ARS: Arylsulfatase; SULF: Sulfatase; HSPG: Heparin Sulfate Proteoglycan; FGF: Fibroblast Growth Factor; WNT: Wingless-related Integration; VEGF: Vascular Endothelial Growth Factor; GDNF: Glial Cell Line-Derived Neurotrophic Factor; PNH: Phosphonate Monoester Hydrolase; BLAST: Basic Local Alignment Search Tool; BLAT: Blast-Like Alignment Tool; NCBI: National Center for Biotechnology Information; UCSC: University of California Santa Cruz; KO: Knock Out; AceView: NCBI Based Representation of Public mRNAs; SWISS-MODEL: Automated Protein Structure Homology- Modeling Server; TFBS: Transcription Factor Binding Sites; UTR: Untranslated Region

Extracellular sulfatases 1 and 2 (SULF1; SULF2; E.C.3.1.6.-; also described as heparan sulfate 6-O-endosulfatase; SUL1; SUL2) are members of the sulfatase enzyme family, for which seventeen genes have been described on the human genome [1,2]. SULF1 and SULF2 are secreted enzymes which carry out essential roles in the extracellular environment by catalysing endoglucosamine-6-sulfatase activity and removing 6-O sulfate groups from Heparin Sulfate Proteoglycans (HSPGs) [3,4]. These perform several key roles: modulating the activity of growth factor receptors and cell signaling pathways, such as FGF, VEGF, GDNF and WNT signaling pathways, which initiate gene transcription signals through cell surface receptors [5-11]; serving essential roles in vertebrate development [12-20]; and in modulating microbial (Chlamydia muridarum) infection [21].

Structures for vertebrate SULF1 and SULF2 genes and cDNA sequences have been reported, including human (Homo sapiens) [3]; mouse (Mus musculus) [5,22,23]; rat (Rattus norvegicus) [24,25]; frog (Xenopus laevis) [13-14]; and zebra fish (Danio rerio) [16] SULF genes. Human SULF1, which spans 194.3 kilobases and comprises 22 exons, is localized on chromosome 8; whereas human SULF2 spans 128.7 kilobases and comprises 21 exons on chromosome 20 [26,27]. Both of these genes are widely expressed in the body, consistent with their overlapping and essential roles in cell signaling pathways, skeletal muscle regeneration, neonatal development and survival, metastasis and wound repair [9,18,19,23].

This paper reports the predicted gene structures and amino acid sequences for several vertebrate SULF1 and SULF2 genes and proteins, the predicted secondary and tertiary structures for human SULF1 and SULF2 protein subunits, and the structural, phylogenetic and evolutionary relationships for these genes and enzymes. Evidence is also presented for SULF2 playing a significant role in female reproductive tissues involving the estrogen receptor localized within the SULF2 promoter [28,29].

Gene and protein identification

BLAST studies were undertaken using web tools from the NCBI (http://www.ncbi.nlm.nih.gov) [30]. Protein BLAST analyses used human ARS sequences (Group 1: ARSA, ARSG, GALNS; Group 2: ARSB, ARSI, ARSJ; Group 3: ARSD, ARSE, ARSF, ARSH, STS; Group 4: SULF1, SULF2, GNS; Group 5: ARSK; Group 6: SGSH; Group 7: IDS); and other vertebrate SULF1 and SULF2 amino acid sequences previously described (Tables 1 and 2) [2-4]. Predicted SULF1 and SULF2-like protein sequences were obtained in each case and subjected to protein and gene structure analyses.

Table 1: Vertebrate SULF1 and SULF2 and Caenorhabditis elegans SUL-1 genes and proteins.

Table 2: Proposed classification of human arylsulfatase genes and proteins.

BLAT analyses were undertaken for each of the predicted SULF1 and SULF2 amino acid sequences using the UCSC Genome Browser (http://genome.ucsc.edu) with the default settings to obtain the predicted locations for each of the vertebrate SULF-like genes, including exon boundary locations and gene sizes [27]. The structures for the major human SULF1 and SULF2 trancripts were obtained using the AceView website (http://www.ncbi.nlm.nih.gov/ieb/research/acembly/) [26]. Alignments of SULF sequences with human SULF1 and SULF2 protein sequences were assembled using the Clustal Omega multiple sequence alignment program [31]. Predicted micro-RNA binding sites (miR) and CpG islands [32] were examined using the UCSC Genome Browser [27]. Predicted human SULF1 and SULF2 transcription factor binding sites (TFBS) were obtained from the PAZAR (OregAnno) dataset [33] (http://www.oreganno.org).

