Compartmentalized processing of catechols during mussel byssus fabrication determines the destiny of DOPA

Significance

Catechol chemistry has emerged as a cornerstone of bioinspired polymers and adhesives due to its versatility in creating diverse covalent and dynamic noncovalent interactions (including metal coordination). The concept initially arose from the discovery that mussels use catechol moieties of 3,4-dihydroxyphenylalanine (DOPA) to mediate robust surface adhesion in seawater and to reinforce tough and self-healing biopolymer fibers. Currently, difficulties controlling DOPA redox chemistry limit its synthetic application; yet, mussels overcome this challenge daily through apparent physical and chemical process control. Here, we reveal that mussels employ several different processing pathways that predetermine the cross-linking fate of DOPA-containing proteins via spatiotemporal control of microenvironments in secretory vesicles and later in mature threads—with key significance for advanced polymer design.

Abstract

Inspired largely by the role of the posttranslationally modified amino acid dopa (DOPA) in mussel adhesion, catechol functional groups have become commonplace in medical adhesives, tissue scaffolds, and advanced smart polymers. Yet, the complex redox chemistry of catechol groups complicates cross-link regulation, hampering fabrication and the long-term stability/performance of mussel-inspired polymers. Here, we investigated the various fates of DOPA residues in proteins comprising mussel byssus fibers before, during, and after protein secretion. Utilizing a combination of histological staining and confocal Raman spectroscopy on native tissues, as well as peptide-based cross-linking studies, we have identified at least two distinct DOPA-based cross-linking pathways during byssus fabrication, achieved by oxidative covalent cross-linking or formation of metal coordination interactions under reducing conditions, respectively. We suggest that these end states are spatiotemporally regulated by the microenvironments in which the proteins are stored prior to secretion, which are retained after formation—in particular, due to the presence of reducing moieties. These findings provide physicochemical pathways toward greater control over properties of synthetic catechol-based polymers and adhesives.

In the last 10 y, the use of catechol chemistry in the production of advanced polymers and adhesives with dynamic and adaptive behaviors has increased dramatically (1–6). These materials, many of which have proven potential in the realm of biomedical and tissue engineering (1, 7), gain inspiration largely from specific chemical design principles elucidated from the protein-based adhesive biofibers of the mussel byssus (Fig. 1A) (8). Catechol chemistry, however, is notoriously complex and difficult to control, exhibiting a strong tendency toward oxidation and covalent cross-linking under certain conditions, while under other conditions the catechol group participates in a variety of different noncovalent interactions with surfaces or forms metal coordination complexes with very high stability constants (9). Nature provides important inspiration in this regard as mussels have evolved to use a natural catechol—3,4-dihydroxyphenylalaine (DOPA)—in many of the proteins that comprise the byssus toward desirable material properties such as underwater adhesion, high toughness, and even self-healing (8, 10). Recent investigations of the mussel byssus formation process, however, have clearly revealed that mussels have evolved specific strategies for regulating catechol chemistry during thread fabrication, e.g., by controlling pH and redox conditions (8). By gaining a deeper understanding of the underlying physical and chemical principles controlling how the mussel is able to guide the fate of its many DOPA-rich proteins, one can potentially extract new design principles for improving and tuning the properties of catechol-based materials with clear implications for biomedical engineering.

Fig. 1.

Localization of DOPA in mussel byssal threads and mussel foot tissue. (A) Mussels attach to surfaces using a collection of biopolymeric fibers known as a byssus. Threads are secreted by an organ known as the foot and possess distinctive regions with different function—the core (a tensile tether), the cuticle (a protective coating), and the plaque (a foamy adhesive). (B) NBT staining of thread core and cuticle and plaque tissue can be used to localize DOPA residues. (C) Raman spectra acquired from the core, cuticle, and plaque of a mature byssal thread. Dashed lines indicate characteristic DOPA–metal resonance Raman peaks corresponding to oxygen–metal vibrations (500 to 700 cm⁻¹) and DOPA ring vibrations (1,270, 1,322, and 1,476 cm⁻¹). (D) NBT staining of a transverse cryosection of a mussel foot, showing the presence of DOPA in the plaque gland (pg), core gland (cg), and cuticle gland (ctg), respectively. Proteins stored in these glands are eventually secreted into the ventral foot groove (fg). (E) Confocal Raman map on a similar section to D showing localization of the different secretory glands (colors correspond to spectra in F). (F) Raman spectra corresponding to the glands in E before (dashed) and after (solid) incubation with 1 mM FeCl₃ solution. Vertical dashed gray lines in C and F indicate resonance Raman peaks corresponding to DOPA–metal complex.

