Methods for isolating and analyzing physiological hyaluronan: a review

Abstract

The carbohydrate hyaluronan (or hyaluronic acid, HA) is found in all human tissues and biofluids where it has wide-ranging functions in health and disease that are dictated by both its abundance and size. Consequently, hyaluronan evaluation in physiological samples has significant translational potential. Although the analytical tools and techniques for probing other biomolecules such as proteins and nucleic acids have become standard approaches in biochemistry, those available for investigating hyaluronan are less well established. In this review, we survey methods related to the assessment of native hyaluronan in biological specimens, including protocols for separating it from biological matrices and technologies for determining its concentration and molecular weight.

INTRODUCTION

Hyaluronan (or hyaluronic acid; HA) is a linear glycosaminoglycan (GAG) composed of the alternating disaccharide repeat structure [-4-d-glucuronic acid-β1-3-N-acetyl d-glucosamine-β1-]_n that was first reported nearly a century ago by Meyer and Palmer (1), who derived its name from the biological matrix from which it was sourced (“hyaloid,” for vitreous humor) and the major chemical constituent it contained (uronic acid). Structurally, HA does not exhibit the substituent sulfation that is typical of other GAGs (2–4), resulting in a consistent chemical configuration and negative electrostatic charge at physiological pH. As a semiflexible anionic polymer, HA forms a hydrated entropic coil in solution that can produce nonlinear viscoelastic properties through coil-coil interactions (5). In free solution, this behavior is driven by physical crowding, dictated largely (6) by HA concentration and polymer size (i.e., larger molecules occupy more entropic volume and promote domain overlap). However, in physiological systems, the activity of binding entities is another major contributing factor. Although the only natural covalent alteration known to impact HA is the attachment of the inter-α-inhibitor heavy chain (HC) domain by the catalytic activity of the tumor necrosis factor-inducible-gene 6 (TSG-6) enzyme (7), a variety of noncovalent interactions with proteoglycans and other proteins (8) are also known to occur, facilitating the variation of intermolecular interfaces and consequently of the local physicochemical environment in vivo.

As a component of diverse mammalian tissues and biofluids (9, 10) [as well as other, nonmammalian systems (11–13)], HA exists in various quantities (14) and is generally polydisperse, covering a broad range (9) of molecular weight (MW) from ∼10⁵ to 10⁷ Da (∼250–25,000 disaccharide units). In its ubiquity, HA is known to play critical roles in diverse biological functions like extracellular matrix structure (9), hydration and turgidity maintenance of tissue (9), regulation of innate immunity (15), tissue regeneration and repair (16), and the protection and lubrication of joints (17), each of which can be impacted by its particular size distribution (i.e., low vs. high MW). This diversity also makes clear the corresponding importance of HA to pathological processes, and its abundance and/or size distribution have indeed been identified as a major factor in diseases that include liver fibrosis (18, 19), amyotrophic lateral sclerosis (20), multiple sclerosis (21), acute kidney injury (22), myocardial ischemia (23), osteoarthritis (24–26), and various forms of cancer (27–31). The importance of HA as a potential bioindicator necessitates robust technologies for elucidating its complex interactions with the physiological environment and subsequently for assessing it in clinical specimens toward patient diagnosis and prognosis. Here, we will review the major approaches relevant to this goal, including both the purification of HA from diverse biological specimens and the determination of its properties.

HA ISOLATION AND EXTRACTION METHODS

HA analytical techniques typically require pure samples, either due to intrinsic nonselectivity or potential interference from background contaminants. For this reason, pretreatment of native biospecimens is critical. Broadly, the approaches for HA separation from its physiological matrix can be classified in two groups: isolation, or the removal of nontarget components to leave pure HA behind, and extraction, or the specific retrieval of HA from a complex milieu based on its physical or chemical properties.