Structures and predicted properties of SULF1 and SULF2 proteins

Predicted secondary structures for human and other mammalian SULF1 and SULF2 proteins were obtained using the SWISS-MODEL web-server [34] and the reported structures using bacterial phosphonate monoester hydrolase (PNH) from Silicibacter pomeroyi (PDB:4UPKA) with modeling residue ranges of 42-416 for human SULF1 and 43- 419 for human SULF2 (Figure 1). Predicted secondary structures for the hydrophilic zones for both SULF1 (residues 397-871) and SULF2 (residues 398-870) were undertaken using the PSIPRED web server [35]. Identification of conserved domains for vertebrate SULF1 and SULF2 proteins was made using NCBI web tools [36].

Figure 1: Amino acid sequence alignments for human SULF1 and SULF2 subunits. See Table 1 for sources of SULF1 and SULF2 sequences; *Shows identical residues for SULF subunits; similar alternate residues; .dissimilar alternate residues; leader peptide residues are in dark yellow; predicted helix; predicted sheet; active site residues shown in blue; N-glycosylated Asn residues are in light green; HD refers to hydrophilic C-terminal sequence; acidic amino acids in HD zone are in dark green; basic amino acid residues in HD zone are in pink; bold font shows known or predicted exon junctions; exon numbers refer to human SULF1 gene.

Comparative human SULF1 and SULF2 gene expression

RNA-seq gene expression profiles across 53 selected tissues (or tissue segments) were examined from the public database for human SULF1 and SULF2, based on expression levels for 175 individuals [37] (Data Source: GTEx Analysis Release V6p (dbGaP Accession phs000424.v6.p1) (http://www.gtexportal.org).

Phylogeny studies and sequence divergence

Phylogenetic analyses were undertaken using the http://phylogeny.fr platform [38]. Alignments of SULF1 and SULF2 sequences were assembled using MUSCLE (Table 1) [39]. Alignment ambiguous regions were excluded prior to phylogenetic analysis yielding alignments for comparisons of these sequences. The phylogenetic tree was constructed using the maximum likelihood tree estimation program PhyML [40].

SULF1, SULF2 and other human sulfatase genes and proteins

Table 2 summarises the comparative genomic and proteomic features for 17 human sulfatase genes and proteins, including SULF1 and SULF2, which are members of the human Group 4 ARS genes [2]. These genes were separately located on human chromosomes (chromosomes 8 and 20, respectively. This is in contrast to Group 3 ARS genes (ARSD, ARSE, ARSF, ARSH and STS), which are localized consecutively within a sterylsulfatase gene cluster on the human X-chromosome (Table 2). SULF1, SULF2 and GNS genes have been designated as belonging to ARS Group 4 [2], due to their higher sequence identities (40-67%) than with other human ARS enzymes (12-22% identical), and to similarities in substrate specificities, acting on either endoglucosamine 6-sulfate (SULF1 and SULF2) [3,9,12] or N-acetylglucosamine 6-sulfate (GNS) substrates [41,42], respectively. In addition, these genes have apparently been derived from a common invertebrate ancestral gene, SUL1 (identified in C. elegans) and SULF1 (identified in D. melanogaster) [2,11].

Alignments of SULF1 and SULF2 subunits

Alignments of amino acid sequences for human SULF1 and SULF2 subunits previously reported [3] are shown in Figure 1. The sequences were 66% sequence identitical (Table 3), suggesting that these are products of two related families of genes and proteins, namely SULF1 and SULF2 (Table 2). Studies of the amino acid sequences for other vertebrate SULF subunits have shown that they contained 867-892 residues for SULF1, whereas vertebrate SULF2 subunits contained 870- 888 residues (Table 1), with higher levels of sequence identity observed for subunits from the same gene family, in each case (Table 3). Several key amino acid residues or regions for human SULF1 and SULF2 were recognized (sequence numbers refer to human SULF1 (Figure 1). These included the leader peptide (residues 1-22 for SULF1; 1-24 for SULF2); metal binding residues at the active site (Ca²⁺) (51Asp, 52Asp, 316Asp and 317His); the active site 87Cys, which functions by forming the 3-oxoalanine residue; and seven N-glycosylation sites, located in the N-terminal region (64Asn, 111Asn, 131Asn, 148Asn, 170Asn, 197Asn and 240Asn) and three N-glycosylation sites in the C-terminal region (623Asn, 773Asn and 783Asn). Comparisons of human SULF1 and SULF2 amino acid sequences with other human ARS sequences showed that SULF1 and SULF2 subunits contained extended C-terminal sequences (with >300 additional amino acid residues) (Table 2). Moreover, these C-terminal regions contained high basic amino acid content, assisting the formation of ionic linkages between SULF1 and SULF2 subunits with the heparan sulfate proteoglycan substrates in the extracellular envoronment, where the enzymes operate to modify the structures of the heparan sulfate chains [10,43]. In addition to these clusters of basic amino acid residues, the SULF1 C-terminal region contained a poly-Glu (x5) acidic amino acid zone (Glu560-564) which may be involved in the formation of ionic linkages with the highly basic C-terminus (Figure 1).

