. intestinalis. (A) Hydropathy plot showing extremely similar hydropathy profiles in human, Xenopus, and Ciona putative homologs for RAD51. (B) Diagrammatic comparison of human and C. intestinalis RAD51 inferred protein motifs. (C) Best ML tree for alignment of putative RAD51 homologs, showing Ci-rad51 closely related to RAD51. (D) Forcing Ci-rad51 out of the clade results in a statistically worse tree (P < 0.01). (E) Hydropathy plot of best aligning regions in human, Xenopus, and Ciona putative homologs for FANCS/BRCA1, showing regions of both similar and divergent hydropathy. (F) Diagrammatic comparison of human FANCS/BRCA1 and putative C. intestinalis brca1 inferred protein motifs. The human protein has a RING domain not present in the C. intestinalis candidate. (G) Best ML tree for alignment of FANCS/BRCA1 and putative homologs in C. intestinalis and other eukaryotes. There is weak support for a clade with Ci-brca1 and FANCS/BRCA1.

FANCS/BRCA1

The C. intestinalis candidate for FANCS by RBB has two BRCT (BRCA1 C-terminal domain) domains at the C-terminus, similar to BRCA1 (Fig. 8F). BRCT domains typically mediate interactions with phosphopeptides. The hydropathy plot of the C-terminal 500 residues of the C. intestinalis, human, and frog proteins show a good degree of similarity (Fig. 8E). However, the rest of the sequence of the 1172 aa predicted C. intestinalis protein (from the ANISEED database as KH2012:KH.C9.487) toward the N-terminus has little resemblance to the human FANCS/BRCA1. Most likely because of this lack of alignment for a large part of the sequence, the ML analysis does not group the C. intestinalis protein with FANCS at a robust level (Fig. 8G). In fact, moving the C. intestinalis sequence either within the vertebrate BRCA1 clade or to the more distant branch of the tree makes for a statistically worse topology (data not shown). Because part of the protein is similar to its putative homolog while over half is not, we cannot say with complete confidence that “Ci-brca1” is a true homolog. However, it may still be the case that this protein in combination with one or more others is fulfilling the function served in humans by BRCA1.

FA and FA-associated proteins not found in Ciona

Our analyses found 11 FA or FA-associated proteins present in vertebrates but not in Ciona. These results were based on the four major criteria outlined for each of the predicted FA homologs, as outlined above, namely, BLAST search, structural motif similarity, hydropathy, and phylogenetic (ML) analysis. The FA proteins that we did not find homologs for in Ciona were as follows: FANCA, FANCB, FANCC, FANCF, FANCG, FANCD1/BRCA2, FANCN/PALB2, and FANCP/SLX4. We also failed to find the FA-associated proteins FAAP20, FAAP24, and FAAP100.

For 10 of the 11 cases, RBB and RSD failed to match a Ciona protein sequence with an FA-related protein (Table 1). The exception is FANCD1/BRCA2, for which a match comes up in RBB as an uncharacterized protein LOC100185089 (Table 1). However, the ML analysis results in another C. intestinalis protein showing a closer relationship to FANCD1/BRCA2. Rearranging the trees so that the best BLAST match is moved out of the FANCD1 clade altogether, or switching the first and second most similar C. intestinalis proteins in the tree, does not result in statistically worse trees, indicating that the evidence for homology of the C. intestinalis proteins is weak (data not shown). In addition, the hydropathy analysis shows a low correlation (R² = 0.117, Table 2). A Prosite scan indicates that LOC100185089 has two BRC repeats, which may explain why it comes up in the BLAST search. However, FANCD1 is a much larger protein (3418 aa vs. 724 aa) and has eight BRC repeats. These BRC repeats represent the major sites of interaction between RAD51 and BRCA2.⁷⁸ In addition, BRCA2/FANCD1 has an α-helical region, an oligonucleotide/oligosaccharide-binding domain, a TOWER domain, and a second oligonucleotide/oligosaccharide-binding domain. C. intestinalis LOC100185089 possesses two BRC repeats only. None of these other domains are present.

There is a possibility that the predicted Ciona protein in the NCBI database is not the full-length sequence. However, we searched a 22 kb region in the Ciona genome, which includes LOC100185089 and flanking regions. No signifi-cant similarity to the human sequence outside the region that aligns with LOC100185089 was found, even when the protein sequence not included in LOC100185089 was blasted against the translated Ciona genomic sequence. Thus, we infer that Ciona does not have a complete ortholog of human BRCA2.

For the other 10 of the 11 cases of unlikely homology, the hydropathy R² statistics are lower numbers than those for the putative homologs, ranging from 0.034 to 0.177 vs. 0.291 to 0.566, respectively (Table 2). Similarly, we did not find good evidence for homology to any C. intestinalis proteins by any of the other three analytical methods used (Table 1, and data not shown). Therefore, we conclude that these 11 FA and FA-associated proteins are missing from C. intestinalis.