Toward isolation, a variety of chemical and biochemical techniques have been adapted to HA since the earliest reports of its retrieval. These approaches necessitate some protocol differences based on the properties of the starting material; for example, bone (32) is flash frozen and powdered by milling, blood (33) or conditioned media (34) are centrifuged to remove cells, and egg shell (35, 36) is decalcified by acetic acid or EDTA. However, outside of these pretreatments, a common set of treatments is generally used, either concurrently (37) or more typically in a stepwise manner. First, lipids can be removed by a variety of solvents, including acetone (1, 13, 38–41) or chloroform-methanol (35, 42–44) followed by enzymatic digestion of protein components, often using broad-spectrum proteases such as proteinase K or pronase (40, 42, 45–51). It has been noted (45) that some broad-spectrum proteases can induce HA degradation (likely due to contaminating glycanases, highlighting the utility of process validation with “spiked” HA internal standards) and so a more directed protease (32, 35, 36, 41, 44, 52) may also be selected depending on the specimen type. Finally, nucleic acids are digested by nuclease treatment (33, 47, 53–56) and then all utilized enzymes are commonly inactivated by heat (41, 48, 53), leaving GAGs in solution with the remaining digest and cleavage products. If necessary (dependent on protease selection), the GAGs may be further processed to remove core protein peptide fragments through β-elimination (40, 44) or other chemical treatments (13, 32), but are then generally separated from residual material via chloroform extraction and/or ethanol precipitation (13, 34, 49, 52) or with detergents (35).

Because all GAGs coprecipitate with these recovery strategies, the challenge then becomes removal of off-target species to isolate HA alone. In principle, this could be achieved with selected GAG lyases (57), but in practice, many tend to have cross-reactivity with HA that can result in unintended degradation. To circumvent this, the unsulfated nature of HA has been exploited through strategies designed to specifically remove sulfated GAGs, including chemical fractionation with either cetylpyridinium chloride (58) or ethanol (59) and ion exchange chromatography (39).

Although historically critical to the study of HA, isolation strategies have significant drawbacks. For example, they are limited in their ability to achieve ultrahigh purity HA samples, because non-HA GAGs can display variable sulfation (2, 60), off-target contaminants can be co-segregated with HA (51). In addition, the multiple steps and purifications generally required for the implementation of these approaches can be lossy, reducing HA yield, especially in biological matrices with low relative starting abundance. For these reasons, extraction techniques are more common in modern processes.

Immunoprecipitation (IP) has been used commonly for small-scale affinity purification and isolation of proteins. In the general technique, target-specific antibodies are immobilized on a substrate (e.g., on gel/resin or beads) and used to bind to the protein of interest selectively from a mixture. Once captured, excess material is removed (or, in the case of beads, the substrate is pelleted and washed) and target proteins are eluted to recover the target material, providing high-purity product. The IP concept was first applied to HA extraction by Yuan et al. (46) who substituted antibodies with hyaladherins to achieve selective capture of HA from human breast milk. As a substrate, commercial streptavidin-coated paramagnetic beads were used after conjugation to highly HA-specific versican G1 domains (VG1) via a biotin moiety, enabling fast pelleting under an applied magnetic field. To avoid interference, isolation steps (like those described earlier) were used to remove or degrade macromolecules (lipids, proteins, and nucleic acids) and salts from the raw specimen. However, cleanup requirements were less stringent due to VG1 specificity. Collected HA could subsequently be eluted via heat denaturation of the VG1. This versatile approach (or slight variations of it) has been extended to extract HA from several complex biological specimens including synovial fluid (50, 61), blood, urine, ovarian tissue (62), and mouse brain (56).

Some key challenges remain to be addressed with the existing IP protocol. For example, elution by heat denaturation renders the VG1 permanently inactive, preventing repeat captures and increasing reagent costs. Other approaches like proteinase K degradation have proven viable but have the same difficulty. In addition, the recovery fidelity of HA capture must be more fully validated; due to the nature of diffusion kinetics, low-MW (LMW) HA moves faster through solution than high-MW (HMW) HA and this can allow LMW HA to access binding sites more efficiently, creating potential bias that will persist through downstream analyses. Care must be taken to ensure captured HA is representative of the native distribution.

QUANTITATIVE MEASUREMENT APPROACHES FOR HA

Because HA lacks chemical variation, analytical techniques focus on two main values of critical importance to its biological activity: abundance and MW. In this section, we describe approaches for probing each of these factors in turn.

