Viewing Hyaluronan

Abstract

Hyaluronan (HA) is an ubiquitous extracellular matrix polymer that plays many roles in health and disease. The ability to view the spatial and temporal expression of HA in tissues and on/in cells has provided researchers with insights into the tremendously diverse biological processes in which HA is involved. Biochemical extraction, quantity, and size measurement of HA can tell part of the story, but these techniques are incomplete in placing HA at the scene of a biological event and determining which other molecules are likely to be cooperating. HA, however, is not immunogenic, so preparing antibodies for histochemistry is problematic. Fortunately, a probe for HA was devised based on the HA binding region of aggrecan, and today this probe is commercially available and very useful for histochemistry. This article discusses the conditions and considerations that the authors’ lab and others have developed for optimal HA staining in many tissues and cell types.

Despite its uncomplicated chemistry, or perhaps as some investigators believe because of it, hyaluronan (HA) is central to many complex biological functions in chordates, including promoting embryogenesis (Camenisch et al. 2000; Shukla et al. 2010), acting as part of the structural extracellular matrix (Timmons et al. 2010; Alaniz et al. 2009), and signaling innate host defense mechanisms (Forteza et al. 2001; Noble and Jiang 2006). HA is a chemically simple, non-branching polymer of repeating units of N-acetyl-glucosamine and glucuronic acid. In normal tissue, HA is usually present as very large molecules (10⁶−10⁷ Da), although during inflammatory processes, the polymers can be cleaved to fragments of lower molecular weight, which take on new cell signaling roles (Stern et al. 2006).

The ability to see the location, time of appearance, and structure of HA in tissues and on/in cells has provided researchers with insights into the tremendously diverse biological processes in which HA is involved. Biochemical extraction, quantity, and size measurement of HA can tell part of the story, but these techniques are incomplete in placing HA at the scene of a biological event and determining which other molecules are likely to be cooperating at the same time. Two examples emphasize this point well (de la Motte et al. 1999; de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003; Kessler et al. 2008). When comparing cultures of intestinal smooth muscle cells that were untreated to those activated with a viral mimic, poly I:C, a 10-fold or greater increase in HA-mediated leukocyte adhesion was consistently observed. Yet only a 2.5-fold increase in cell-associated HA was measured (de la Motte et al. 1999). Through the use of imaging, we gained the insight that much of the induced HA formed a cable structure that could bind non-activated leukocytes, whereas the cell glycocalyx that excludes particles was not leukocyte adhesive. This observation led to the investigation of whether incorporation of binding proteins into the HA matrix—specifically, the heavy chains of inter-alpha-trypsin inhibitor (IαI; de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003) and versican (de la Motte, Hascall, Drazba, Strong 2003; Potter-Perigo et al. 2010)—is important for cable formation and function. Blocking the incorporation of either of these two proteins inhibits leukocyte adhesive function (de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003; Potter-Perigo et al. 2010). However, blocking IαI binding to HA results in loss of cable structure formation, yet coat structures still form. IαI heavy chains are now known to increase the adhesiveness of HA (Zhuo et al. 2006) for leukocytes, which suggests that a major reason the pericellular matrix does not interact with leukocytes while cables structures do is based on the incorporation of IαI heavy chains. Although versican is incorporated in both pericellular coat and cable HA, we envision that versican, as presented on the cable structure, is more accessible for helping to facilitate leukocyte adhesion.

In vivo, similar shifts in HA deposition have been observed in intestinal tissue that would not be reflected in HA measurements. Mouse large intestine normally contains organized HA in the epithelial layer, likely to aid water absorption and maintain flexibility of the organ. During intestinal inflammation, however, the epithelial HA disappears, and new deposition in subepithelial tissue occurs (Kessler et al. 2008), although total HA content in the intestinal wall does not change appreciably. Once again, HA was found to associate with binding proteins, the heavy chains of IαI, in the inflammatory context. Figure 1 is an example of how tissue imaging gave many clues to important inflammation-associated changes in HA that biochemical techniques would not have uncovered. Panels 1a and 1c show different magnifications of cross sections through a normal mouse colon, where organized HA surrounds the epithelial-lined (“e”) intestinal crypts. During inflammation (panels 1b and 1d), the crypts drop out, and differently structured HA is densely deposited beneath the damaged epithelial layer. In addition, HA is deposited in the submucosal space (“s”), which greatly expands during inflammation. Higher magnification of the submucosal tissue shows a change in HA deposition as well as IαI binding protein incorporation during inflammation (typified by the cellular infiltrate, as seen by the increase in the number of blue-stained nuclei).

Figure 1.

Changes in hyaluronan (HA; green) arrangement and distribution are among the most dramatic alterations in the colon during induction of inflammation. (a, c) Untreated control; (b, d) induced colitis. High-magnification images (c, d) reveal that heavy chains of inter-alpha-trypsin inhibitor (IαI; red) are heavily co-localized (yellow) with HA in the submucosa (star) in the inflamed intestine (d) but little in the mucosa (arrowhead) or in the normal control section (c). HA was detected with biotinylated hyaluronan binding protein and Alexa 488–streptavidin and imaged with a Leica confocal microscope. e, epithelial layer; s, submucosa; m, external muscle layer.