Discussion

In this study, we have established that the model marine invertebrate, C. intestinalis, appears to contain all of the necessary functional components to reconstitute a simplified FA pathway (Fig. 9). Of the FA core complex group I proteins, we identified orthologs of FANCL, FANCT/UBE2T, and FANCM, and possibly FANCE. FANCL and FANCT/UBE2T are the E3 ubiquitin ligase and E2 ubiquitin conjugase enzymes, respectively, that monoubiquitinate FANCD2 and FANCI.^5–7,10 While FANCD2 and FANCI monoubiquitination are largely defective in FA patient cells with mutations in any of the core complex genes (FANCA, B, C, E, F, G, L, and T), several studies have established that FANCL and FANCT/UBE2T, in the presence of an E1 ubiquitin-activating enzyme and DNA, can readily promote FANCD2 and FANCI monoubiquitination in vitro.^8,79–81 The roles of the other FA core complex proteins in promoting FANCD2 and FANCI monoubiquitination in vivo remain unknown. The functions provided by these other core complex proteins may be unnecessary in C. intestinalis, or may be provided by other proteins. Interestingly, previous studies have established that the FANCE protein directly interacts with FANCD2, thereby bridging the core ubiquitin ligase machinery and the substrate. C. intestinalis fance may fulfill an analogous function. Similar to human FANCM, C. intestinalis fancm contains an N-terminal DEAH domain-containing Walker A and B motifs typical of an SF2 family translocase. These proteins are capable of movement along DNA in the absence of helicase activity. FANCM translo-case activity is necessary for replication fork stability and ATR-CHK1 checkpoint signaling.^82,83 The C-terminus of human FANCM contains a degenerate ERCC4 endonuclease domain, which is also the site of binding of its heterodimeric partner FAAP24; yet, this region appears absent in C. intestinalis fancm (Fig. 3B). Since C. intestinalis appears to lack a FAAP24 homolog, it is not surprising that Cifancm lacks the binding site. It has been speculated that the FANCM-FAAP24 heterodimer plays an important DNA-targeting function, and why the formation of a heterodimer might be unnecessary in C. intestinalis is unclear.^57,84 However, the categorization of FANCM as a true FA gene remains controversial.

Figure 9.

(A) A model of the FA pathway in humans. Following exposure to DNA damaging agents or during S-phase of the cell cycle, the FA core complex (group I) proteins catalyze the monoubiquitination of the FANCD2 and FANCI (group II) proteins. Following their monoubiquitination, FANCD2 and FANCI function together with the downstream FA (group III) proteins to repair damaged DNA. Modified from Cybulski and Howlett, 2011.¹¹⁴ (B) A model of a hypothetical simplified FA pathway in C. intestinalis based on the reduced complement of FA gene homologs found by this study. C. intestinalis possesses the critical E3 ubiquitin ligase (Fancl) and E2 ubiquitin-conjugating enzyme (Fanct) to monoubiquitinate Fancd2 and Fanci, as well as a minimal set of FA group III effector proteins. Proteins shown in gray have lower support for existence in C. intestinalis.

The evidence for structural and functional conservation of the FANCD2 and FANCI proteins appears quite strong, with several protein domains and important sites of posttranslational modification being highly conserved (Fig. 4D, E, J, and K). This is consistent with the previous finding indicating considerable depth in their conservation in all eukaryotes.²⁶ The monoubiquitination of these proteins is a critical step in the activation of the FA pathway and in ICL repair.^5–7 In the case of FANCD2, monoubiquitination of K561 has been implicated in the recruitment of the FAN1 and FANCP/SLX4 proteins, which participate in, or facilitate, several key nucleolytic processing steps during ICL repair.^13–15,17 Conservation of the FANCD2 K561 and FANCI K523 monoubiquitination sites, as well as several other important sites of posttranslational modification, strongly suggests that this central step is intact in C. intestinalis.

Of the group III proteins, the evidence points to the existence of C. intestinalis orthologs of FANCJ/BRIP1, FANCQ/ERCC4, FANCR/RAD51, FANCO/RAD51C, and FANCS/BRCA1. The heterodimeric binding partner of FANCQ/ERCC4, ERCC1, is also present, as is the FANCD2-associated nuclease FAN1. Conservation of FANCR/RAD51 and FANCS/BRCA1 is not surprising, given their key roles in multiple cellular processes, including meiotic and mitotic recombination. Targeted disruption of either gene results in early embryonic lethality in mice.^85,86 However, the absence of FANCD1/BRCA2 is particularly surprising, given its strong conservation among eukaryotes.²⁶ FANCD1/BRCA2 plays a critical role in regulating FANCR/RAD51 nucleoprotein filament formation and DNA strand exchange.^78,87–89 It is also intriguing that C. intestinalis apparently lacks FANCN/PALB2. FANCN/PALB2 interacts directly with FANCD1/BRCA2 and promotes its chromatin localization.²³ Studies of the Ustilago maydis homolog of BRCA2 indicate that BRCA2 promotes RAD51 nucleation at junctions of single-stranded and double-stranded DNA.^90,91 However, lower eukaryotes such as Saccharomyces cerevisiae and Schizosaccharomyces pombe lack homologs of both FANCS/BRCA1 and FANCD1/BRCA2, indicating that the functions provided by these proteins are unnecessary in certain organisms or may be provided by other proteins.

