Modulation of receptor binding to collagen by glycosylated 5‐hydroxylysine: Chemical biology approaches made feasible by Carpino's Fmoc group

Abstract

The creation of the 9-fluorenylmethoxycarbonyl (Fmoc) group by the Carpino laboratory facilitated the synthesis of peptides containing acid-sensitive groups, such as O-linked glycosides. To fully investigative collagen biochemistry, one needs to assemble peptides that possess glycosylated 5-hydroxylysine (Hyl). A convenient method for the synthesis of Fmoc-Hyl(ε-tert-butyloxycarbonyl (Boc),O-tert-butyldimethylsilyl (TBDMS)) and efficient methods for the synthesis of Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] have been developed. Glycosylated Fmoc-Hyl derivatives were used to construct a series of types I-IV collagen-model triple-helical peptides (THPs) that incorporated known or proposed receptor binding sites. Glycosylation of Hyl was found to strongly down-regulate the binding of CD44 and the α3β1 integrin to collagen, while the impact on α2β1 integrin binding was more modest. Molecular modeling of integrin binding indicated that Hyl glycosylation directly impacted the association between the α3β1 integrin metal ion-dependent adhesion site (MIDAS) and the receptor binding site within type IV collagen. The Fmoc solid-phase strategy ultimately allowed for chemical biology approaches to be utilized to study tumor cell interactions with glycosylated collagen sequences and document the modulation of receptor interactions by Hyl posttranslational modification.

Graphical Abstract

INTRODUCTION

The collagen family of proteins includes at least 28 different types.^1–5 The common feature to all collagens is the formation of a three stranded, right-handed superhelix. The superhelix incorporates three α chains of primarily Gly-Xaa-Yaa repeating triplets. The Gly-Xaa-Yaa repeating sequence induces the individual α chains to adopt left-handed polyPro II helices, which in turn intertwine to form the right-handed superhelix. The stability and biological activity of collagen is dependent on several posttranslational modifications. Lys residues within collagens are often posttranslationally modified to 5-hydroxylysine (Hyl).^6–8 Subsequent posttranslational glycosylation at the 5-hydroxyl group results in the attachment of either a β-D-galactopyranosyl monosaccharide or an α-D-glucopyranosyl-(1→2)-β-D-galactopyranosyl disaccharide (Figure 1).^8–14 Hyl is the major glycosylation site within collagens. Hyl glycosylation directly impacts (a) collagen fibril assembly, as carbohydrate content modulates fibril diameter,^6,12,15–17 (b) cross-linking in fibrillar collagens, as the mature trivalent cross-links are primarily monoglycosylated,¹³ (c) mineralization of fibrillar collagens, as collagen with lower levels of glycosylation exhibit delayed mineralization,^12,17 and (d) secretion, assembly, and distribution of types IV and VI collagens and assembly of the basement membrane, as all of these processes are impaired by underglycosylation.^18,19 It has been further suggested that collagen glycosylation patterns may impact autoimmune diseases such as rheumatoid arthritis and systemic sclerosis,⁸ based on T-cell responses to a specific glycosylated collagen fragment (see also below).²⁰

Figure 1.

Structure of glycosylated 5-Hyl.

Glycosylated Hyl is critical for the insulin-sensitizing activity of adiponectin, a Clq family member that possesses an N-terminal collagen-like domain.^21,22 5-Hyl is found in histones as a result of Jumonji domain containing 6 (JMJD6) hydroxylating Lys residues.²³ The Lys hydroxylation inhibits N-acetylation and N-methylation, potentially impacting epigenetic regulation.²³

Studies of Lys hydroxylation and subsequent glycosylation, as well as the role of glycosylated Hyl in collagen assembly and receptor recognition of triple-helices, can be greatly advanced by the development of efficient methods for the synthesis of Hyl-containing peptides. The mild conditions of 9-fluorenylmethoxycarbonyl (Fmoc) chemistry are, in general, better suited for glycopeptide synthesis than tert-butyloxycarbonyl (Boc) chemistry, as repetitive acid treatments of the latter can be detrimental to sugar linkages.²⁴ The Fmoc group was developed in the laboratory of Louis Carpino in the early 1970s.^25–28 The adaptation of the Fmoc group for solid-phase synthesis was subsequently championed by the groups of Johannes Meienhofer (1929–1993) and Robert (Bob) Sheppard (1932–2019) in the later 1970s and early 1980s.^29–42 The use of Fmoc chemistry has allowed our research group to assemble collagen-model sequences that incorporate Hyl and glycosylated Hyl and use these sequences to study cell surface receptor interactions with collagen.