Of the more than 10 proteins extracted and characterized from the byssus, almost all of them contain DOPA to varying degrees (1, 8). DOPA is encoded in the DNA sequence as a tyrosine residue, which is then posttranslationally modified with a tyrosinase enzyme to DOPA via addition of a hydroxyl group in the 3 position of the aromatic ring (11). These DOPA residues are implicated in a number of different functional roles in the byssus, which have been validated through diligent study over many years, most prominently through the combined work of Waite and Israelachvili and coworkers (12–15). At the adhesive interface that the byssus forms with hard surfaces on the rocky seashore, DOPA residues in the proteins mfp-3 and mfp-5 (mfp, mussel foot protein) have been shown to mediate adhesion via formation of a number of transient and covalent interactions with surfaces including hydrogen bonding, metal coordination, hydrophobic interactions, and oxidative covalent cross-linking (12–16). In the foamy bulk of the plaque and in the thin protective outer coating of the byssus, known as the cuticle, intermolecular Tris–DOPA–metal complexation by the proteins mfp-2 and mfp-1, respectively, has been shown to provide mechanical reinforcement, providing 80% of the stiffness and hardness in the cuticle and contributing to cohesion within the plaque foam (10, 17, 18). In the self-healing fibrous core of the byssus threads, the role of DOPA (which has a relatively low content of ∼1 mol%) is much less understood. However, it has been suggested that the well-conserved DOPA residues at the termini of the collagenous core proteins (known as the preCols) are responsible for forming covalent cross-links via aryl coupling with other DOPA residues or Michael addition/Schiff base reaction with other amino acids driven by the spontaneous oxidation of DOPA–catechol to its quinone form (11, 16, 19).

Given the many different proposed destinies and functions attributed to DOPA in the byssus fiber, the question naturally arises of how this is all controlled at the micrometer scale in a biopolymer processing event that occurs in just minutes and a postsecretion curing process that occurs subsequently in seawater, with no further biological intervention (20). As mentioned, this is especially relevant given the challenges of controlling DOPA chemistry that have emerged in efforts to make mussel-inspired materials (6, 21). This question turns out to be much more difficult to answer given that many of the steps of byssus formation occur within the confines of the mussel foot (Fig. 1A). During byssus formation, liquid protein precursors are secreted into a groove in the mussel foot where they self-assemble and cross-link to form a rigid thread (20). Waite and coworkers have made significant progress in understanding the controls in place for regulating DOPA chemistry at the adhesive interface through a combination of in vivo and in vitro studies (8, 22), which was supported by in vivo measurements of Miserez and coworkers (23). As mentioned, strong adhesion is favored by various interactions between the reduced, catechol form of DOPA and the substrate (16). DOPA–catechol is believed to be maintained at low pH or in the presence of reducing agents (e.g., Cys-rich protein mfp-6), functioning within microenvironments formed by liquid–liquid phase separation of protein precursor (8, 16, 22, 24). It seems plausible that a similar mechanism may function in the cuticle; however, the processing pathway of DOPA in the core is not clear.

In the present work, we follow the byssus secretion and curing process in order to investigate the various fates of DOPA residues in the proteins comprising the different parts of the thread—with a major focus on DOPA oxidation in the core. To achieve this, we utilize histological and spectroscopic assays for monitoring DOPA viability over time. We validate our assay and generate further insights into this process using synthetic peptides derived from the conserved protein domain postulated to be integral in cross-linking the thread core. Our findings indicate that DOPA oxidation is spatiotemporally controlled, guiding the cross-linking fate. Specifically, DOPA oxidation in the thread core triggers covalent cross-linking and contributes to fiber formation and maturation. In contrast, DOPA oxidation is prevented in the plaque bulk and cuticle, enabling metal chelation toward dynamic dissipative properties. These findings have strong relevance to the field of mussel-inspired polymers where improved control over DOPA cross-linking could improve current materials.

Results

Localization of DOPA in Byssal Threads and Mussel Foot Tissue.

In order to detect the presence of DOPA in the various parts of the byssal thread and its producing organ (the mussel foot) (Fig. 1A), we utilized nitro blue tetrazolium (NBT) staining and confocal Raman spectroscopic mapping. NBT is a redox active dye which can be utilized to detect DOPA or other catechol moieties in histological tissue sections (25), and has been previously applied to study DOPA localization in byssal foot and thread tissues (20). In native threads, the dark blue staining indicative of the redox reaction of DOPA with NBT was observed in the cuticle and plaque, but not in the core (Fig. 1B). However, NBT staining of thin sections of the mussel foot revealed the presence of DOPA in the three glands that store the proteins that are eventually excreted to form the core, cuticle, and plaque (Fig. 1D). Surprisingly, the core gland stains the darkest, even though the prominent core proteins contain only up to 1 mol% DOPA and despite the fact that no staining was observed in the native thread core. In contrast, cuticle and plaque proteins which typically contain >15 mol% DOPA (1), stain only light blue. This phenomenon was proposed previously to arise from the low pH and effective reducing conditions within the vesicles involved in plaque and cuticle biogenesis (8, 20), although the exact internal conditions of these vesicles remains unclear.