Table 3: Percentage identity matrix for vertebrate and Caenorhabditis elegans SULF amino acid sequences.

Predicted secondary structures of SULF1 and SULF2 subunits

Analyses of predicted secondary structures for human SULF1 and SULF2 sequences were obtained using the SWISS-MODEL web-server [32] and the reported tertiary structures using bacterial Phosphonate Monoester Hydrolase (PNH) from Silicibacter pomeroyi (PDB:4UPKA) (Figure 1). Several α-helix and β-sheet structures were observed for the human SULF1 and SULF2 subunits examined, with 11 β-sheet and 7 α-helices predicted. Of particular interest was the prediction of β-sheet and α-helix structures at the N-terminal end of the SULF subunits, in comparison with extended hydrophilic C-terminal sequence. Secondary structures were readily apparent near key residues or functional domains including the β-sheet and α-helix structures near the substrate binding active site (87Cys) and the metal binding residues at the active site (Ca²⁺) (51Asp, 52Asp, 316Asp and 317His) [3,10].

The predicted secondary structures for human SULF1 and SULF2 showed similarities to structures previously reported for other ARS proteins, including human ARSA [44], ARSB [45], STS [46] and SGSH [47]. The active site for SULF1 was centrally located with two β-sheet structures (β1, β6) and the metal binding residues at the active site (Ca+) (51Asp, 52Asp, 316Asp and 317His). The hydrophilic C-terminal region was absent in the ARSA, ARSB, STS and SGSH proteins previously reported [44-47]. The positively charged Hydrophilic Domain (HD) domain has been previously characterized as having high affinity with heparan/heparan sulfate, with specific regions influencing different aspects of heparan sulfate binding, cellular localization and enzyme function [4].

Predicted gene locations and exonic structures for vertebrate and invertebrate SULF genes

Table 1 summarizes the predicted locations for vertebrate SULF1 and SULF2 genes based upon BLAT interrogations of genomes using the reported sequences for human, mouse and frog SULF1 and SULF2 [3-5,14] and the predicted sequences for other SULF1 and SULF2 proteins and the UCSC Genome Browser [27]. Human SULF1 and SULF2 genes were located on different chromosomes (chromosomes 18 and 20, respectively), which is the case for all vertebrate genomes examined (Table 1). Of particular interest to the evolution of SULFlike genes in invertebrate genomes, the worm (Caeborhabditis elegans) showed evidence of having only one gene which was similar to the vertebrate SULF1 gene which encoded a SULF-like gene (designated as sul1). This amino acid sequence also encoded a leader peptide, similar to that for human SULF1 and SULF2 proteins.

Figure 1 summarizes the predicted exonic start sites for human SULF1 and SULF2 genes which contained 18 or 20 exons, respectively, in identical or similar positions, with the exception of 2 additional exons (exons 19 and 20) encoded at the C-terminus end of SULF2. In each case, exon 1 encoded the leader peptide and the double aspartate (Asp51-Asp52) Ca²⁺ binding site; exon 2 encoded the active site 87Cys, which functions by forming the 3-oxoalanine residue; exon 6 encoded two other active site residues, 316Asp and 317His; and exons 9-18 (or 20, in the case of SULF2) encoded the hydrophilic C-terminus region.

Figure 2 illustrates the predicted structures of mRNAs for human SULF1 and SULF2 transcripts for the major transcript isoforms in each case [26]. The genes cover 194.3 and 130.5 kilobases in length, respectively, with 18 introns and 20 exons present for the mRNA transcripts. The human SULF1 gene promoter contained six predicted TFBS (Figure 2 and Table 4), including 3 binding sites for FOXA1, encoding hepatocyte nuclear factor 3-alpha, which participates in embryonic development and directs tissue-specific gene expression [48]; 2 binding sites for TFA2PC, encoding transcription factor AP-2 gamma, which is involved in eye, face, body wall, limb and neural tube development [49]; and a binding site for EGR1, encoding early growth response protein 1, a gene regulator which regulates the transcription of several genes involved in early vertebrate development [50]. Three of these TFBS were also observed for the SULF2 promoter, including FOXA1, TFA2PC and EGR1, although three others were found in this region, including ESR1, encoding the estrogen receptor [51]; HNF4A, encoding hepatocyte nuclear factor 4-alpha, controlling several genes essential for the development of liver, intestine and kidney [52]; and CTCF, encoding CCCTC-binding factor, which is necessary for memory formation and for basal and experience-dependent gene regulation [53].