SYNTHESIS OF HYL DERIVATIVES

Hyl is typically obtained commercially as a D/L mixture. Methods have been described for isolating or synthesizing the desired (2S,5R)-5-Hyl.^22,43,44 The subsequent synthesis of Hyl derivatives requires several considerations.⁴⁵ The first consideration is the solubility of the intermediates generated during Hyl derivative assembly. Partial protection of Hyl results in a compound that has both hydrophilic and hydrophobic components and thus has poor solubility in both organic and aqueous environments.^46,47 The second consideration is that activation of a Hyl derivative that does not incorporate side-chain protection of the hydroxyl group leads to intramolecular lactone formation.^45,48–50 The Hyl lactone form can be used for synthesis of dipeptides and tetrapeptides in solution,^48,49,51,52 but it is poorly incorporated when assembling longer peptides by solid-phase synthesis approaches. The third consideration is the method by which copper complexed Hyl derivatives can be disrupted. The solubility problems described above are observed with copper complexed Hyl derivatives. Although copper complexed amino acid derivatives may be disrupted by Chelex 100 resin (in either Na⁺ or H⁺ form),^53,54 ethylenediaminetetraacetic acid,^55,56 thioacetamide,^57,58 thioureas,⁵⁹ or tetrahydrothiazole-2-thione,⁶⁰ solvent systems specific for copper complexed Hyl derivatives must be developed to allow for the copper to be removed efficiently.

Several groups have described the syntheses of hydroxyl-protected and glycosylated Hyl derivatives for incorporation into peptides (see below). Also described are preparation of analogs of Hyl, such as 4-hydroxylysine,⁶¹ N⁶-hydroxylysine,⁶² and Nε-acetyl-Nε-hydroxylysine,⁶³ and analogs of glycosylated Hyl, such as disaccharide-containing 5-hydroxynorvaline⁶⁴ and monosaccharide-containing dihydroxynorleucine,⁶⁵ (2S,5S)-5-hydroxylysine,⁶⁵ (2S,5S)-5-azido-6-hydroxynorleucine,⁶⁵ and (2S,5R)-5-hydroxy-5-methyllysine.⁶⁵

SYNTHESIS OF FMOC-HYL DERIVATIVES

Initial syntheses of Hyl-containing peptides utilized the same protection strategy for both the α- and ε-amino groups.^46,48,51 Our group sought to synthesize O-protected Fmoc-Hyl in a convenient fashion.⁶⁶ tert-butyldimethylsilyl (TBDMS) protection of the Hyl 5-hydroxyl group was utilized, as the TBDMS group can be introduced easily to secondary alcohols and is well-suited for Fmoc chemistry, being base stable and acid labile.^67–69 Prior to TBDMS introduction, the amino and carboxyl functionalities of Hyl were protected in a copper complex. Boc₂O was utilized to protect the ε-amino group in the complex (Figure 2). Addition of TBDMS-Cl with imidazole catalysis in N,N-dimethylformamide (DMF)⁶⁷ to incorporate the TBDMS group was not successful, partially due to disruption of the Boc-protected copper complex. A prior study demonstrated sucessful incorporation of TBDMS to a hydroxyl group using TBDMSOTf and 2,6-lutidine as base in CH₂Cl₂.⁷⁰ However, the Boc-protected copper complex had low solubility in CH₂Cl₂. Use of pyridine as the solvent and 4-dimethylpyridine (DMAP) as the base proved to be successful. Acylation of sterically hindered secondary and tertiary alcohols is effectively catalyzed by DMAP.⁷¹

Figure 2.

Synthesis of Fmoc-Hyl(ε-Boc,O-TBDMS). Reproduced by permission of John Wiley and Sons.⁶⁶

Disruption of the Boc-protected complex was achieved as described⁵³ using a pyridine:H₂O mixture and Na⁺ Chelex 100 resin (Figure 2). Treatment of the product with Fmoc-OSu followed by flash chromatography purification provided Fmoc-Hyl(ε-Boc,O-TBDMS) in 67% overall yield (Figure 2). The purity and composition of Fmoc-Hyl(ε-Boc,O-TBDMS) was confirmed by NMR and mass spectra.⁶⁶