Complementary to NBT staining, Raman spectroscopy can be used to detect the presence of DOPA–metal interactions, which produce very intense resonant Raman peaks at 550, 596, and 637 cm⁻¹ (catechol–metal interaction) and 1,270, 1,322, and 1,476 cm⁻¹ (catechol ring vibrations) (10, 17, 26). In native threads, these DOPA–metal interactions are present in the cuticle and plaque (17, 18), but not in the core (Fig. 1C). In contrast, Raman spectra of proteins in all three mussel foot glands show no DOPA–metal resonance Raman peaks, indicating that, as previously proposed (20), metal ions such as Fe and V are not stored in the glands with the proteins. However, by briefly incubating the foot sections with 1-mM solutions of either FeCl₃ or VCl₃, resonance Raman peaks for DOPA–metal interactions are easily detected in all three glands, although the resonance signal is markedly higher in the cuticle and plaque where DOPA content is also suspected to be higher (Fig. 1 E and F and SI Appendix, Fig. S2 A and B). The fact that this contradicts the staining intensity of NBT in the different glands (namely, that the core exhibited the highest NBT staining) offers further support for highly effective reducing conditions within the plaque and cuticle gland vesicles, which would be expected to inhibit the redox reaction of DOPA with NBT (20). With this method, the specific DOPA–metal Raman peaks are increased up to a millionfold, providing a highly sensitive assay for revealing trace quantities of DOPA–catechol and its coordinated metals (26). We, therefore, use this assay throughout this study alongside traditional NBT staining to probe the state of DOPA at various stages during the thread formation process.

Although DOPA is present and able to form interactions with metal ions in the core proteins in the secretory glands prior to secretion, no metal interactions are observed in the native thread core, even after the thread is incubated with metal ion solutions (Fig. 1 B and C). It has been previously proposed that conserved DOPA residues in the core proteins might react with other amino acid residues including DOPA, cysteine, histidine, or lysine to form covalent cross-links (9, 19, 27–29). This is thought to proceed via oxidation of DOPA–catechol to DOPA–quinone, which can subsequently react with electron-rich amino acid moieties either via spontaneous radical-initiated or polar nucleophilic addition processes (8, 9). Since DOPA–quinone is a much less effective chelator of metal ions (16), we posit that the lack of a resonance Raman signal indicates oxidation and subsequent cross-linking of DOPA residues within the core with other amino acid side chains (e.g., Lys, His, and other DOPA residues) or conversion of DOPA–quinone to the dehydro-DOPA form (9, 30, 31). Indeed, it is expected that the DOPA–quinone state will be short-lived, as it is highly reactive and transient (32, 33). This is consistent with the fact that NBT does not stain the native core (20) (Fig. 1B). In contrast to the thread core and consistent with previous findings (17, 18), DOPA residues in the cuticle and plaque are not oxidized, but rather form reversible DOPA–metal cross-links after secretion of the proteins (Fig. 1 B and C). It is worth noting that the Raman spectra from both native cuticle and plaque indicate the presence of DOPA–V coordination, rather than DOPA–Fe based on the position of specific resonance peaks (Fig. 1C). This is consistent with several recent findings performed on several mytilid mussel species, which have clearly demonstrated spectroscopically and mechanically that vanadium ions are localized in the granules coordinated with DOPA, where they contribute significantly to cuticle hardness and stiffness (10, 34). These observations have intriguing implications that must be addressed in the future considering the very different oxidative properties of V and Fe, and the very different viscoelastic response of DOPA–V and DOPA–Fe mussel-inspired polymers (35, 36).

Processing of DOPA.

Above, we observed that all protein vesicles have active DOPA capable of NBT staining and metal binding (Fig. 1); yet in the mature thread, these tendencies are altered differently in different parts of the thread on the microscale. In order to better understand the root cause and regulation of this differential cross-linking process and to reveal at what point these various DOPA-based cross-links are formed, we investigated thread assembly and curing by artificially inducing the thread formation process via injection of a potassium chloride solution in the base of the foot as described previously (20, 37). Using our two assays to detect the presence of DOPA–catechol, we then followed DOPA processing over time and under different treatment conditions aimed at perturbing or accelerating the putative DOPA-based cross-linking process.

We observed that freshly secreted induced threads show the presence of DOPA via NBT staining in both the core and the cuticle (Fig. 2 B and C). However, consistent with previous studies (20), confocal Raman spectroscopy revealed that no DOPA–metal interactions were present in the core or cuticle of freshly induced threads (Fig. 2 D–F), even though both parts of the thread are clearly able to coordinate metal ions after incubation of either FeCl₃ or VCl₃ solutions (Fig. 2 E and F and SI Appendix, Fig. S2 C and D). These data reveal two key points: 1) immediately after secretion, DOPA–catechol is still present in both the core and cuticle; and 2) metals are not added during the initial protein secretion step of the thread fabrication, as previously proposed (20).

Fig. 2.

DOPA localization in induced byssal threads. (A) Unstained longitudinal cryosection of an induced mussel foot with induced thread (it) in the foot groove (fg) surrounded by the core gland (cg) in which the core-forming preCol proteins are stored in secretory vesicles. (B) NBT-stained sample similar to A highlighting the induced thread (it) in the foot groove (fg). (C) Magnified image of NBT-stained induced thread from B. (D and E) Raman spectroscopic imaging of an unstained induced thread in a foot groove before (D) and after (E) incubation with 1 mM FeCl₃. Colored overlay corresponds to Raman spectral mapping localizing the core (blue) and cuticle (red) of the induced thread shown in F. (F) Corresponding Raman spectra of the core and cuticle regions of the induced threads before and after incubation with 1 mM FeCl₃, showing the ability of DOPA in both regions to bind Fe³⁺, based on characteristic DOPA–Fe resonance peaks highlighted with dashed lines.