HA Concentration Determination

Physiological HA experiences a high rate of turnover (9) that is governed by associated metabolic (63, 64) and catabolic (65–67) pathways. In healthy individuals, a homeostatic balance is generally maintained such that HA concentration tends to be relatively consistent in a given tissue or biofluid (10). However, pathologies that impact either the production or (more often) the clearance of HA can have a significant impact on its abundance in vivo. For this reason, changes in HA concentration have been associated with several disease states in humans, including liver diseases [cirrhosis (68, 69) and fibrosis (70)], osteoarthritis (71, 72), and certain cancers (73, 74), and is currently in clinical use as a urinary biomarker of bladder cancer (75, 76).

There are several options for HA quantification in biospecimens. Although conventional techniques used in biochemistry have been applied, such protocols generally have critical limitations. For example, high-performance liquid chromatography (HPLC) has recently been implemented with ultraviolet detection to probe HA viscosupplements (77), but the potential for interference from background constituents in complex biological specimens reduces specificity and necessitates comparative analyses on the same sample both with and without hyaluronidase treatment to assess HA. As a result, this method has reduced translational capacity. Mass spectrometry has also been applied (78, 79) but its dynamic range has an upper limit of ∼10–50 kDa. Consequently, it typically requires complete hydrolysis of the HA, similar to HPLC, thus precluding downstream size analysis. For these reasons, alternative methods for quantification are typically used for HA derived from biospecimens.

Imaging modalities.

A number of imaging techniques have been used for probing HA. For example, histological analysis, through which HA in cell culture or tissues is labeled for direct imaging, has been demonstrated for a wide range of specimens (80–86). Because HA-specific antibodies are not available (HA is a “self” molecule) (87), labeling of HA has been enabled instead by highly specific HA binding proteins, especially the cartilage link protein (88), the aggrecan terminal fragment globular domain 1-interglobular domain-globular domain 2 (G1-IGD-G2 or HA binding protein, HABP) (89), and VG1 (90). Of these, HABP and VG1 have comparably high specificity and affinity, but the cartilage link protein is less efficacious and in modern protocols is generally avoided (and is a common contaminant in HABP isolated from bovine nasal cartilage, for example). Although histological HA assessment is often qualitative, content determination can be performed either fluorometrically (91) or using a secondary radiolabeling technique (81), both of which provide the added benefit of also localizing HA in the tissue. However, among the significant disadvantages of this approach are its inability to deliver direct quantification, instead of assessing HA levels relative to an internal standard or between conditions, and its limitations with viable biological matrices (i.e., biological fluids cannot be assessed). A second imaging modality that has been applied to HA quantification is magnetic resonance imaging (MRI) supplemented with chemical exchange saturation transfer (CEST) (92) wherein proton spins in HA are saturated and exchanged with the bulk water surrounding them, resulting in an enhancement in image contrast. Quantification of HA can be derived subsequently from the integrated CEST signal. Due to the noninvasive nature of MRI, this approach has the major advantage of being applicable to live imaging of tissues (23), especially monitoring the behavior and lifetime of implanted HA-biomaterials (i.e., macroscopic amounts of relatively pure HA). However, a significant drawback is its limited specificity for HA; CEST results in an image enhancement for exchangeable protons in general and so off-target analytes like proteins and other GAGs can contribute to the signal, strongly impacting accurate HA quantitation. In addition, like all imaging modalities, CEST cannot provide HA MW information.

ELISA-like assays.

The most widely used techniques for the determination of HA concentration are extensions of the enzyme-linked immunosorbent assay (ELISA), the antibody-based approach initially developed for quantifying protein antigens (93). As described earlier for histological imaging, ELISA quantification uses HA binding proteins like the cartilage link protein, the HABP portion of aggrecan, and VG1. Two ELISA-like assay formats are typical for HA: sandwich assays (94–99) (Fig. 1A) in which sample HA is introduced to a surface-bound binding agent and a second, labeled binding agent is then provided to enable quantification through double binding, or competitive assays (97, 100–104) (Fig. 1B) in which labeled binding agent is introduced to surface-bound HA and then its displacement through competition with sample HA is used to determine quantity. Several strategies have been demonstrated for labeling HA binders, including radiolabeling, fluorescence, biotinylation (with subsequent streptavidin labeling), and chemiluminescence. Although both sandwich and competitive ELISA-like assays are common and offer high concentration resolution (typically 25 ng/mL) (105), it has been noted that sandwich assays in particular are biased toward high-molecular weight HA (96, 99, 106) and thus fail to quantify smaller fragments accurately.