Synthesis of Fmoc-Hyl(ε-Boc,O-TBDMS) has been previously described.⁷⁰ The initial steps, (a) copper complexation of Hyl and (b) treatment with Boc₂O to protect the Nε-amino group of copper-complexed Hyl, were similar to our synthetic scheme (Figure 2). However, the prior method had several additional steps, as TBDMSOTf treatment was performed only after (a) disruption of the copper complex, (b) Fmoc protection of the Nα-amino group by treatment with 9-fluorenylmethyl chloroformate, and (c) protection of the carboxyl group by treatment with Bzl-Br. After the addition of the TBDMS group, the Bzl group was removed via hydrogenation. The overall yield was 49%. A significant improvement in our method was the direct incorporation of the TBDMS group to copper complexed Hyl(ε-Boc) using pyridine as the solvent. This “one pot” approach has been used previously for synthesis of Lys and Hyl derivatives,^56,72 including one in which Fmoc-Hyl(ε-Cbz,O-TBDMS) (where Cbz is benzyloxycarbonyl) was synthesized using 9-borabicyclononane complexation of Hyl.⁷³ One pot approaches are highly dependent upon obtaining appropriate solvent conditions. Preparation of Hyl derivatives has also been achieved using precursors such as N^α-Cbz-Lys-OCH₃ or Boc-Asp-OtBu, but these derivatives incorporate the hydroxyl functionality at either the N^ε-amino or 4-position rather than the 5-position.^61,63

SYNTHESIS OF MONOGLYCOSYLATED FMOC-HYL DERIVATIVES

Methods for preparing glycosylated Hyl have been reviewed previously.^22,45 Our group pursued several methods for the synthesis of glycosylated Fmoc-Hyl derivatives.^66,72 We found that either the allyloxycarbonyl (Aloc) or Boc group could be utilized for Hyl side-chain protection, but Cbz was not appropriate due to issues of derviative solubility.⁴⁷ Copper complexation of Hyl was followed by Aloc or Boc protection of the Hyl ε-amino group (Figure 3). 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide was added to the copper complex of Hyl(ε-Boc) (Figure 3). Na⁺ ion exchange resin was used to remove the copper, and Hyl(ε-Boc,O-2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)-OH was eluted with CH₃OH−H₂O (1:1). Hyl(ε-Boc,O-2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)-OH was then reacted with Fmoc-OSu in H₂O-acetone in the presence of NaHCO₃ (Figure 3). Fmoc-Hyl(ε-Boc,O-2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl) was synthesized in overall 29% yield.

Figure 3.

Synthesis of Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)]. Reproduced by permission of Elsevier.⁷²

Although effective, one difficulty with the above method was the marginal CH₃CN solubility of the various reactants, which made the yield of the desired product highly variable. The efficiency of the Koenigs-Knorr, thioglycoside, and trichloroacetimidate methods for the preparation of glycosylated Fmoc-Hyl derivatives was compared in order to find the optimal synthetic route.⁶⁶ Fmoc-Hyl(ε-Boc)-OBzl was produced in 60% yield by (a) preparing the Boc-protected copper complex as described above, (b) using H⁺ Chelex 100 resin in MeOH:H₂O to remove the copper, (c) dissolving Hyl(ε-Boc) and NaHCO₃ in H₂O, (d) adding Fmoc-OSu predissolved in acetone to produce Fmoc-Hyl(ε-Boc), and (e) dissolving Fmoc-Hyl(ε-Boc) in CsCO₃:EtOH and reacting with Bzl-Br (Figure 4).

Figure 4.

Synthesis of Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] using the inverse Koenigs-Knorr procedure. Reproduced by permission of John Wiley and Sons.⁶⁶

Starting from Fmoc-Hyl(ε-Boc)-OBzl, the most efficient synthetic route was found to be the “inverse” Koenigs-Knorr procedure. 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide and Fmoc-Hyl(Boc)-OBzl were added to a suspension of AgOTf in 1,2-dichloroethane at −15 °C, and the reaction proceeded for 15 min. 2,6-di-tert-butyl-4-methylpyridine was added and the β-glycoside was isolated in 45% yield (Figure 4). Fully protected glycosylated Hyl underwent catalytic hydrogenation at room temperature in EtOAc, producing Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] in 61% yield (Figure 4). Unlike a prior report,⁷⁰ the Fmoc group was not removed under these conditions. The purity and composition of Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] was confirmed by NMR and mass spectra.

In prior studies, Hyl glycosylation was achieved by treatment of a protected Hyl derivative with (a) 2-O-levulinyl-3,4,6-tri-O-acetyl-α-D-galactopyranosyl bromide or 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide and Hg(CN)₂ in dry toluene^{49,52,74–76} or (b) 2-O-chloroacetyl-3,4,6-tri-O-benzyl-α-D-galactopyranosyl trichloroacetimidate and TBDMSOTf in diethyl ether.^77,78 These approaches did not yield amino acid building blocks compatable with solid-phase methodology.^{49,52,74–78}

Glycosylation of protected Hyl to create an appropriate building was achieved by treatment with 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide or 2,3,4,6-tetra-O-pivaloyl-α-D-galactopyranosyl bromide and silver silicate in CH₂Cl₂.^65,73,79 Unfortunately, inadequate stability of the ε-Boc group resulted in low yields⁷⁹ or significant orthoester formation occurred.⁶⁵ With the Koenigs-Knorr inverse procedure, there was no observed loss of the Boc group or orthoester formation.