Oxidation of DOPA.

As mentioned, it has been proposed that DOPA present in the core proteins forms covalent cross-links with other amino acid residues due to its spontaneous oxidation to DOPA–quinone. For this reason, we monitored the presence of available DOPA–catechol over time in induced threads stored under various conditions. In the core, NBT staining is diminished over time in induced threads stored in buffered solutions at pH 8, which mimics the pH of seawater. In contrast, threads stored at pH 5 (mimicking the suspected acidic protein storage conditions prior to secretion) (22, 24) maintained a constant NBT staining level over 14 d (Fig. 3A). A key step in the chemical mechanism of the NBT staining reaction depends on oxidation of DOPA–catechol to DOPA–quinone (25), and thus, the intensity of staining is directly related to the amount of free DOPA–catechol remaining in the induced threads following the various treatment conditions. It is well established that catechols will spontaneously oxidize to their quinone form under basic conditions (1, 8), which is consistent with our observation that threads stored under acidic conditions exhibit positive NBT staining, but those stored under basic conditions similar to the natural seawater pH do not (38, 39). To further validate the oxidation of DOPA, freshly induced threads were exposed to both reducing and oxidizing agents. No change in NBT staining is observed over time in the presence of a reducing agent (ascorbic acid, ref. 40), whereas even brief treatment of threads with the oxidizing agent sodium periodate (NaIO₄), prevents staining immediately and completely (Fig. 3A). Raman spectroscopy of metal-incubated samples following these various redox and pH treatments corroborated the results from the NBT assays. The induced core stored at pH 5 could still coordinate metal ions after 14 d (suggesting abundant DOPA–catechol) (Fig. 3B), but if stored at pH 8, no metal coordination is detected in the core with Raman spectroscopy. Considering that we are adding an excess of metal ions to the system in these experiments, these observations clearly indicate that the DOPA residues are no longer available for metal binding, likely due to spontaneous oxidation at basic pH and subsequent covalent cross-linking (Fig. 3C). In contrast, NBT staining and metal coordination were still observed in the outer cuticle of induced threads incubated for 14 d in pH 8 buffer (Fig. 3 A–C), suggesting that a mechanism exists to hinder DOPA oxidation in the secreted cuticle—even in the absence of metal binding (see below).

Fig. 3.

Maturation of induced byssal threads under different storage conditions. (A) Time series showing NBT staining of induced threads within a foot groove at various time points when stored under a range of conditions, including basic (pH 8), acidic (pH 5), reducing (ascorbate), and oxidizing (NaIO₄). Inset of pH 8 at 14 d shows that the cuticle still stains positive for NBT, while the core does not. (B and C) Raman spectra of core and cuticle regions of induced threads incubated with 1 mM FeCl₃ after storage for 14 d under acidic (B) and basic conditions (C). Under acidic conditions, both core and cuticle exhibit DOPA–metal complexation, whereas only the cuticle does under basic conditions (a.u., arbitrary units).

PreCol Peptide Analysis.

DOPA is present in the core proteins in the gland and in early stages of thread assembly, but diminishes over time in the secreted thread core. The majority (∼98%) of the distal region of the core consists of preCol proteins (41), which contain up to 1 mol% DOPA. The DOPA moieties in preCol proteins are located entirely at the N and C termini of the protein. This localization pattern is highly conserved between the different preCol variants and between different species and thus, it has been proposed that the DOPA residues are necessary for forming covalent cross-links between the ends of axially adjacent preCol proteins (19, 42). To investigate this, we synthesized a peptide corresponding to the N-terminal sequence of preCol-D from Mytilus edulis that we named pC-DN, which contains four DOPA residues in positions that are coded as tyrosine in the DNA sequence (Fig. 4A). Similar to the induced thread core, the dried peptide blotted on nitrocellulose paper stains with NBT, but the staining becomes diminished over time at slightly basic pH or in the presence of oxidation agents. In addition, while the freshly made peptide solution is clear, it turns slightly yellow over time at pH 8.

Fig. 4.

Fate of DOPA in mussel mimetic peptides. (A) Short peptide sequence based on the DOPA-enriched N terminus of preCol-D (named pC-DN) was synthesized. NBT staining of pC-DN peptide blotted on nitrocellulose paper and stored under various conditions shows that DOPA residues are oxidized under basic conditions, but not under acidic conditions. (B) Time-dependent UV/vis spectra of peptide solutions in different buffering solutions. The development of an absorption peak around 350 nm (arrow) is associated with DOPA oxidation. (C) UV/vis spectra of peptide solutions mixed FeCl₃ before and after storage under basic conditions (pH 8). Peak at ∼500 nm is associated with DOPA–Fe complexation. (D) Raman spectra of peptide with and without FeCl₃ added.