Figure 1.

A: sandwich enzyme-linked immunosorbent assay (ELISA)-like assay wherein surface-bound hyaluronan (HA) binding proteins i) are contacted with specimen HA ii) and labeled (green) secondary HA binding proteins are introduced for quantification iii). B: competitive ELISA-like assay in which surface-bound HA binding proteins preloaded with standard HA and labeled secondary HA binding proteins i) are contacted with specimen HA ii) that competes for labeled entities and reduces signal. C: AlphaScreen construct composed of a donor bead with bound HA (right) that creates singlet oxygen upon optical excitement (ɣ), diffusing to an acceptor bead (left) held proximal by a surface-bound HA binding protein thereby producing emission (yellow). Competitive binding with specimen HA will reduce signal as in (B).

AlphaScreen.

An emerging alternative approach to HA quantification is an adaptation of the Luminescent Oxygen Channeling Immunoassay (107), conducted commercially for drug discovery under the name AlphaScreen. In the general procedure (Fig. 1C), two types of beads are used: donor beads loaded with chromophores that can generate singlet oxygen when excited optically; and acceptor beads loaded with separate chromophores that react with singlet oxygen to chemiluminescence. The reaction is proximity-dependent due to the short lifetime of singlet oxygen and so only occurs efficiently when the two beads are physically separated by no more than ∼200 nm.

In the modified assay developed for HA (108), the donor beads are conjugated to short biotinylated HA (50 kDa) through streptavidin coupling and the acceptor beads are bound to HABP, enabling the HA-HABP interaction to keep the beads in close proximity and producing a robust chemiluminescent signal. Introduction of unknown sample HA competes for HABP, releasing the bead complex and reducing the signal in a concentration-dependent manner. Using this approach to quantify HA, AlphaScreen provides similar concentration resolution to ELISA techniques (30 ng/mL) but requires less sample volume (108) (2.5 μL vs. 100 μL) and avoids labor-intensive washing steps of the plate-based methods.

The ELISA-like assays and the AlphaScreen assay are both somewhat tolerant of sample purity due to their respective specificities of binding and so only factors that could impact measurement efficacy need to be managed. For example, HA binding proteins in solution will compete for HA with assay probe proteins and active aggrecanases (109) can cleave HABP and have detrimental effects on HA binding. Therefore, aside from protein degradation, only moderate purification steps are required and should in fact be avoided as possible due to the potential for sample loss that could impact accurate HA quantification. However, a central limitation for each of these techniques is their inability to assess MW as the second major figure of merit for HA. For this, additional techniques are required.

Molecular Weight Determination

HA is naturally polydisperse in all organisms, but the particular MWs present in a given specimen are critical to its behavior. Indeed, a remarkable dichotomy exists in this size-function relationship, with the activity of HMW HA relative to LMW HA (110–116)—or in some cases a narrow distinct MW range relative to other sizes (51, 117)—being contradictory. A critical example of this phenomenon is in inflammation. In its usual HMW conformation in the glycocalyx and extracellular matrix (9), HA acts as a tissue integrity signal and is thus anti-inflammatory in a number of ways (118), including interacting with cell surface receptors to prevent immune cell recognition (119), reducing inflammatory cytokines (120), and inducing STAT5 signaling in regulatory T-cells to promote their maintenance (121). Under inflammatory conditions, however, an abundance of LMW HA (<∼500 kDa) is found, likely due to damage and/or degradation. It has been shown that LMW HA is instead proinflammatory in several respects, including upregulating the transcription of inflammatory genes (e.g., TNF-α, IL-1β) (122, 123) and stimulating growth factors (124) and other inflammatory elements (125, 126) in macrophages. Stark contrasts in size-dependent HA functions have also been reported in processes like cell migration/invasion (127–129), angiogenesis (130, 131), and joint lubrication (120), among others. Consequently, determination of MW is critical to the validity of HA as a translational bioindicator of disease.

Viscometry.