Glycosylation has also been achieved by protected Hyl treatment with 6-O-(tert-butyldiphenylsilyl)-3,4-O-isopropylidene-D-galactal, dimethyldioxirane, and ZnCl₂ in acetone:CH₂Cl₂:tetrahydrofuran.^50,70 This approach was based on (a) preparation of the α−1,2-anhydrosugar by epoxidation of the corresponding glycal and (b) Lewis acid-catalyzed opening of α−1,2-anhydrosugars.^70,80 A desired result of this reaction was to create both a β-glycosidic linkage and a free hydroxyl group at the C-2 center of the sugar for further attachment of an α-D-glucosyl moiety. The α−1,2-anhydrosugar was obtained by galactal epoxidation with dimethyldioxirane, followed by ZnCl₂ promotion to yield the β-galactoside.⁷⁰ Cbz ε-amino group protection resulted in an anomeric mixture (β/α = 3.8:1) in 51% yield⁵⁰. Unfortunately, as mentioned earlier, lactone formation is more readily obserived with Hyl(ε-Cbz) derivatives than the corresponding Hyl(ε-Boc) derivatives,^45,50 and the 2-O-acetylated donor leads to orthoester formation.⁸¹ The Koenigs-Knorr inverse procedure avoids side reactions such as lactone formation, loss of the Boc group, or orthoester formation.

Lactone and orthoester formation were also not observed when the 9-borabicyclononane complex of Hyl(ε-Cbz) was reacted with 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide in CH₂Cl₂ under silver silicate promotion in a one pot Koenigs-Knorr glycosylation.⁷³ Unfortunately, Fmoc-Hyl[ε-Cbz,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] is not as conveniently deprotected following solid-phase synthesis as Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)].

SYNTHESIS OF DISACCHARIDE-CONTAINING FMOC-HYL DERIVATIVES

Cbz group Hyl ε-amino protection was utilized to assemble a α-D-glucopyranosyl-(1→2)-β-D-galactopyranosyl Hyl building block.⁵⁰ To create the disaccharide, the thioethyl glucoside was reacted with glycosylated Hyl to achieve α-galactosylation. Coupling to the glycosylated Hyl HO-2 was facilitated by activation with N-iodosuccinimide and silver trifluoromethane sulfonate in CH₂Cl₂. The desired glycoside was obtained in 47% yield from an anomeric mixture of glycosides (α/β = 3.3:1). Deprotection gave a diglycosylated Hyl suitable for solid-phase peptide synthesis.

Disaccharide-containing Hyl was prepared in solution phase synthesis using the Cbz-protected lactone of (2S,5R)-5-Hyl.^49,52 The 1,2-trans-galactosidic linkage was achieved by neighboring group participation from the levulinyl group. Removal of the levulinyl group by mild hydrazinolysis provided the corresponding alcohol without affecting other esters. This alcohol was coupled with 1-O-(N-methyl)-actimidyl-2,3,4,6-tetra-O-benzyl-β-D-glucopyranose to form the O-(2-O-α-D-glucopyranosyl)-β-D-galactopyranoside of Hyl-Gly.^49,52 Preparation of disaccharide-containing Hyl has also been achieved by coupling protected 5-Hyl with α-trichloroacetimidates.^22,78

SYNTHESIS OF HYL- AND GLYCOSYLATED HYL-CONTAINING PEPTIDES

Fmoc-Hyl(ε-Boc,O-TBDMS) was used for the automated Fmoc solid-phase synthesis^66,82 of Thr-Pro-Gly-Glu-Hyl-Gly-Glu-Hyl-Gly-NH₂ (a potential GLT25D1 substrate⁸³), Gly-Tyr-Hyp-(Gly-Pro-Hyp)₂-Gly-Leu-Hyp-Gly-Ala-Hyl-Gly-Glu-Ala-Gly-Leu-Hyp-(Gly-Pro-Hyp)₃-NH₂ (a potential GLT25D1 substrate⁸³), and Gly-Phe-Hyp-Gly-Leu-Hyp-Gly-Ala-Hyl-Gly-Glu-NH₂ (a potential lysyl hydroxylase product⁸⁴). Fmoc-Hyl(ε-Boc,O-TBDMS) was coupled using 2-fold molar excesses of Fmoc-amino acid and 1-hydroxybenzotriazole (HOBt), a 1.8-fold molar excess of 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU), and a 4-fold molar excess of N,N-diisopropylethylamine (DIEA). Peptide-resin cleavage and side-chain deprotection was achieved with H₂O-trifluoroacetic acid (TFA) (1:19) or H₂O-thioanisole-TFA (1:1:18) for 2 h. Peptides were purified by preparative RP-HPLC, and the products were homogeneous by analytical RP-HPLC. MALDI MS analysis gave the desired masses for the three peptides (Table 1).

