Synthesis of the genes for the enzymes PAPS synthase 1 and sulfotransferase C4ST-1 and their subsequent cloning into plasmids pET-28b(+) and pET-32b(+), respectively, was performed by GenScriptTM (Piscataway, NJ). The original nucleotide sequence of both enzymes was optimized to implement their expression in E. coli (OptimumGene™ algorithm). E. coli strain BL21 (DE3) was obtained from Promega Biotech Ibérica S.L. (Madrid, Spain). Plasmid purification was performed using the GenElute™ Plasmid Miniprep Kit from Sigma-Aldrich (Darmstadt, Germany). Promega 1 kb DNA Step Ladder Molecular Weight Markers and restriction enzymes were purchased from ThermoFisher Scientific Inc. (Waltham, MA). Kanamycin, pyruvate kinase/lactic dehydrogenase mixture, pyrophosphatase, DNase I, and other reagents were purchased from Sigma-Aldrich (Darmstadt, Germany). Isopropyl-β-d-thiogalactopyranoside (IPTG) was purchased from Applichem GmBH (Darmstadt, Germany). Imidazole and NADH were purchased from Thermo Fisher Scientific Inc. (Waltham, MA). Bradford reagent and acrylamide/bis-acrylamide solution of 30% (29:1) used for protein analysis by SDS-PAGE were purchased from Bio-Rad (Hercules, CA). The Low Molecular Weight Calibration Kit from GE Healthcare was used. SDS-PAGE Gels were run on a MiniProtean^® Tetracell cuvette from Bio-Rad (Hercules, CA). Agarose gels were run in a RunOne™ Electrophoresis Cell cuvette from EmbiTec (San Diego, CA) using SYBR Safe dye from ThermoFisher Scientific (Waltham, MA) for staining DNA in electrophoresis gels. Densitometric analysis of acrylamide genes was performed using a Gene Flash Bio Imaging Photodocumenter from Syngene Ltd. (Bengaluru, Karnataka) and GeneTools 3.07 software. Spectrophotometric assays were carried out in a UV-Visible SPECTRAmax-384 PLUS, from Molecular Devices, LLC (San José, CA). Iminodiacetic acid agarose (IDA-agarose) was purchased from Agarose Bead Technologies (Miami, FL). Solvents were of analytical grade. DNA manipulation was performed according to standard procedures (Sambrook et al., 1989). Unsulfated chondroitin (CS-0S) was chemically prepared following the procedure previously described in Benito-Arenas et al. (2018). ¹H-NMR was performed at room temperature using deuterated water (D₂O) as solvent in a Varian INOVA-500 (¹H 500 MHz) instrument. Chemical shift values are given in parts per million (δ, ppm). In addition, the two-dimensional NMR experiment Heteronuclear Single Quantum Correlation (HSQC) was performed to further confirm the signal assignments. The MestReNova v.11.0.1-17801 program was used to process all NMR spectra, both one-dimensional and two-dimensional.

Cloning of papss1 gene was performed in the expression plasmid pET-28b(+), which allows induction of protein overexpression with IPTG. In addition, the protein is expressed fused to a 6-His tag at the N-terminal end, allowing its one-step purification by ion metal affinity chromatography (IMAC) from cell-free extract (CFE). The plasmid, designated pET-28b(+)-papss1, was then transformed into electrocompetent E. coli BL21 (DE3) cells. The transformed cells were growth in LB agar plates containing 30 μg/ml of the antibiotic kanamycin. The plates were kept overnight at 37°C. Three randomly selected colonies were grown in 5 ml LB containing kanamycin (30 μg/mL) for plasmid purification. The purified plasmids were subjected to a double digestion with the restriction enzymes EcoRI and HindIII in a reaction volume of 10 μl according to the protocol specified by the manufacturer. The different DNA fragments were analyzed by electrophoresis in .8% agarose gel in TAE (40 mM Tris-acetate, 1 mM EDTA, pH 8.0). Visualization of the gel with SYBR Safe dye revealed two bands (Supplementary Figure S1): a band of 5,368 bp corresponding to plasmid pET-28-b(+) and a band of 1,872 bp corresponding to the expected size of the papss1 gene.