To investigate this more quantitatively, we performed UV-visible (UV-vis) absorption spectroscopy experiments on solutions of the pC-DN peptide and observed the development of a broad peak centered at ∼350 nm (Fig. 4B). The oxidation of DOPA–catechol to DOPA–quinone normally produces an absorption band centered at ∼395 nm (29, 30). The 350-nm peak is consistent with the further conversion of DOPA–quinone to α,β-dehydro-DOPA, which has been previously observed in peptidyl-DOPA analogs as an intermediate for sclerotization (30, 43). It is worth noting that the absorption peak of dehydro-DOPA was reported between 320 nm and 350 nm depending on the buffer system (30, 31). Alternatively, the 350-nm absorption could reflect Schiff base formation between DOPA and a lysine residue (44). Notably, this peak does not appear under acidic conditions (pH 5), further suggesting that the color change is related to spontaneous oxidation and possible cross-linking of DOPA under basic conditions. We also observed that the freshly prepared synthetic pC-DN peptide forms Tris–DOPA–metal coordination interactions, as evidenced by characteristic UV-vis absorption and resonance Raman peaks (Fig. 4 C and D). The absorption at ∼500 nm is likely associated with the formation of the Tris–catechol–iron complex with a λ_max at 482 nm (45). However, this ability to bind metals decreases over time at high pH presumably due to oxidation of DOPA. These findings are consistent with our measurements with the induced thread.

We next attempted to detect covalently cross-linked pC-DN oligomers using liquid chromatography electrospray ionization mass spectrometry (LC-ESI-MS). For these studies, the pH of a sample of pC-DN peptide in water was adjusted to ∼8 using dilute ammonium hydroxide. At desired time points, aliquots were removed from this reaction, quenched by the addition of formic acid, and analyzed by LC-ESI-MS as described in Materials and Methods. In this assay, we observed rapid (but incomplete) oxidation of the four DOPA residues to DOPA–quinone (Fig. 5A). As time progressed, additional MS signals corresponding to dehydrated pC-DN peptides appeared, as well as a number of signals that appear to be oxidized pC-DN adducts (green spheres in Fig. 5A). The oxidized pC-DN adducts likely arise from reaction of the peptide with H₂O₂ generated in situ during the spontaneous two-electron oxidation of DOPA–catechol to DOPA–quinone (46). The dehydrated adducts could result from the intramolecular nucleophilic addition of one of the three lysine residues located in the pC-DN peptide onto a nascent DOPA–quinone moiety (Fig. 5E). Consistent with this claim, the time scale for the formation of the dehydrated peptides is similar to the time scale for the formation of the absorption at 350 nm, reaching completion within about 90 min (Fig. 5B). Furthermore, the loss of a hydroxyl group from the DOPA moiety in this adduct is consistent with the loss of a resonance Raman signal, which results from chelation of the metal ion by both hydroxyl groups (26).

Fig. 5.

Characterization of DOPA-containing peptides from preCol-D sequences. (A) Mass spectral time course of 100 µM of pC-DN, a DOPA-containing peptide from the N terminus of preCol-D, incubated at pH 8.0 for the indicated time. The Inset illustrates the rapid broadening of the isotope distribution of the starting material toward smaller m/z values over the first 5 min of the reaction, indicating the oxidation of the four DOPA–catechol residues to DOPA–quinone. The red numbers denote the monoisotopic peaks of the peptides containing zero to four DOPA–quinone residues. At longer reaction times, dehydrated peptides and oxidized adducts (marked with green dots) were detected. (B) Time-dependent formation of oxidized DOPA in the preCol-D peptide based on a chromophore that absorbs maximally at 350 nm. These data were collected from a portion of the same reaction mixture used to generate the data in A. (C) Tandem MS fragmentation spectrum of the starting material, [M], from the t = 0-min time point. Numerous b and y fragment ions were observed and are mapped onto the pC-DN peptide sequence. (D) Tandem MS fragmentation spectrum of the [M-1H₂O] peak from the 5-min time point. The fragmentation efficiency is significantly reduced, especially in the N-terminal portion of the peptide. This is consistent with the presence of Lys–DOPA–quinone cross-links. (E) Cross-linking reaction leading to loss of a water molecule.

In an attempt to confirm the presence of intramolecular cross-links, we performed tandem mass spectrometry studies on the starting material and on the one-fold dehydrated peptide (labeled as [M] and [M-1H₂O], respectively, in Fig. 5A). Fragmentation by collision-induced dissociation (CID) resulted in numerous b and y ions for the starting material and clearly confirmed the presence of the four l-DOPA residues (Fig. 5C). In contrast, the fragmentation efficiency of the [M-1H₂O] peptide was significantly reduced, especially in the N-terminal region of the peptide where the DOPA–quinone and lysine residues are clustered (Fig. 5D). This is most strongly evidenced by the complete lack of the b₄-b₁₆ ion series in the fragmentation spectrum of the dehydrated peptide. This observation is consistent with the presence of intramolecular cross-links in the pC-DN peptide, which would greatly diminish generation of b and y fragment ions in this region. Surprisingly, we were unable to detect intermolecularly cross-linked pC-DN oligomers under the LC-ESI-MS conditions employed (a broad band mass spectrum showing an extended mass-to-charge ratio [m/z] range is shown in the SI Appendix, Fig. S3). However, the facile intramolecular cross-linking observed here clearly suggests that this chemistry is possible within the core region of the byssus fiber, where preCol-D would be highly concentrated and coaxially organized. Future investigations will focus on elucidating the in vitro reactivity of oxidized preCol-D in more detail.