Viscometry is a method to analyze and measure bulk HA solution flow that can yield average molecular weight information (132). As HA MW (and/or concentration) increases, increased polymer chain entanglement will increase viscosity and impede the free flow of the liquid in a way that can be measured. Typically, a purified HA solution is either: 1) flowed through a calibrated tube or capillary with the force of gravity and the speed of transit between two marks is measured; or 2) positioned between one mobile (rotor) and one stationary (stator) plate or cylinder and the forces for starting or stopping the rotor element is monitored. In both cases, HA MW is determined using the Stokes–Einstein equation and tables based on previous model HA studies using biophysical sedimentation. However, viscometry has several practical challenges, including the requirement for large sample sizes (up to 1 g of HA, depending on chain size and the device) and variability induced by the dependence of viscosity on experimental factors like ionic strength and temperature. Moreover, viscometry can only provide a mean MW rather than distribution. Still, industry often uses the technique for product quality checks as it is a simple, well-established method to obtain average MW information from HA.

Multi-angle light scattering.

Light scattering (LS), a method to analyze HA MW from the angular distribution and intensity of photons interacting with macromolecules, is a relatively sensitive process that has the advantage of not requiring comparison to a standard. Two modes of LS are practiced: multi-angle light scattering (MALS) (133) and dynamic light scattering (DLS). Typically, the former is used for HA MW analysis (from ∼7 kDa to several MDa); the latter is more suited for particle sizing (e.g., cross-linked gels). From a practical standpoint, the required HA sample size for MALS depends on the relative MW of the polymer since larger molecules scatter light more strongly than smaller ones. In the simplest embodiment, a known concentration of purified HA (as contaminants can skew results) is injected into a liquid cell and the MW and polydispersity results are obtained based on the intensity and angular distribution of scattered light. The more angles of scattering data that are used, the better the MW determination; currently, modern detectors simultaneously measure from 3 to 17 angles.

Typically, most modern MALS instruments used a HPLC-size exclusion chromatography (HPLC-SEC) approach as both a separation method and “pre-filter” (dust and contaminant particles can overshadow the target HA) as well as an independent means to assess size distribution. In addition, after the MALS detector, an in-line tandem refractometer is used to measure the concentration of HA; this is especially useful for smaller samples (e.g., precious patient extracts and trials of synthetic products) where the exact concentration is not known. To determine MW, the amount of scattered light and concentration are related through a refractive index increment constant (dn/dc) that is derived empirically by injecting a series of HA solutions of known concentration in a defined buffer solution into the MALS cell sequentially. Many workers in the field use a known buffer such as neutral phosphate buffered saline (0.15 M NaCl) with an established dn/dc value (ranging from 0.153 to 0.167) to avoid unknowns and promote reproducibility. In addition, this buffer is often regarded as compatible with physiological processes. A monodisperse protein standard like bovine serum albumin (66 kDa) is often used to align the outputs of the two detectors, the refractive index, and light scattering cells, as well as calibrate MW and thus validate system performance.

Field flow fractionation (FFF) is also used as a separation method for very high MW [>1–2 MDa HA polymers (134)]; in the MDa range, most existing size exclusion resins are not as useful and can cause some chain shearing, leading to a lower apparent MW value. In a simplistic description of a useful permutation of FFF, the HA sample flows down an open channel similar to chromatographic approaches, but the device simultaneously uses an additional perpendicular fluid flow across a semipermeable membrane along the axis of the channel. Because larger molecules are more affected than the smaller molecules by the cross-flow, HA molecules can be separated according to size and quantified separately. Currently, these devices are more expensive and less numerous than HPLC-SEC setups.

One challenge with MALS approaches is their requirement for significant amounts of HA, dependent on MW and size distribution. By way of example, HA synthesized chemoenzymatically in vitro (135, 136) can achieve a polydispersity index (PDI, or the ratio of the number-average MW to the weight-average MW) of 1.005–1.05, close to a truly monodisperse (PDI = 1.0) polymer. For such quasi-monodisperse HA >500 kDa, ∼5–20 μg may be needed for analysis whereas polymers <50 kDa may require 75–100 μg. In physiological HA, the PDI is typically ∼1.3–2.0, and since SEC does not allow the majority of HA to be in the MALS detector cell at once (i.e., the wide peak is spread over the elution time), ∼2–5 times more sample may be needed. MALS measurements are intrinsically rapid but are limited by the HPLC-SEC or FFF step such that each test will typically take ∼15–60 min, and at least duplicate runs should be performed. The absolute MW obtained through MALS is widely recognized as the gold standard, but the accuracy of the size distributions it can deliver is somewhat limited by the separation method resolution.