For PAPSS1 enzyme expression, preinocula of recombinant colonies were prepared in 5 ml LB medium containing 30 μg/ml kanamycin and incubated overnight at 37°C with shaking (160 rpm). The culture was extended and incubated under the same conditions until the exponential phase (OD_600nm = .5–.7) was reached. At this point, protein expression was induced with IPTG at a final concentration of 1 mM. Orbital agitation was kept constant, and the temperature was lowered to 30°C to avoid the formation of inclusion bodies. The culture was then centrifuged at 2,500 × g for 20 min at 4°C to collect the cells. The pellet was resuspended in Na₂HPO₄ buffer (50 mM, 300 mM NaCl, pH = 8.0). Cell disruption was performed by sonication (70% amplitude, 5 s pulse and 20 s pause between pulses). The lysed cells were separated by centrifugation of the mixture at 8,000 rpm for 20 min. The supernatant obtained was treated with DNaseI (10 μg/ml cells) and MgCl₂ (.95 μg/ml cells) on ice for 20 min. Streptomycin (1% wt/vol) was then added and allowed to act for an additional 20 min to precipitate the resulting nucleotides. Finally, the cell-free extract (CFE) containing the soluble protein fraction (separated from precipitates and inclusion bodies) was obtained by centrifugation at 8,000 rpm for 20 min at 4°C.

Purification of the enzyme was carried out according to two strategies. First, it was performed by affinity chromatography with divalent metals (IMAC) using high-density Co₂ ⁺-IDA-agarose resin. The resin was packed into a propylene chromatography column using a suitable column filter. The resin was washed and equilibrated with 50 mM Na₂HPO₄, 300 mM NaCl, 10 mM imidazole, pH 8.0 buffer. The CFE was added to the resin at a ratio of 1:1 (v/v). The resin was washed with 3 volumes of the same buffer to eliminate non-specific binding proteins. Finally, the protein was eluted with a column volume of 50 mM Na₂HPO₄, 300 mM NaCl, 2 M imidazole, pH 8.0. Subsequently, the imidazole was removed from the medium by ultrafiltration so that it was exchanged for 50 mM Na₂HPO₄ pH 8.0 buffer. To achieve an even higher degree of purity, enzyme purification by size exclusion chromatography was performed. HiLoad 26/60 Superdex 200 PG size exclusion columns were used, controlled by a AKTA-FPLC system from GE Healthcare Life Science. 50 mM Na₂HPO₄ and 150 mM NaCl buffer at pH 7.2 with an isocratic flow of 1.0 ml/min were used as mobile phase. Fractions containing PAPSS1 were pooled and concentrated. All fractions were analyzed by SDS-PAGE on 11% acrylamide/bis-acrylamide gels and by spectrometric activity and protein concentration determination.

The activity of recombinant PAPSS1 was measured using a PK/LDH-coupled spectrophotometric assay that allows monitoring of ADP generated during APS phosphorylation by the concomitant oxidation of NADH. This assay allows not only spectrophotometric monitoring of the reaction, but also regeneration of ATP consumed by APS kinase, thus avoiding its inhibition by ADP. The enzyme pyruvate kinase (PK) catalyzes the transfer of a phosphate group from phosphoenolpyruvate (PEP) to ADP formed in the APS phosphorylation reaction. At the same time, the pyruvate formed is reduced to lactate by LDH consuming NADH, which can be measured spectrophotometrically (Figure 2).

Reactions were carried out in 1 ml of Tris-HCl buffer (50 mM pH = 8.0) containing ATP (12.5 μmol), K₂SO₄ (15 μmol), MgCl₂ (25 μmol), PEP (1.0 μmol), NADH (.3 μmol), and the enzymes LDH (5 μg/ml), PK (5 μg/ml), and PAPSS1 (126.5 μg/ml). Because of the strong inhibition of ATP sulfurylase activity by PPi, PPase (10 μg/ml) was also added to degrade the pyrophosphate formed (Figure 2). In parallel, reactions were monitored by HPLC using a Dionex PDA-100 chromatograph with a C18 Discovery column (250 mm × 4.6 mm; Ø = 5 μm). The mobile phase used was 20 mM Na₂HPO₄ buffer pH 6.0 with isocratic flow .5 ml/min and detection at 260 nm. Initially, the separation method was optimized using standards for ATP and ADP (Supplementary Figure S2). Reactions were monitored for 18 h, with aliquots taken at 15 min, 30 min, and 18 h. As a control, similar reactions were analyzed but without the ATP regeneration system (PK).

For the cloning of the C4ST-1 enzyme from R. norvegicus, the procedure described by He et al. (2017) for the cloning of the human enzyme was followed as the sequence of the R. norvegicus enzyme shows 96.88% similarity to the human C4ST-1 sequence. The first 60 amino acids of the C4ST-1 protein (1Met-60Leu) were removed because they have been identified as a putative transmembrane domain (He et al., 2017). The chst11 gene modified in this way was cloned between the restriction enzymes BamHI and XhoI of the plasmid pET-32b(+). This plasmid allows the N-terminal end of C4ST-1 to be fused to TrxA to increase the solubility of the recombinant protein, and additionally to a 6-histidine peptide to allow its purification by IMAC chromatography. The resulting pET-TrxA-chst11 plasmid was transformed into competent E. coli BL21 (DE3) cells which were incubated on LB-agar plates with 250 μg/ml ampicillin. The plates were kept at 37°C overnight. Three randomly selected colonies were grown in 5 ml of LB and ampicillin (250 μg/ml) to purify their plasmids and analyzed by double digestion with the restriction enzymes BamHI and XhoI in a reaction volume of 10 μl according to the protocol indicated by the manufacturer. The different DNA fragments were analyzed by electrophoresis in agarose gel at .8% in TAE (Tris-acetate 40 mM, EDTA 1mM, pH = 8.0). Two bands were observed in each of the colonies (Supplementary Figure S3): one of the bands, with an approximate size of 6,000 bp, corresponds to the plasmid pET32b(+); the other band, with an approximate size of 900 bp, is consistent with the expected size for the chst11 gene encoding the C4ST-1 protein (892 bp).