Figure 2: Gene Structures for the Human SULF1 and SULF2 genes. Derived from the AceView website http://www.ncbi.nlm.nih.gov/IEB/Research/Acembly/ [26]; the major isoform variants are shown with capped 5’- and 3’- ends for the predicted mRNA sequences; introns (pink lines) and exons (pink boxes) are shown; the length of the mRNAs (as kilobases or kb) are shown; a CpG island (CpG165) is shown for the SULF2 promoter; 21 miR-binding sites were observed in for the 3’UTR of the human SULF1 gene (Supplementary Table 1); a miRNA-202 binding site was identified for the 3’UTR of the human SULF2 gene; the direction for transcription is shown; TFBS refers to transcription factor binding sites located within the SULF1 and SULF2 gene promoters (Table 4); individual TFBS were identified within these promoter regions, including the estrogen receptor (ESR1) within the SULF2 promoter.

Table 4: Transcription factor binding sites (TFBS) identified for human SULF1 and SULF2 gene promoters.

Many microRNA binding sites were located in the 3’-UTR of human SULF1, which are potentially of major significance for the regulation of this gene (Supplementary Table 1 and Figure 3). A recent study of miR- 19 has shown that it contributes to the regulation of newborn neuronal cell migration and is enriched in neural progenitor cells [54]. Several other miR binding sites within the 3’-UTR of human SULF1 have been reported with significant roles in regulating cell proliferation during carcinogenesis, including miR-26, miR-205, miR-130, miR-148, miR- 26, miR-1, miR-200, miR-140, miR-145, miR-17, miR-202, miR-433 and miR-137 (Supplementary Table 1). In addition, miR-202, located in the 3’-UTR of human SULF2, has been shown to inhibit the progression of human cervical cancer [55].

Figure 3: Comparative Tissue Expression for Human SULF1 and SULF2 genes. RNA-seq gene expression profiles across 53 selected tissues (or tissue segments) were examined from the public database for human SULF1 and SULF2, based on expression levels for 175 individuals [37] (Data Source: GTEx Analysis Release V6p (dbGaP Accession phs000424.v6.p1) (http://www.gtexportal.org). Tissues: 1. Adipose-Subcutaneous; 2. Adipose-Visceral (Omentum); 3. Adrenal gland; 4. Artery-Aorta; 5. Artery-Coronary; 6. Artery-Tibial; 7. Bladder; 8. Brain-Amygdala; 9. Brain-Anterior cingulate Cortex (BA24); 10. Brain-Caudate (basal ganglia); 11. Brain-Cerebellar Hemisphere; 12. Brain-Cerebellum; 13. Brain-Cortex; 14. Brain-Frontal Cortex; 15. Brain-Hippocampus; 16. Brain-Hypothalamus; 17. Brain- Nucleus accumbens (basal ganglia); 18. Brain-Putamen (basal ganglia); 19. Brain-Spinal Cord (cervical c-1); 20. Brain-Substantia nigra; 21. Breast-Mammary Tissue; 22. Cells-EBV-transformed lymphocytes; 23. Cells-Transformed fibroblasts; 24. Cervix-Ectocervix; 25. Cervix-Endocervix; 26. Colon-Sigmoid; 27. Colon- Transverse; 28. Esophagus-Gastroesophageal Junction; 29. Esophagus- Mucosa; 30. Esophagus-Muscularis; 31. Fallopian Tube; 32. Heart-Atrial Appendage; 33. Heart-Left Ventricle; 34. Kidney-Cortex; 35. Liver; 36. Lung; 37. Minor Salivary Gland; 38. Muscle-Skeletal; 39. Nerve-Tibial; 40. Ovary; 41. Pancreas; 42. Pituitary; 43. Prostate; 44. Skin-Not Sun Exposed (Suprapubic); 45. Skin-Sun Exposed (Lower leg); 46. Small Intestine-Terminal Ileum; 47. Spleen; 48. Stomach; 49. Testis; 50. Thyroid; 51. Uterus; 52. Vagina; 53. Whole Blood.