Table 1.

Sequences, analytical data, and thermal stabilities of Hyl-containing collagen-model peptides.

Peptide designation	Sequence	Receptora	[M+H]+ observed (calcd.)	Tm (°C)a
GLT25D1 substrate	Gly-Phe-Hyp-Gly-Leu-Hyp-Gly-Ala-Hyl-Gly-Glu-NH2	NA	1076.56 (1077.2)	NA
GLT25D1 substrate	Gly-Tyr-Hyp-(Gly-Pro-Hyp)2-Gly-Leu-Hyp-Gly-Ala-Hyl-Gly-Glu-Ala-Gly-Leu-Hyp-(Gly-Pro-Hyp)3-NH2	NA	2783.70 (2783.00)	NA
Lysyl hydroxylase product	Thr-Pro-Gly-Glu-Hyl-Gly-Glu-Hyl-Gly-NH2	NA	956.19a (955.99)b	NA
α1(I)79–93(Gal) THP	(Gly-Pro-Hyp)4-Gly-Thr-Ala-Gly-Leu-Hyp-Gly-Met-Hyl(Gal)-Gly-His-Arg-Gly-Phe-Ser-(Gly-Pro-Hyp)4-NH2	DDR2?	3806.28 (3806)	25
C10-α1(I)79–93(Gal) THP	C10-(Gly-Pro-Hyp)4-Gly-Thr-Ala-Gly-Leu-Hyp-Gly-Met-Hyl(Gal)-Gly-His-Arg-Gly-Phe-Ser-(Gly-Pro-Hyp)4-NH2	DDR2?	3960.33 (3960)	30
C16-α1(I)79–93(Gal) THP	C16-(Gly-Pro-Hyp)4-Gly-Thr-Ala-Gly-Leu-Hyp-Gly-Met-Hyl(Gal)-Gly-His-Arg-Gly-Phe-Ser-(Gly-Pro-Hyp)4-NH2	DDR2?	4044.56 (4044)	37
C16-α1(IV)1263–1277(Gal) THP	C16-(Gly-Pro-Hyp)4-Gly-Val-Hyl(Gal)-Gly-Asp-Lys-Gly-Asn-Pro-Gly-Trp-Pro-Gly-Ala-Pro-(Gly-Pro-Hyp)4-NH2	CD44/CSPG	3992.0 (3989.1)	48.5
α1(IV)1263–1277(Gal) THP	(Gly-Pro-Hyp)4-Gly-Val-Hyl(Gal)-Gly-Asp-Lys-Gly-Asn-Pro-Gly-Trp-Pro-Gly-Ala-Pro-(Gly-Pro-Hyp)4-NH2	CD44/CSPG	3752.4 (3751.1)	42
C10-α1(IV)382–393(Gal) THP	C10-(Gly-Pro-Hyp)4-Gly-Ala-Hyp-Gly-Phe-Hyp-Gly-Glu-Arg-Gly-Glu-Hyl(Gal)-(Gly-Pro-Hyp)4-Tyr-NH2	α2β1	3865.1714 (3865.8273)	37
C10-α1(IV)531–543(Gal)(Gal) THP	C10-(Gly-Pro-Hyp)5-Gly-Glu-Phe-Tyr-Phe-Asp-Leu-Arg-Leu-Hyl(Gal)-Gly-Asp-Hyl(Gal)-(Gly-Pro-Hyp)5-NH2	α3β1	4770.0703 (4770.2896)	37

^aNA = not applicable.

^b[M+Na]⁺.

Morpholine is sometimes prefered over piperidine for removing the Fmoc group during glycopeptide solid-phase synthesis, as the lower pK_a of morpholine (8.3) compared with piperidine (11.1) makes the former potentially less detrimental to side chain glycosyls.^24,85,86 This is less of a concern for glycosylated Hyl, as the Gal-Hyl O-glycosidic linkage is stable to strong alkali, in contrast to O-glycosidic linkages to Ser and Thr.⁸⁷ Thus, peptide syntheses incorporating glycosylated Hyl have typically used piperidine as the Fmoc removal agent (see below).^{44,70,79,80,88,89}

Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] was used for the solid-phase synthesis of the α1(IV)1263–1277 sequence from type IV collagen (Table 1).⁴⁴ This sequence contains a glycosylated Hyl residue in position 1265.⁹⁰ α1(IV)1266–1277 was assembled using Fmoc solid-phase chemistry on an automated synthesizer. The last three amino acids were coupled manually in an orbital shaker. Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] was coupled using 3-fold molar excesses of Fmoc-amino acid and 1-hydroxy-7-azabenzotriazole (HOAt), a 2.7-fold molar excess of 1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate (HATU), and a 6-fold molar excess of DIEA in DMF for 18 h. The last two amino acids from the α1(IV)1263–1277 sequence (Fmoc-Val and Fmoc-Gly) were coupled using 4-fold molar excesses of Fmoc-amino acid and HOBt, a 3.6-fold excess of HBTU, and an 8-fold molar excess of DIEA in DMF for 1 h. Fmoc groups were removed with piperidine-DMF (1:4) for 1 h. Peptide-resin cleavage and side-chain deprotection proceeded with H₂O-TFA (1:19) for 1.5 h. The peptide was purified by preparative RP-HPLC and deacetylated with 2 M methanolic sodium methoxide⁹¹ for 1 h at 20 °C. The product was homogeneous by analytical RP-HPLC, and MALDI-MS analysis gave the desired mass (Table 1). Carbazole-sulfuric acid treatment of the peptide was positive for carbohydrate.

Fmoc-D,L-Hyl[ε-Cbz,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)]-OPfp was used for the automated, microwave-assisted solid-phase synthesis of two integrin binding sequences from type IV collagen.⁹² The synthesis of Fmoc-D,L-Hyl[ε-Cbz,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)]-OPfp was as previously described⁷³ starting from the racemate of D,L-5-Hyl. Fmoc-amino acids were coupled using 5 equiv of each amino acid, 4.9 equiv O-(1H-6-chlorobenzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate, and 8 equiv N-methylmorpholine (microwave power of 25 W at 50°C, 300 s). Fmoc-D,L-Hyl[ε-Cbz,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)]-OPfp was incorporated manually using 3 equiv of amino acid and 6 equiv DIEA with a reaction time of 17 h. The N-termini of all glycopeptides were modified with n-dodecanoic acid to ensure collagen-model triple-helical structure and to facilitate their attachment to plastic surfaces during cell adhesion assays.^93,94

Removal of side chain protecting groups and peptide-resin cleavage were carried for 3 h in Ar using 7 mL of H₂O-thioanisole-phenol-1,2-ethanedithiol-TFA (2:2:1:1:34).⁴⁴ Cleaved glycopeptides were precipitated in cold methyl tert-butyl ether, centrifuged, and lyophilized. Peptide products were subjected to acetyl protecting group removal using 0.1 M NaOH solution for 15 min. The solution was neutralized with HCl and lyophilized. Crude glycopeptides were purified using preparative RP-HPLC. The products were homogeneous by analytical RP-HPLC, and MALDI-MS analysis gave the desired masses (Table 1).

Collagen is a ligand for discoidin domain receptors (DDR1 and DDR2).^95,96 Gly-Val-Met-Gly-Phe-Hyp has been identified as a DDR binding motif in collagen.^97,98 However, DDR2 activation was initially reported to require glycosylated collagen,^96,99 and DDR binding sites in addition to Gly-Val-Met-Gly-Phe-Hyp have been noted.^100–102 Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)]⁶⁶ was used for the solid-phase synthesis of several variants of the α1(I)79–93 sequence (previously unpublished; Table 1). This sequence contains a glycosylated Hyl residue at α1(I)87,^11,13,103 and thus was considered as a possible DDR2 binding site. Peptide-resin assembly was performed on an automated peptide synthesizer using NovaSyn TGR resin with an initial load of 0.25 mmol/g. Standard Fmoc chemistry was used throughout with a 4-fold molar excess of the acylating amino acids, and HBTU and HOBt as coupling reagent. Fmoc-Hyl[ε-Boc,O-(2,3,4,6-tetra-O-acetyl-β-D-galactopyranosyl)] was coupled manually in 2-fold molar excess to reduce consumption of this glycosylated amino acid. Removal of the Fmoc protective group was performed using mixture of 1% 1,8-diaza-bicyclo[5.40.]undec-7-ene and 5% piperidine in N-methylpyrrolidinone. Peptides were cleaved from the resin with thioanisole-H₂O-TFA (1:1:18) for 2 h and purified by preparative RP-HPLC. The elution gradient was 0–90% B over 80 min, where A was 0.1% TFA in H₂O and B was 0.1% TFA in CH₃CN. The products were homogenous by analytical RP-HPLC and MALDI MS analysis indicated the desired masses (Table 1; Figure 5).

Figure 5.

MALDI mass spectrum of C₁₀-α1(I)79–93(Gal) THP. In the mass spectrum, m/z 3960.3340 corresponds to [M+H]⁺ (theoretical 3960).