Overexpression of the C4ST-1 enzyme was performed from a colony containing the chst11 gene. The clone was incubated at 37°C with shaking at 160 rpm in LB medium containing 250 μg/ml ampicillin until an OD_600nm between .5–.7 was reached. At this point, expression was induced with .2 mM IPTG and the temperature was reduced to 22°C with the culture shaken at 160 rpm all night. Protein extraction followed the same steps described above for the PAPSS1 enzyme until the cell-free extract was obtained.

The coupling of the PAPS generation module and the sulfation module was performed using a desulfated chondroitin (CS-0S) previously obtained in our laboratory as a substrate to be sulfated (Benito-Arenas et al., 2018). The PAPSS1 enzyme was used after purification by IMAC, whereas the C4ST-1 enzyme was used without purification. For the PAPS generation module, ATP (12.5 μmols), K₂SO₄ (15 μmols), MgCl₂ (25 μmols), PEP (1 μmol), NADH (.3 μmols), and LDH enzymes (5 μg/ml, PK (5 μg/ml), pyrophosphatase (10 μg/ml), and PAPSS1 (126.5 μg/ml) were used. The sulfation module consisted of 6.0 mg/ml CS-0S and a volume of 25 μl of CFE of enzyme C4ST-1. The reaction was performed in 1 ml of 50 mM Tris-HCl buffer at pH = 8.0. The reaction was maintained for 72 h with orbital shaking at room temperature. Purification of the product of coupling the two modules was performed by dialysis against distilled water using a membrane with a cut-off of 3.5 KDa for 24 h. The resulting solution was lyophilized to give the product as a white solid. The product was identified by proton nuclear magnetic resonance (¹H-NMR) experiments at room temperature using deuterated water (D₂O) as solvent and two-dimensional heteronuclear single quantum correlation (HSQC) NMR experiments. Chemical shift values were compared to those previously described for authentic samples of the product (Benito-Arenas et al., 2018).

When PAPSS1 protein overexpression was analysed, the appearance of a band with a molecular mass of approximately 72 kDa was observed, which corresponds to the expected mass of the PAPSS1 protein. Although a large fraction of the recombinant protein is insolubly expressed in the form of inclusion bodies, PAPSS1 in the soluble fraction accounts for 40.6% of the total proteins present, making it the major protein (Figure 3A). As mentioned previously, PAPSS1 was purified using two alternative chromatographic techniques: IMAC and size exclusion chromatography. A high degree of purity was achieved with both techniques. IMAC achieved a purity of 85.5%, whereas size exclusion chromatography yielded an even higher purity of 95.7% (Figure 3B). Although better results were obtained with size exclusion chromatography than with IMAC, the latter technique was used for successive purifications of protein from CFE. This was mainly because of the simplicity of the method and the shorter time required to obtain the pure protein compared with size exclusion chromatography. In addition, no non-specific interactions were observed for the remaining 14.5% of the protein in reactions performed with IMAC-purified PAPSS1.

The activity of PAPSS1 was monitored spectrophotometrically thanks to the coupling of the PK/LDH system, which also allows the regeneration of ATP used for the phosphorylation of APS catalyzed by the APS kinase activity of PAPSS1. The reaction was initiated by adding the K₂SO₄ substrate to the reaction medium (Figure 4).

The decrease in absorbance at 340 nm due to the oxidation of NADH allowed us to indirectly determine the amount of PAPS formed. In this way, it was determined that .06 μmol of PAPS was formed (Figure 5A), corresponding to a yield of 4.8%, whereas only .03 μmol of ATP was consumed (Figure 5B). These results indicate that regeneration of ATP consumed during phosphorylation of APS occurs, thus regenerating at least one of the two equivalents used in the synthesis of PAPS.