Because HA has a very simple, conserved composition and is ubiquitously expressed in all animals that have a developed immune response, HA is not immunogenic. Therefore, there are no antibodies that specifically recognize HA, and traditional immunohistochemical methods of detection of HA are not possible. Fortunately, a very specific and tightly binding protein, the hyaluronan binding protein (HABP), which is composed of the HA binding domain with the link module from aggrecan, was isolated (Hascall and Heinegård 1974; Tengblad 1979) and adapted as an HA probe (Ripellino et al. 1985). HABP is now widely employed for specific detection of HA.

The HABP probe has been successfully used for detection of HA in cell cultures (live [Rilla et al. 2008; Evanko et al. 2009] and fixed [de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003; Kessler et al. 2008; Twarock et al. 2010; Selbi et al. 2006]), as well as in tissue sections (fixed, paraffin-embedded sections as well as cryopreserved, OCT-embedded sections; Noble and Jiang 2006; Stern et al. 2006; McDonald et al. 2008; Auvinen et al. 2005; Lin et al. 1997). Fortunately, HABP coupled to biotin can be purchased from many sources today. The addition of biotin allows detection with common secondary streptavidin reagents that are fluorescent or colorimetric. Radiolabeled streptavidin (e.g., ¹²⁵Iodine-streptavidin) is also available for relative quantification of HA in the cell matrix (de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003). Fixation with alcohols or aldehydes does not appear to affect HABP binding to HA, but the choice of method does affect condensation of the matrix and may alter detection of HA-associated proteins in tissue using specific antibodies. One possible caution investigators must be mindful of is that HA that is highly bound to specific proteins, such as versican, aggrecan, or TSG-6, may be underreported by the HABP.

Prior to the use of HABP, a specific histochemical reaction was used for detection of HA in tissues. In fact, nearly 60 years ago, attempts to determine which reactions actually stained components of mucopolysaccharides, including HA, using histochemical procedures were being reported (Davies 1952). Because the components associated with HA were not fully appreciated at the time, some methods (e.g., periodic acid Schiff method) gave varied results, and the conclusions reported were often imperfect (Zugibe 1962). Staining with the cationic dye Alcian blue became the method of choice for glycosaminoglycan detection, and differential staining of sulfated glycosaminoglycans and HA was reported (Hodson and Prout 1968). The specificity of differential pH staining has been questioned in numerous histological studies (Lin et al. 1997; Derby and Pintar 1978).

Then and now, a critical validation of whether a stained sample contains HA is achieved by abrogating staining with specific hyaluronidase treatment, but not with specific chondroitinase treatment, in replicate samples (Derby and Pintar 1978). Streptomyces hyaluronidase is considered the most specific HA-degrading enzyme, and chondroitinase ABC used at pH 8.6 is the most specific for chondroitin sulfate.

The Process

Fixation

As mentioned above, cell cultures and excised tissue may be preserved in either alcohol-based or aldehyde-based fixatives and successfully used for hyaluronan detection. However, the patterns of staining are specific to the fixative, so direct comparisons between tissues fixed under different conditions are not possible (Lin et al. 1997). Evanko et al. (2009) have recently tested and compared the common in vitro fixation methods used in HA studies. Unlike the other glycosaminoglycans that have covalently linked protein components, HA is not necessarily strongly integrated into the matrix, and washout during processing is possible. Cetylpyridinium chloride (CPC) precipitation of HA in tissue results in superior retention in loose connective tissue (Ripellino et al. 1985; Lütjen-Drecoll et al. 1990; Nishikawa et al. 1996). In relatively dense matrix where HA is heavily linked with other proteins such as CD44, aggrecan, and versican, CPC is not generally used. In addition, CPC is autofluorescent and therefore would interfere with specific fluorescent detection methods.

Our lab methods rely primarily on alcohol-based fixation procedures, with excellent results predominating. Recently, Evanko et al. (2009) have demonstrated in comparison assays that acid-formalin-ethanol fixative (3.7% formaldehyde-PBS, 70% ethanol, and 5% glacial acetic acid, all v/v) described previously (Lin et al. 1997) is superior in preserving the very fine HA matrix structures and resulted in less HA washout than formalin fixation. For cells grown in monolayer culture, cold (−20C) 100% methanol fixation for 10 min followed by air-drying generally works well for preserving the lengthy HA cable structures. Using alcohol fixatives as opposed to aldehydes has the added benefit that antigen retrieval—sometimes required because aldehydes can crosslink proteins and change the antigenic structure—can be avoided when antibody co-staining in the same preparation.