CD spectra were recorded over the range λ = 180–250 nm. Thermal transition curves were obtained by recording the spectra at λ = 222 nm, whereas the temperature was continuously increased in the range of 5–90 °C at a rate of 15 °C/h. For samples exhibiting sigmoidal melting curves, the inflection point in the transition region (first derivative) is defined as the melting temperature (T_m). Alternatively, T_m was evaluated from the midpoint of the transition. All peptides provided CD spectra indicative of triple-helical structure (Figure 6). The peptide T_m values increased with the addition of C₁₀ and C₁₆ alkyl chains (Table 1), an effect observed previously for triple-helical peptides (THPs).⁹⁴

Figure 6.

(A) CD spectra of C₁₀-α1(I)79–93(Gal) THP recorded at 5 °C before (blue) and after (green) the melt. (B) Thermal transition curve. CD spectra were recorded on a JASCO J-810. The peptide concentration was 50 μM in water.

THE ROLE OF GLYCOSYLATED HYL IN RECEPTOR RECOGNITION OF TYPE IV COLLAGEN

Interest in the relationship between levels of collagen glycosylation and cellular activities stems from reports of T-cell specific recognition of a glycosylated type II collagen sequence,^{70,88,89,104,105} the identification of melanoma and breast carcinoma binding sites within type IV collagen that contain glycosylated Hyl residues,^93,106–112 and the need for type I collagen to be glycosylated for osteosarcoma cell proliferation and viability.¹¹³ Our laboratory utilized THPs incorporating the α1(IV)1263–1277 sequence to (a) identify the melanoma cell receptor for this ligand and (b) evaluate the results of single-site glycosylation on melanoma cellular behaviors.⁴⁴ α1(IV)1263–1277 THP was bound by melanoma cell CD44/chondroitin sulfate proteoglycan (CSPG) receptors, and not by the collagen-binding integrins or melanoma-associated proteoglycan/melanoma chondroitin sulfate proteoglycan (MPG/MCSP/NG2). Melanoma cell adhesion to and spreading on the triple-helical α1(IV)1263–1277 sequence was dramatically decreased for glycosylated [replacement of Lys¹²⁶⁵ with Hyl(O-β-D-galactopyranosyl)] versus non-glycosylated ligand (Figure 7). CD44/CSPG did not bind to glycosylated α1(IV)1263–1277.

Figure 7.

Human melanoma cell spreading on (top) 10 μM C₁₆-α1(IV)1263–1277 THP or (bottom) 10 μM C₁₆-α1(IV)1263–1277(Gal) THP at 37 °C. This research was originally published in the Journal of Biological Chemistry,⁴⁴ copyright the American Society for Biochemistry and Molecular Biology.

Glycosylation had a profound effect on CD44/CSPG interaction with collagen. Although structures have been determined for CD44 binding to hyaluronic acid (HA),^114,115 there are no such structural studies for CD44 interaction with collagen. α1(IV)1263–1277 and HA bind to different regions of CD44, as CD44 binding to α1(IV)1263–1277 requires CS but CS interferes with CD44 binding to HA.¹¹⁶ Thus, we could not propose a precise mechanism for modulation of CD44/CSPG binding. Most likely there is a specific, unfavorable carbohydrate-carbohydrate interaction between the CD44 CS and the Hyl¹²⁶⁵ galactose residue, as opposed to glycosylation masking the side-chain charge of residue 1265 given the small size of the carbohydrate.

Modulation of integrin function by glycosylation was examined by synthesizing THPs with Lys substituted by glycosylated Hyl for Lys³⁹³ from the human α1(IV)382–393 gene sequence (α2β1 integrin-specific) and Lys⁵⁴³ and Lys⁵⁴⁰ from the human α1(IV)531–543 gene sequence (α3β1 integrin-specific).⁹² A dose dependent effect on melanoma cell adhesion was observed for the galactosylation of Hyl³⁹³ in α1(IV)382–393 and Hyl⁵⁴⁰ and Hyl⁵⁴³ in α1(IV)531–543 (Figure 8). The integrins were impacted differently, as α2β1 binding was only modestly inhibited [with the primary melanoma cell line (WM-115) being least affected] while the α3β1 interaction was significantly inhibited (Figure 8). Although cell binding to galactosylated ligands can be mediated by galectin-3,^117,118 we observed only non-specific binding of the THPs with galectin-3 (IC₅₀ in the mM range) with no difference between glycosylated and non-glycosylated ligands.⁹²

Figure 8.