Discussion

The most illuminating finding of the present study is that mussels apparently spatiotemporally control the fate of DOPA residues differently in different proteins, which comprise distinct parts of the byssus (e.g., core, cuticle, plaque). Control is likely achieved by regulating the microenvironment of the protein secretory vesicles before secretion, which is a strong determinant of the future fate of the DOPA residues after secretion (assuming the conditions are maintained), free of further biological intervention (Fig. 6). Specifically, we provide clear evidence that DOPA residues undergo at least two distinct processing pathways during byssus formation that either promote oxidation and favor covalent cross-link formation (e.g., in the core), or that hamper oxidation and favor the formation of noncovalent interactions including metal coordination cross-links (e.g., in the cuticle and plaque foam) or adhesive interactions (e.g., at the plaque–surface interface) (Fig. 6). Our findings suggest that the mussel regulates the two different fates of DOPA by controlling the pH and redox environment of the proteins before secretion, which then determines the long-term fates of those proteins. Indeed, we demonstrated that cross-linking of DOPA in the core can be accelerated or prevented under basic and acidic pH, respectively. In contrast, DOPA in the cuticle remains reduced and available for metal binding even after storage for 2 wk under basic conditions that typically favor DOPA oxidation. This is likely related to the recently discovered Cys-rich proteins believed to be present in the cuticle, which are proposed to function as a multifunctional reducing agent (34, 47, 48).

Fig. 6.

Byssus processing and the destiny of DOPA. The mussel produces and stores the byssal proteins in a specialized organ called mussel foot within three different glands. Nearly all proteins contain DOPA which is derived from posttranslational modification of tyrosine by tyrosinase. The oxidation of DOPA within the glands is hindered by storing the proteins at acidic pH. After secretion and assembly of proteins into a byssal thread, DOPA residues are exposed to the slightly basic pH of seawater. In the thread core, this leads to oxidation of DOPA to DOPA–quinone and subsequently, formation of covalent cross-links between DOPA and other amino acid residues. In the cuticle and plaque, however, oxidation is prevented, likely due to the presence of cysteine-rich proteins that reduce DOPA–quinone to DOPA–catechol, which is important for the adhesion of the plaque and for the formation of DOPA–metal complexes in the cuticle.

Oxidation of DOPA in the Core.

Before the secretion and formation of the thread, all proteins are believed to be stored in vesicles at low pH in the three mussel foot glands, thereby preventing DOPA oxidation prior to secretion. This is supported by the low pH (∼2) measured in the plaque secretion (24), although this could be different in the different vesicles. In addition, the proteins are stored separately from metal ions like iron or vanadium, since this also could lead to DOPA oxidation at low pH (49). After the secretion of the thread proteins, the threads are exposed to seawater (pH 8)—conditions under which DOPA is typically prone to oxidation (16). Under basic conditions, we observed relatively rapid DOPA oxidation in the core of induced threads that continues over a time span of up to 2 wk. DOPA oxidizes in the presence of oxygen to DOPA–quinone, which is very reactive and can form covalent cross-links to other amino acids. Interestingly, the major proteins of the thread core—the preCol proteins—contain conserved DOPA residues in their N and C termini. Using a model peptide pC-DN derived from the N terminus of preCol-D containing four DOPA residues, we observed spontaneous oxidation of DOPA and formation of covalent cross-links, possibly due to Schiff base formation with lysine residues, which are also highly conserved in the preCol termini. While these findings are very suggestive of a regulated time-dependent curing of the thread core after release into seawater, further research is required to reveal where and between which amino acid residues cross-links are formed in native threads, as this is expected to significantly influence the mechanical properties of the mature thread.

Delayed Oxidation of DOPA toward Metal Binding in the Plaque and Cuticle.

In contrast to the proposed spontaneous covalent cross-linking of DOPA in the core, DOPA oxidation is clearly prevented or at least, decelerated in the thread cuticle based on its ability to bind metal ions and undergo NBT staining, even after a 2-wk incubation at pH 8. In the native threads, it is conceivable that the coordination of DOPA–catechol with metal ions prevents their oxidative cross-linking (6); however, the lack of oxidation to DOPA–quinone was observed in induced threads after 2 wk at pH 8—even in the absence of metals. Thus, other mechanisms must be functioning here. One intriguing possibility is that DOPA residues in the cuticle are isolated in a locally reducing microenvironment, even when the bulk solution is basic and oxidizing. This hypothesis is supported by the work of Waite and coworkers on microenvironments in the thread plaque mediated by a cysteine-rich protein known as mfp-6, which is secreted at the adhesive surface, but apparently does not participate in adhesion (22). It was shown that the oxidation of DOPA is prevented during plaque formation by mfp-6, due to its dynamic redox activity in which the sulfhydryl groups of the Cys residues actively reduce DOPA–quinone back to DOPA–catechol (22). This step is crucial for the adhesion of mfp-3 and mfp-5 as DOPA–catechol is a much more efficient adherent than DOPA–quinone (16).