To analyze the effect of ATP regeneration on PAPSS1-catalyzed PAPS synthesis, two reactions of PAPS synthesis were followed for 18 h. In one of the reactions, the ATP regeneration system was coupled (Figure 2), whereas in the other this system was eliminated. Since this second reaction could not be analyzed spectrophotometrically, both reactions were analyzed by HPLC (Figure 6). In the reaction in which the ATP regeneration module was eliminated (Figure 6A), a peak with a retention time of 14.8 min appears, corresponding to the retention time of the ADP pattern, and the peak corresponding to PAPS is practically missing. On the other hand, in the chromatogram of the reaction with the coupled ATP regeneration system (Figure 6B), the appearance of the peak corresponding to ADP was not observed and the peak corresponding to PAPS increased significantly.

These results confirm that the PK/LDH system is efficient in ATP regeneration and, furthermore, that this regeneration is necessary for the synthesis of PAPS because the APS kinase activity of the PAPSS 1 enzyme is strongly inhibited competitively by APS (Yanagisawa et al., 1998; Harjes et al., 2004; Lansdon et al., 2004; Harjes et al., 2005; Sekulic et al., 2007; Mueller and Shafqat, 2013). This inhibition occurs through the formation of a dead-end complex of the enzyme with the APS and the ADP resulting from the phosphorylation reaction. The formation of this complex is favored by the fact that the affinity of the enzyme-ADP complex for APS is greater than that showed by the apoenzyme (Mueller and Shafqat, 2013).

SDS-PAGE analysis of the soluble and insoluble fractions of the cell-free extract obtained from the E. coli strain transformed with the plasmid pET32b(+)-chst11 revealed a major protein in the insoluble fraction with a molecular mass of approximately 50 kDa, consistent with the size previously estimated for the C4ST-1 enzyme (54 kDa). Unfortunately, the majority of the recombinant protein was present in the form of inclusion bodies (Figure 7; lanes 2–4), whereas its expression in the soluble fraction was low (Figure 7; lanes 5–7). These results are consistent with others previously reported in the literature, such as mouse C4ST-1, which had to be expressed in P. pastoris due to the null soluble expression in E. coli (Zhou et al., 2018). The difficulty in expressing this enzyme in E. coli may be due in part to the fact that N-linked oligosaccharides bound to C4ST-1 contribute to the production and stability of the active form of the enzyme (Yusa et al., 2005).

Given the low expression of the enzyme in the soluble fraction, it was decided to use the soluble fraction of the cell-free extract directly for the sulfation reactions of CS-0S without further purification.

Once both the PAPS generation module with the regeneration system of one of the ATP molecules used in its synthesis and the enzyme C4ST-1 responsible for catalyzing the regioselective sulfation of position 4 of the N-acetylglucosamine unit (sulfation module) were available, the two modules were coupled as shown schematically in Figure 8. In this way, the PAPS generated by the PAPSS1 enzyme from ATP and inorganic sulfate can be used in situ by the C4ST-1 enzyme to perform the sulfation of chondroitin in a single step without the need for purification of the reaction intermediates.

After purification of the reaction product by dialysis followed by freeze-drying, it was analyzed by NMR. Analysis of disaccharides was discarded because a complete structural characterization of various chondroitin sulfates, including CS-0S and CS-4S, had already been performed in our previous work (Benito-Arenas et al., 2018). This characterization included determination of the composition of the polysaccharides by enzymatic digestion with chondroitin lyase ABC, followed by HPLC analysis and comparison with authentic samples. In addition, preparative HPLC analysis was performed coupled with mass spectroscopy detection. Finally, the samples were analyzed by NMR spectroscopy (¹H and ¹³C NMR spectra). In this way, we had authentic and confirmed samples of both CS-0S and CS-4S with which to compare the results of our NMR experiments. Under these conditions, the agreement of the chemical shifts of our samples with those of the authentic samples described in our previous work serves to confirm unequivocally the identity of the products of the sulfation reaction described in our manuscript (Table 1). Thus, two compounds were identified, the minor compound corresponding to chondroitin sulfated in position 4 of N-acetylglucosamine (CS-4S).

In the ¹H-NMR spectrum (Figure 9A), the characteristic signals of the chondroitin disaccharide unit are observed, and the presence of sulfated chondroitin cannot be confirmed. However, if position 4 of the N-acetylglucosamine subunit of the disaccharide were sulfated, the signal corresponding to this position would not be screened and would appear below the signal for deuterated water from the solvent. Therefore, it cannot be concluded from the ¹H-NMR spectrum whether sulfation of the compound has occurred. Therefore, a 2D-NMR HSQC experiment was performed (Figure 9B). In this experiment, a minor correlation (δ ¹H = 4.77 δ ¹³C = 77.3 ppm) attributable to position 4 of the sulfated N-acetylglucosamine subunit was observed, confirming that the developed modular cascade was able to catalyze the regioselective sulfation of chondroitin generating in situ the PAPS from ATP and inorganic sulfate.