There is considerable precedent suggestive of the efficacy of studying the FA pathway in C. intestinalis. Study of several human diseases have benefited from the use of invertebrate model organisms. In particular, the genetically tractable invertebrates, such as Drosophila and Caenorhabditis elegans, have been used extensively.^92,93 Notably, it has recently been shown that even very simple animals, such as sponges and sea anemones, have homologs of many human disease genes.^94,95 C. intestinalis has only recently emerged as a model system. However, it has already been used to study certain human disease pathways, such as Huntington’s Disease^96,97 and Alzheimer’s disease (AD).⁹⁸ In the case of Alzheimer’s, transgenic C. intestinalis were produced expressing the human APP gene mutant associated with familial AD. The transgenic protein resulted in the formation of amyloid-β plaques in less than 24 hours in the rapidly developing C. intestinalis larval brain. This result contrasts with a 2–8-month time period for plaques to form in mouse AD models. For FA, study of the pathway in invertebrate model organisms has proven valuable in several cases.⁹⁹ For example, the function of FANCJ in maintaining poly(G)/poly(C) tract stability during DNA replication was first shown in the nematode worm C. elegans.¹⁰⁰ It was later demonstrated that human FANCJ has the same helicase function.¹⁰¹

It is important to note that of all the three major constellations of FA patient phenotypes, namely, developmental defects, bone marrow failure, and increased cancer risk, the molecular bases of the developmental defects are the most poorly understood. A C. intestinalis model for FA could provide unique insights into these defects. Temporospatial aspects of FA gene expression and developmental consequences of disruption of FA genes using CRISPR/Cas9 or TALEN systems^102–106 could be highly informative for FA patient developmental defects. Furthermore, another unique benefit to exploring a C. intestinalis model for FA is the prospect of discovering the physiological function(s) of this pathway. While it is well established that FA patient cells are hypersensitive to ICL-inducing agents, the relevance of ICLs in the physiological setting is unclear. Recent studies have established an important role for the FA proteins in mitigating endogenously arising aldehyde-mediated DNA damage.^107–109 Exploring the pathway in other model systems may lead to a broader understanding of the true function(s) of these key proteins. C. intestinalis, as a tunicate, is in the most closely related invertebrate group to the vertebrates.^27,110 As such, in spite of being anatomically simpler than a vertebrate, they are genetically more similar than other eukary-otes. However, it is possible that C. intestinalis may deploy its FA homologs differently than they function in humans. If this is the case, it may still be relevant to understanding human disease, as it will point to alternative ways of dealing with DNA lesions and may provide information on some of the other defects seen in FA patients.

In summary, our study provides compelling evidence for the existence of a simplified and potentially functional FA pathway in the model chordate C. intestinalis. C. intestinalis is an excellent model for the study of developmental processes because it is anatomically simple, its gametogenesis and development are well studied, it has a small and well-annotated genome and abundant gene expression data, and good transgenic technology exists.^{29,30,111–113} Future studies will seek to determine the patterns and timing of FA gene expression in C. intestinalis and the developmental impacts of disruption of the pathway.

Supplementary Material

Supplemental Table S1. Accession Numbers for sequences used in analyses.

Table 2.

Hydrophobicity plot correlations between identities and positive matches.

Group	Name	Id%	Hydropathy Plot ID% R2	Pos%	Hydropathy Plot Pos% R2
I	FANCA	36%	0.052	58%	0.027
	FANCB	37%	0.045	52%	0.026
	FANCC	32%	0.121	46%	0.052
	FANCE	19%	0.082	34%	0.051
	FANCF	47%	0.039	65%	0.021
	FANCG	56%	0.034	74%	0.021
	FANCL	36%	0.373	56%	0.199
	FANCM	52%	0.369	66%	0.321
	FAAP20	18%	0.053	28%	0.027
	FAAP24	26%	0.177	45%	0.098
	FAAP100	35%	0.114	53%	0.053
	UBE2T	34%	0.311	58%	0.222
II	FANCD2	25%	0.304	44%	0.222
	FANCI	29%	0.331	50%	0.237
III	FANCD1	31%	0.127	49%	0.046
	FANCJ	37%	0.291	53%	0.236
	FANCN	18%	0.099	45%	0.061
	FANCO	41%	0.328	44%	0.345
	FANCP	32%	0.050	49%	0.030
	FANCQ/XPF	50%	0.566	68%	0.446
	FANCR/RAD51	82%	0.844	92%	0.847
	FANCS/BRCA1	27%	0.117	46%	0.322
	ERCC1	47%	0.396	65%	0.446
	FAN1	41%	0.447	63%	0.387

Note: ID% refers to the Delta-BLAST result for identical amino acid matches. Pos% refers to Delta-BLAST results for positive amino acid matches, eg, aa from the same functional groups.