Along these lines, a family of putative cysteine-rich proteins were recently identified in a transcriptome of the cuticle gland from a related species (Mytilus californianus) (mfp-16 to 19), one of which has been recently localized in the cuticle secretory vesicles via proteomics (48). Recent elemental analysis of the cuticle with energy dispersive X-ray spectroscopy (EDS) has also revealed a high sulfur content within the cuticle colocalized with the DOPA-containing protein mfp-1 (34). Together with the findings of the current study, these data suggest that the Cys-rich proteins in the cuticle may serve a similar role as an effective reducing agent, favoring the formation of metal coordination in the cuticle. How and when the metal ions are added to the cuticle (and plaque) is another interesting question for future investigation; however, our current findings reveal that the DOPA residues in these regions are awaiting their arrival in a reduced catechol form under conditions that typically favor DOPA oxidation. Furthermore, if the metal coordination bonds break during normal wear and tear in the marine environment, the reducing microenvironment produced by the Cys-rich proteins should help to combat DOPA oxidation (8).

Conclusion

The versatile chemistry of the noncanonical amino acid DOPA has earned a great deal of attention in the realm of bioinspired materials, fueled largely by discoveries of its function in biological adhesives, most prominently the mussel byssus (8). Yet, efforts to effectively translate these discoveries into technical and biomedical applications have been hampered by challenges in controlling the highly reactive chemistry of DOPA. Thus, understanding how mussels control DOPA destiny before, during, and after byssus fabrication may offer new insights in this respect. In the present study, we provide clear evidence for at least two distinct processing pathways resulting either in oxidation and covalent cross-linking of DOPA or in delayed oxidation preferring formation of noncovalent metal complexes and/or adhesive interactions with surfaces. Our findings in the light of recent studies suggest that these different fates are a result of the specific physicochemical microenvironments of the secretory vesicles in which the protein precursors are stored (in terms of pH and redox conditions), which are controlled biologically. After secretion (at which point proteins become free from further biological intervention), these storage conditions determine the eventual destiny of the DOPA residues, favoring one of the two pathways mentioned above, based entirely on physical and chemical considerations. By elucidating these pathways, we provide important design parameters for the further development of DOPA-containing bioinspired smart polymers with important biomedical and technical applications.

Materials and Methods

Sample Preparation.

Mussels of the species M. edulis were acquired from farms near Prince Edward Island and prior to preparation were kept in a marine aquarium with artificial seawater at 15 °C. Mussels were dissected open and were induced to secrete thread proteins via injection of 0.56 M potassium chloride solution in the base of the mussel foot at the bottom of the pedal groove, as described previously (20). After 5 to 10 min, the mussel foot was cut out and immediately frozen in optimal cutting temperature (OCT) medium with liquid nitrogen-cooled isopentane. Afterward, 5-µm thick longitudinal or transverse cross-sections were cut with a cryotome at −20 °C.

Thread Cross-Linking Studies.

To investigate how DOPA-based cross-linking in freshly secreted threads proceeds over time, longitudinal sections of induced threads were incubated under different conditions for up to 2 wk. Specifically, freshly cut sections were fixed for 10 min in cold methanol and then stored in one of four buffering conditions: 1) 100 mM phosphate buffer, pH 8; 2) 100 mM citrate buffer, pH 5; 3) 25 mM ascorbic acid in 100 mM phosphate buffer, pH 8 (40); or 4) 25 mM NaIO₄ in 100 mM Tris buffer, pH 8. After 1, 5, and 14 d, sections were briefly washed and investigated with NBT staining and with confocal Raman spectroscopy.

Raman Spectroscopy.

Raman spectroscopy was performed on 5-µm-thick dry mussel foot sections on glass slides using a confocal Raman microscope (Alpha 300R, WITec). A 532-nm green laser was used at a power between 3 and 10 mW and focused with an 100× objective (Zeiss, numerical aperture [NA] = 0.9). Spectra were collected with a thermoelectrically cooled CCD detector behind a 600-g/mm grating. For image scans, an integration time of 5 s per pixel was used and for single spectra, a 1-s integration time and 60 accumulations were used. Data were collected with WITec ControlFIVE 5.1 software and processed using WITec Plus software to remove cosmic rays and subtract the background signal. Distinct characteristic spectra for different regions within the tissue glands were identified using linear combinations of single spectra. The presence of DOPA–catechol in mussel foot and thread sections was further investigated by briefly incubating samples with a metal chloride solution (either 1 mM FeCl₃ or 1 mM VCl₃), followed by washing with distilled water. DOPA–metal binding was assessed by confocal Raman spectroscopy, given the strong resonance Raman signal of the complex (17, 26, 50).

NBT Staining.

NBT staining was performed according to previously described protocols (25). Tissue sections where first fixed in cold methanol for 10 min and subsequently stained with a freshly made solution of 0.2 mg/mL NBT in 2 M potassium glycinate buffer (pH 10) for 40 min in the dark. After staining, sections were quickly washed with water and then dehydrated in an ascending alcohol series (95% ethanol, 100% ethanol; NeoClear) and mounted with NeoMount.

Peptide Synthesis.