Gas-phase electrophoretic mobility molecular analysis.

Gas-phase electrophoretic mobility molecular analysis (GEMMA) measures the electrophoretic mobility of single-ionized, spherical molecules and constructs in air to estimate MW. This technique has previously been used to differentiate MWs of globular proteins and viruses (137–139). However, Malm et al. (140) demonstrated that the logarithmic dependence of electrophoretic mobility diameter (EMD) for HA (with the exception of small HA oligomers) can be used to produce a reliable assessment of MW in the range of 30–2,400 kDa, spanning most of its typical physiological range. Like MALS, GEMMA requires a highly purified HA sample as contaminants will skew the results. But in contrast to HPLC-SEC, only a single calibration of analyte electrophoretic mobility is needed rather than routine calibrations, and the resolution of GEMMA is comparable with SEC such that its ability to produce a full MW distribution is limited; it is most effective for comparing distributions of GEMMA spectra between groups (e.g., healthy and pathological tissues).

Measurement time for one spectrum is rapid (5–10 min) and requires as little as 6 pg of total HA, making it an excellent tool for determining the MW of HA from biological specimens with low HA concentrations (141), including tissue samples from small rodents (e.g., mice) and small volume biological fluids (e.g., cerebrospinal fluid and bronchial lavage). However, one intrinsic limitation of the GEMMA approach is the assumption of spherical molecules: variation in instantaneous polymer conformation can produce deviations from this model that may cause measurement inaccuracies. For example, the technique has been shown (140) to underestimate MW for HA molecules > 70 kDa.

Gel electrophoresis.

Gel electrophoresis is a simple and inexpensive method to estimate the MW distribution of unlabeled HA samples ranging in size (142–150) from a few kDa up to at least 6,000 kDa that has been used to analyze HA isolated from numerous tissues and biofluids including skin (49, 151), skeletal muscle (151), heart (151), lung (47, 151), brain (56), small and large intestine (151), mammary gland (54), cervix (152), synovial fluid (10, 50, 61, 144), vitreous (144), milk (46), lymph fluid (153), and the conditioned medium of cultured cells (53, 63, 154–162). The charge-to-mass ratio is constant for all HA molecules due to the homogeneous repeating primary structure containing one negatively charged carboxyl group per disaccharide. Thus, like DNA or denatured proteins, migration of HA through a gel matrix under the influence of an applied electric field allows sieving on the basis of size. At the end of the migration time, cationic dyes like Stains-All can be used to immobilize and label HA in the gel (163). Densitometric analysis of the stained pattern can then allow determination of the mean MW and distribution of MWs (polydispersity) of HA in the sample.

Generally, HA greater than ∼100 kDa is best size-separated and analyzed on agarose gels (Fig. 2) using total HA mass in the range of 0.1–7 μg, depending on polydispersity and staining method. The applicable MW range depends on agarose concentration and buffer. Concentrations of 0.5%–2% agarose are used most commonly, typically with Tris-borate-EDTA (TBE) or Tris-acetate-EDTA (TAE) (150) as buffer. HA from ∼5–100 kDa can be size separated and analyzed best by polyacrylamide gel electrophoresis (Fig. 3) using total HA mass of ∼0.5–1.0 μg. Simple gels containing a constant acrylamide concentration (e.g., 10%) can be used, but commercially available gradient gels (e.g., 4%–20% acrylamide) in TBE buffer can yield excellent separation. All HA larger than ∼100 kDa moves slowly, but the resulting sharp bands can be analyzed to quantify as little as 0.1 μg HA by densitometric analysis. At the other extreme, HA oligosaccharides smaller than ∼11 disaccharides (4.4 kDa) cannot be detected quantitatively by dyes (142) (possibly due to poor immobilization in the gel), but can be analyzed by the fluorophore-assisted carbohydrate electrophoresis (FACE) method (78, 147, 164–170), which uses reducing-end labeling of the oligosaccharides with a fluorescent dye before electrophoresis so that no additional staining and de-staining procedure is needed.