Comparative SULF1 and SULF2 human tissue gene expression

Figure 3 shows comparative gene expression for various human tissues obtained from RNA-seq gene expression profiles for human SULF1 and SULF2 genes obtained for 53 selected tissues or tissue segments for 175 individuals [37] (Data Source: GTEx Analysis Release V6p (dbGaP Accession phs000424.v6.p1) (http://www.gtex.org). These data supported a much higher level of tissue expression for human SULF1 in arterial and fibroblast cells, and for SULF2 in female reproductive tissues, including cervix, ovary, uterus and vagina. The presence of multiple TFBS within the SULF1 gene promoter (EGR1, FOXA1 and TFA2PC) and the SULF2 (ESR1, EGR1, HNF4A, CTCF and FOXA1) gene promoter may contribute to this high level in expression level for these genes. In addition, the presence of the binding site for the estrogen receptor (ESR1) within the SULF2 promoter, which is highly expressed in female reproductive tissues, is potentially of major significance for this enhanced SULF2 expression profile.

Phylogeny and Divergence of Vertebrate SULF1 and SULF2

A phylogenetic tree (Figure 4) was calculated by the progressive alignment of human and other vertebrate SULF1 and SULF2 amino acid sequences with an invertebrate (worm: Caeborhabditis elegans) sequence (SUL1). The phylogram was ‘rooted’ with this C. elegans SUL1 sequence and showed clustering of the SULF-like sequences into two groups: vertebrate SULF1 and SULF2 sequences. Overall, these data suggest that the vertebrate SULF1 and SULF2 genes arose from a gene duplication event of an ancestral invertebrate SULF-like gene, resulting in two separate lines of vertebrate gene evolution for SULF1-like and SULF2-like genes. This is supported by the comparative biochemical and genomic evidence for vertebrate SULF1 and SULF2-like genes and encoded proteins, which shared several key features of protein and gene structure, including having similar alpha-beta secondary structures (Figure 1). In addition, the locations of vertebrate SULF1 and SULF2 genes on separate chromosomes (Table 1) may reflect on a possible mechanism for ancestral vertebrate SULF gene duplication by whole-genome duplication rather than by an unequal crossover event of a single ancestral chromosome, as exemplified by studies supporting at least two rounds of whole genome duplication during early vertebrate evolution [56].

Figure 4: Phylogenetic tree of vertebrate SULF1 and SULF2 amino acid sequences. The tree is labeled with the gene name and the name of the animal. Note 2 major clusters for the vertebrate SULF1 and vertebrate SULF2 sequences. The tree is ‘rooted’ with the worm (Caenorhabditis elegans) SUL1 sequence. Table 1 for details of sequences and gene locations. A genetic distance scale is shown (% amino acid substitutions). The number of times a clade (sequences common to a node or branch) occurred in the bootstrap replicates is shown. Only replicate values of 0.9 or more which are highly significant are shown with 100 bootstrap replicates performed in each case. An evolutionary model for a proposed gene duplication event of an ancestral invertebrate SULF (SUL-1) gene is shown.

In conclusion, the results of the present study suggested that vertebrate SULF1 and SULF2 genes and encoded SULF1 and SULF2 enzymes represented a distinct arylsulfatase enzyme and gene family which share key conserved sequences and structures with those reported for other arylsulfatase gene families [1,2]. SULF1 has been recognized as a major extracellular sulfatase expressed in many tissues of the body, particularly in arterial and fibroblast cells, which plays a specific role in removing sulfate from heparan sulfate proteoglycan extracellular substrates, catalysing the hydrolysis of endoglucosamine-6-sulfate residues. SULF2 has also been described as a second major extracellular sulfatase expressed in many tissues of the body, particularly in female reproductive tissues, also with a specific endoglucosamine-6-sulfatase role [3]. Bioinformatic methods were used to predict the amino acid sequences, secondary and tertiary structures and gene locations for SULF1 and SULF2 genes and encoded proteins using data from several vertebrate genome projects. Vertebrate SULF protein subunits shared 59-93% sequence identities and exhibited sequence alignments and identities for key SULF amino acid residues as well as conservation of predicted secondary structures with those previously reported for a bacterial phosphonate monoester hydrolase from Silicibacter pomeroyi (PDB:4upk). Phylogenetic analyses demonstrated the relationships and potential evolutionary origins of the vertebrate SULF1 and SULF gene families which were related to a worm (Caeborhabditis elegans) extracellular sulfatase (SUL1) gene and protein. These studies indicated that SULF1 and SULF2 genes may have appeared early in vertebrate evolution following gene duplication of an ancestral SUL-like gene, following whole-genome duplication in the vertebrate ancestor.