A synthetic peptide with a sequence corresponding to the N terminus of preCol-D (DY*GRKY*GKPSY*GEY*GGKRGGGRVSGAVAHAHAHAHA; Y* = DOPA), which we refer to as pC-DN peptide, was synthesized on solid support based on fluorenylmethyloxycarbonyl (Fmoc) chemistry and a 2-(6-chloro-1-H-benzotriazole-1-yl)-1,1,3,3-tetramethylaminium hexafluorophosphate (HCTU) coupling strategy. Briefly, all amino acids (3 equivalents) were coupled using an in situ activation with HCTU and diisopropylethylamine (DIEA) in dimethylformamide (DMF). Each coupling was monitored by the ninhydrin test and coupling was repeated until completion. Fmoc‐Dopa(acetonide)‐OH was incorporated at positions 2, 6, 11, and 14. The peptide was cleaved from the Rink Amide AM resin with a mixture of trifluoroacetic acid (TFA), ethanedithiol, phenol, and water (92.5/2.5/2.5/2.5 [vol/vol]). Crude peptide was purified by high-performance liquid chromatography (HPLC) using a C18 column and a linear gradient of acetonitrile in H₂O/TFA (0.06%, vol/vol). Collected fractions were analyzed with analytical HPLC using a C18 column and “time of flight” mass spectrometry (LC/MS-TOF). Fractions corresponding to the desired peptide with a purity of over 95% were pooled and lyophilized (SI Appendix, Fig. S1).

Peptide Cross-Linking Studies.

The synthetic DOPA-containing peptide derived from the N terminus of preCol-D, pC-DN, was investigated with several techniques under different treatment conditions. Peptide solutions of ∼1 mg/mL in water were diluted twofold with either: 1) 10 mM Tris buffer, pH 8; 2) 10 mM citrate buffer, pH 5; 3) 25 mM NaIO₄ in 10 mM Tris buffer, pH 8; or 4) 10 mM dithiothreitol (DTT) in 10 mM Tris buffer, pH 8. Immediately after mixing, UV/vis spectra were collected over a 24-h interval using a Cary 100 Bio UV-vis spectrometer (Agilent) in a 100-µL quartz cuvette with a 1-cm path length. Absorption was measured every 1 nm with a 600-nm/min scan rate.

To study the DOPA–catechol content, 2 µL of each of the treated peptide solutions was blotted onto nitrocellulose paper after 0, 1, and 4 d and stained with NBT according to Paz et al. (25). After a 40-min staining, the nitrocellulose paper was briefly washed and stored in 0.1 M sodium borate buffer, pH 10. DOPA–metal interactions of the peptide were also investigated with UV/vis and Raman spectroscopy. First, 1 mM FeCl₃ was added to ∼0.5 mg/mL peptide solution to a final concentration of 74 µM and then NaOH was added to shift the solution to basic pH, favoring Tris–DOPA–Fe coordination (45). UV-visible absorption spectra were recorded both before and after the addition of ferric iron to either a fresh peptide sample, or to a sample that had been aged for 30 h in 10 mM Tris buffer, pH 8. For Raman spectroscopy, peptide solutions in water with and without ferric iron were dried on a glass surface and single spectra were taken with 1-s integration time and 60 accumulations.

Mass Spectrometry.

Synthetic DOPA-containing preCol-D peptide pC-DN was prepared at a concentration of 100 µM in water and the pH was adjusted to 8.0 by the addition of dilute NH₄OH. At desired time points, 10-µL aliquots were removed and quenched with 90 µL of 0.1% formic acid in water. A 25-µL portion of this sample was injected into a Waters BEH C18 ultra-performance liquid chromatography (UPLC) column (1 × 100 mm, 1.7 µM, 300-Å pores) coupled to a Synapt G2-Si quadrupole time-of-flight mass spectrometer. The UPLC separation was performed using a water/acetonitrile solvent system containing 0.1% formic acid. The peptide was eluted (retention time 2 min) with a linear gradient of 2 to 100% acetonitrile over 10 min. Column eluate was directed to an ESI source operated with the following settings: capillary voltage = 3 kV, cone voltage = 40 V, source offset = 80 V, and source temperature = 150 °C. Mass spectral data were collected in continuum, positive ion, and sensitivity modes. A [Glu1]-fibrinopeptide external standard was used to correct the measured m/z values. For tandem MS studies, the quadrupole was operated with an LM resolution = 25, and was centered on m/z values of either 632.3 or 628.1, corresponding to the center of the isotope distribution for the peptide signals corresponding to the 6⁺ ions of the starting material and the onefold dehydrated peptide, respectively. After passing through the quadrupole, the selected peptide ions were fragmented by collision-induced dissociation using argon as the collision gas and by applying a collision energy ramp of 20 to 30 V applied over a 0.5-s interval. Tandem MS spectra were smoothed and deconvoluted to the [M+H]¹⁺ state using the MaxEnt3 algorithm available in Mass Lynx software (Waters).

Data Availability.

All data discussed in the paper are presented in the figures and SI Appendix. Experimental protocols are outlined in Materials and Methods. Materials will be made available to readers upon reasonable request.