Figure 2.

A: electrophoresis of hyaluronan (HA) on 0.5% agarose in Tris-acetate-EDTA (TAE) buffer, stained with Stains-All dye. The chemoenzymatically synthesized HA standards appear as sharp bands, but HA isolated from tissue or bacterial sources appears highly polydisperse. The viscosity-average molecular weight (MW) of the two polydisperse samples is indicated. B: calibration plot for electrophoretic migration of HA standards. Migration distance for each band was normalized to the migration distance of the smallest HA species. Adapted with permission from Ref. 144. Copyright 2011, Elsevier Inc.

Figure 3.

A: electrophoresis of hyaluronan (HA) on a 4%–20% polyacrylamide gradient gel in Tris-borate-EDTA (TBE) buffer, with detection by Stains-All dye. Chemoenzymatically synthesized HA standards appear as discrete bands, but HA samples obtained by controlled degradation of high-molecular weight (HMW) HA appeared highly polydisperse. The viscosity-average molecular weight (MW) values of the polydisperse samples are indicated. B: calibration plot for electrophoretic migration of HA standards. Adapted with permission from Ref. 143. Copyright 2011, Elsevier Inc.

HA standards must be co-electrophoresed in the same gel used for unknown samples to create a standard curve for size analysis. Chemoenzymatically synthesized quasi-monodisperse HAs (135, 136) are good for this purpose. Sample purification is also a consideration but HA does not need to be highly pure to be analyzed by gel electrophoresis, especially if specific detection is used. However, it must be free of strongly bound proteins like aggrecan or HC, which can cause an electrophoretic mobility shift (149), and non-HA anionic contaminants that can be stained by nonspecific dyes like Stains-All or Alcian Blue +/− silver staining. An HA-specific hyaluronidase from Streptomyces [but not testicular hyaluronidase as it can also digest chondroitins (171)]) can be used to confirm the identity of bands (46). Notably, a detailed protocol for agarose gel electrophoresis of HA has been published previously (150). Instructions and spreadsheets for analysis of densitometric scans, and calculation of weight-average and number-average molecular masses have been included as Supplemental Material files in a published method (148).

Nanopore analysis.

Nanopore analysis is an emergent single-molecule approach in which a thin membrane containing an individual, nanometer-scale aperture is positioned between two reservoirs of electrolyte solution and an applied voltage is used to transport molecules introduced to one reservoir through the pore electrophoretically (Fig. 4A). The translocation creates a temporary reduction in the measured ionic current passing through the pore (Fig. 4A, insets), the characteristics of which can convey properties of the passing molecule: for example, signal duration indicates molecular length and signal amplitude (blockage depth) corresponds to molecular diameter. Currently, there are two classes of nanopore platforms: biological nanopores (173–175), composed of a porin protein inserted in a suspended lipid membrane, and solid-state (SS-) nanopores (176, 177), composed of single apertures fabricated in a (usually silicon-based) thin-film membrane. Although biological pores have been applied to HA measurement (178), they are generally delicate (due to the lipid membrane) and have yielded limited size capacity. Consequently, our discussion here will focus on their artificial counterparts.

Figure 4.

A: schematic of the SS-nanopore measurement approach showing a hyaluronan (HA) (green) pulled electrophoretically through a nanometer-scale pore (blue). Top inset: example current trace showing molecular translocation signals (spikes). Bottom inset: zoom of a translocation signal indicating event charge deficit (ECD) (green). B: calibration plot of ECDs for chemoenzymatically synthesized HA standards. Line is a power law fit to the data. Reprinted from Ref. 50 under the Creative Commons CC BY license. C: example molecular weight (MW) distribution histograms collected for HA derived from bacterial membrane fragments for indicated times. Reprinted from Ref. 172 under the Creative Commons CC BY license.