Adhesion of melanoma cells as a function of (upper left) C₁₀-α1(IV)382–393 THP, (upper right) C₁₀-α1(IV)382–393(Gal) THP, (lower left) C₁₀-α1(IV)531–543 THP, and (lower right) C₁₀-α1(IV)531–543(Gal)(Gal) THP concentration. Cells were allowed to adhere for 1 h at 37 °C. This research was originally published in the Journal of Biological Chemistry,⁹² copyright the American Society for Biochemistry and Molecular Biology.

Modeling α2β1 integrin interaction with α1(IV)382–393(Gal) THP indicated that (Gal)Hyl³⁹³ was at the outer interface of the integrin interaction site, and thus galactosylation only slightly decreased binding (Figure 9). Modeling α3β1 integrin interaction with α1(IV)531–543 THP indicated that the 531–543 region bound across the α3β1 interface with Asp⁵⁴² binding to the metal ion-dependent adhesion site (MIDAS) motif within the β1 subunit (Figure 10). The significant inhibition of integrin binding to α1(IV)531–543 when glycosylated was mostly likely due to the proximity of the galactosylated residues (Hyl⁵⁴⁰ and Hyl⁵⁴³) to the electrostatic/metal binding interactions via Asp⁵⁴².

Figure 9.

Molecular modeling of α1(IV)382–393(Gal) THP interaction with the α2 integrin subunit I-domain. This research was originally published in the Journal of Biological Chemistry,⁹² copyright the American Society for Biochemistry and Molecular Biology.

Figure 10.

Molecular modeling studies of α3β1 integrin with α1(IV)531–543 peptide. (A) A small deviation of ϕ/ψ angles (10°) due to flexibility of this region results in strand separation. (B) Docking studies of α1(IV)531–543 peptide using three different sets of ϕ/ψ angles. Docked peptide (backbone in gray, Lys in blue, and Asp/Glu in yellow) binds across the α/β integrin interface (magenta/cyan, respectively). (C) Refined docking simulations where only Gly⁵⁴¹ and/or Gly⁵⁴⁴ were allowed to be flexible. This research was originally published in the Journal of Biological Chemistry,⁹² copyright the American Society for Biochemistry and Molecular Biology.

Overall, collagen glycosylation was found to modulate CD44/CSPG and integrin binding. While inhibition of cellular activities due to glycosylation has not been commonly observed, tumor cell binding to type IV collagen was reduced by the presence of sialic acid on sialoglycoprotein P2B.^119,120

CONCLUSIONS

In order to study the effects of collagen glycosylation on a variety of activities, the synthesis of collagen-model peptides containing Hyl and glycosylated Hyl was required. The mild conditions of Fmoc chemistry were particularly well suited to achieve this goal. The development of the Fmoc group by Louis Carpino was thus critically important for furthering our knowledge of collagen biochemistry. In addition, other synthetic achievements of the Carpino laboratory, such as the development of the peptide coupling reagent HATU and additive HOAt^121–123 and the 2,2,4,6,7-pentamethyldihydrobenzofuran-5-sulfonyl (Pbf) Arg side chain protecting group¹²⁴ facilitated the assembly of glycosylated collagen-model peptides. With these tools in hand, we were able to use chemical biology to demonstrate the modulation of tumor cell receptor interactions with type IV collagen via ligand glycosylation.

Abbreviations used are:

Aloc

allyloxycarbonyl

Boc

tert-butyloxycarbonyl

Boc₂O

di-tert-butyl dicarbonate

Bzl

benzyl

Cbz

benzyloxycarbonyl

CSPG

chondroitin sulfate proteoglycan

DDR

discoidin domain receptor

DIEA

N,N-diisopropylethylamine

DMAP

4-dimethylaminopyridine

DMF

N,N-dimethylformamide

Fmoc

9-fluorenylmethyloxycarbonyl

Fmoc-OSu

9-fluorenylmethoxycarbonyl-N-hydroxysuccinimide ester

HA

hyaluronic acid

HATU

1-[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxide hexafluorophosphate

HBTU

2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate or N-[(1H-benzotriazol-1-yl)(dimethylamino)methylene]-N-methylmethanaminium hexafluorophosphate N-oxide

HOAt

1-hydroxy-7-azabenzotriazole

HOBt

1-hydroxy-benzotriazole

Hyl

5-hydroxylysine

Hyp

4-hydroxyproline

MALDI

matrix-assisted laser desorption/ionization

MIDAS

metal ion-dependent adhesion site

MS

mass spectrometry

NMR

nuclear magnetic resonance

Pfp

pentafluorophenyl

RP-HPLC

reversed-phase high pressure liquid chromatography

TBDMS

tert-butyldimethylsilyl

TBDMSOTf

tert-butyldimethylsilyl trifluoromethanesulfonate or tert-butyldimethylsilyl triflate

tBu

tertiary-butyl

TFA

trifluoroacetic acid

THP

triple-helical peptide