SS-nanopores can be engineered to various dimensions, but HA analysis is typically performed using pores ∼7–10 nm in diameter (50). Because the cross-section of a single HA chain is much less than this (<1 nm) (179, 180), molecules are unlikely to translocate linearly, but rather pass in a random, folded conformation that results in complex electrical signatures. To account for this, signal area [called event charge deficit (181), or ECD; see Fig. 4A, bottom inset] is considered—since folding (higher amplitude) will necessarily be offset by shortened signal duration, this metric is a constant for a given molecular size. Indeed, when quasi-monodisperse HA are analyzed, ECD is found to scale with MW according to a power-law dependency (50), providing a quantitative size calibration (Fig. 4B). Consequently, translocation signals can be used to determine HA MW on a molecule-by-molecule basis and a full distribution can be attained by analyzing a representative sampling of HA in a specimen. Commonly, this is achieved through consideration of at least 500 molecules (and more often many thousands), which is measurable in a typical timeframe of 15–20 min from HA samples as small as 10 ng total (50). There is also potential for this sensitivity to be improved further through enhanced extraction or manipulation of experimental variables (e.g., buffer conditions, voltage, etc.).

Notably, the lower MW limit for SS-nanopore measurements is determined by the signal-to-noise ratio (SNR) of the electrical signal. Under common measurement conditions, this enables HA as low as 54 kDa to be resolved readily (50). However, several strategies could be applied to improve this value, including ultra-low noise SS-nanopore devices (182–184) and high bandwidth electrical systems (185, 186). In principle, there is no upper MW limit to the assessment, but extremely large molecules can promote transient clogging of the pore due to entanglement, potentially resulting in an overestimated ECD and thus MW. Nonetheless, measurements of HA up to ∼20 MDa have been demonstrated (172).

Understanding the molecular translocation process is critical to ensuring accurate data; for example, if LMW HA is more likely to transit the pore than HMW, the resulting distribution will be biased. In the case of HA, a series of measurements have been performed to determine the relevant dynamics, finding that the rate of translocation signals for a given concentration varies linearly with voltage (50). This is indicative of diffusion-limited translocation (187, 188), such that there is no MW dependence of molecular capture. Consequently, SS-nanopores yield faithful size distributions for arbitrary mixtures of HA; a capacity that has been applied to material from a variety of sources that includes equine synovial fluid (50, 61), human ovarian tissue (62), and HA synthesized in vitro from bacterial membrane fragments (172) (Fig. 4C). In addition, the rate of measured translocation signals is known to correlate with HA concentration in solution (50), indicating that SS-nanopores could deliver both MW distribution and quantification in a single measurement if data analysis could compensate for the impact of extraction on the observed concentration.

Conclusions

The myriad physiological roles of HA drives a rising translational interest in the molecule and supports a growing number of use cases that range from structural biomaterial to treatment to diagnostic target. This interest underscores the need for robust analytical approaches to characterize the molecule and as the demand for HA assessment increases, the most appropriate purification strategies and tools will be dictated by the particular needs of the task. For example, in the emerging area of clinical HA diagnostics, evaluation of patient tissues and biofluids will require not only high fidelity and high selectivity extraction processes, but also rapid and sensitive analyses that can provide multidimensional information about the characteristics of physiological HA—especially concentration and polymer size. Of the methodologies surveyed in this review, we believe that SS-nanopore technology may have unique promise for translation given its short measurement time, high sensitivity, broad dynamic range for MW determination, and potential for integrative assessment and automation. However, innovations with other existing technologies continue to be developed as well, suggesting ultimately that improvements in the overall capabilities and availability of HA assessment can be expected. Consequently, it is clear that techniques for probing HA quantity and size will continue to provide new insights into a variety of healthy biological processes and disease pathologies.

GRANTS

The authors acknowledge support from the National Institutes of Health (NIH) Grants R01 GM134226 and P41 EB020594.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

This article is part of the special collection “Deciphering the Role of Proteoglycans and Glycosaminoglycans in Health and Disease.” Liliana Schaefer, MD, served as Guest Editor of this collection.

AUTHOR CONTRIBUTIONS

M.K.C. and A.R.H. prepared figures; F.R., D.E., I.S., E.R., P.L.D., M.K.C., and A.R.H. drafted manuscript; F.R., D.E., I.S., E.R., P.L.D., M.K.C., and A.R.H. edited and revised manuscript; F.R., D.E., I.S., E.R., P.L.D., M.K.C., and A.R.H. approved final version of manuscript.