For colon tissue preservation of HA, we obtain excellent results by fixation with Histochoice, a product that contains a low concentration of the aldehyde, glyoxal. Freshly dissected tissue submerged in Histochoice (>10 volumes Histochoice/1 volume of wet tissue sample) may remain for 1 and 7 days at room temperature before processing and paraffin embedding. However, Histochoice may not be optimal for preservation of HA in all tissues, and most investigators recommend directly comparing fixatives for specific tissues and species initially before conducting large experiments. Formalin fixed or fresh-frozen tissues also successfully stain for HA. In our application, formalin fixation results in higher background autofluorescence that interferes with our confocal fluorescence microscopy results, whereas the fresh-frozen method compromises tissue morphology. The choice of preservation method can also be dependent on which antibodies are being used for co-detection of molecules in addition to HA.

Staining

Commercial preparations of HABP form a very stable complex with HA. As mentioned, the addition of biotin allows easy detection of the probe. (There is widespread, successful use of the commercial reagents from Associates of Cape Cod, Inc. [East Falmouth, MA] and Calbiochem [La Jolla, CA].) Most published accounts use the HABP probe at a concentration of 1 to 5 µg/ml (de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003; Rilla et al. 2008; Evanko et al. 2009), generally suspended in a saline solution (e.g., PBS, Hank’s balanced salt solution) containing protein (usually non-immune serum or albumin) to reduce nonspecific attachment of the binding probe to tissue or the slide. We and other investigators have found that binding of the HABP probe to HA does not seem to occur as rapidly as antigen-antibody reactions. Optimal staining is achieved after 12- to 16-hr incubation times. For the HA labeling step, the sample is incubated at 4C to prevent bacterial growth and in a humidified chamber (i.e., a sealed plastic container with damp paper towels in the bottom) to prevent drying. After incubation, slides or cover slips carrying the samples are washed and reincubated (30 min, ambient temperature) with streptavidin that is conjugated to a fluorescent label (e.g., Alexa 488 Streptavidin [green] or Alexa 568 Streptavidin [red], both from Molecular Probes [Eugene, OR], used at 1:500 dilution in a saline solution) for fluorescence microscopy detection. Alternatively, using streptavidin linked to an enzyme label (e.g., peroxidase) that is further processed for a colorimetric reaction can also yield clear HA detection results (Auvinen et al. 2005).

Imaging

Hyaluronan can form huge (at least on a cellular scale) structures in tissue and in the matrix of cell cultures (de la Motte, Hascall, Drazba, Bandyopadhyay, et al. 2003). The HA structures are three-dimensional, even in fixed cells or tissue where the matrix has been condensed. Observing stained HA on a traditional microscope, whether brightfield or fluorescence, is possible with continuous focal plane adjustment, but the observed images all contain out-of-focus light from adjacent planes of focus, and capturing a clear image of the whole three-dimensional molecule is impossible. To avoid this limitation, we use a confocal microscope that allows multiple planes through the sample to be captured—each in focus—and then viewed as a series, integrated into a single image, or used to create three-dimensional reconstructions that can be examined from all sides. Figure 2a illustrates HA structures produced on the surface of tumor necrosis factor (TNF)–α–activated endothelial cells and the important role of confocal microscopy in visualizing the extensive matrix. The single planes capture different information that can be integrated into an overlay that best represents the whole cable structure. When fluorescently tagged antibodies are used to label other proteins in conjunction with HABP labeling, the multicolor confocal images provide a wealth of information about the structural associations of the tagged molecules (Figure 2b).

Figure 2.

Confocal microscopy aids in visualizing different cellular configurations of the hyaluronan (HA; green) matrix. Panel a shows representative serial slices and the projection of the 16-micron z-stack from tumor necrosis factor (TNF)–α–activated microvessel endothelial cells to which leukocytes have bound. Different structural features are evident in different focal planes, whether close to the endothelial surface or wrapped around the leukocytes above the monolayer. The features are more evident in panel b, the same z-series in which cell location is identified by nucleus staining (DAPI stain—blue), and leukocytes are labeled with CD44 antibody (red). Imaging has led to the understanding that activated endothelial monolayers produce an HA matrix that contains different configurations: (1) cable-like (arrowhead) structures that bind leukocytes (red) and (2) cell surface coat-like (star) arrays that are not leukocyte adhesive.

HA can also be tagged and imaged in live-cell cultures. This can be achieved either by (1) adding a fluorescently conjugated protein that binds to HA, such as Neurocan–green fluorescent protein (GFP) fusion protein (Zhang et al. 2004) or fluoresceinated HABP (Ripellino et al. 1985), both used successfully to image real-time HA changes in live cultures, or (2) adding commercially available biotinylated HABP directly to the culture medium, followed by a gentle wash, incubation with a fluorescently tagged streptavidin, and another wash at set times of interest. The live cells are best imaged using a spinning disk confocal or a point-scanning confocal with a resonant scanner that can collect stacks of images very rapidly at regular intervals over time. Such time-lapse imaging allows for the study of factors that affect HA formation and that influence its cellular interactions while the events are occurring.

HA is so much more than a space-filling matrix molecule, although that role is also important. Increasingly, investigators are appreciating that HA is a cue that activates or inhibits physiological as well as disease responses (Stern et al. 2006). In addition to changes in quantity, HA’s spatial and temporal expression, as well as the proteins with which it associates, all affect the function. The ability to specifically label and image HA has aided many groups in defining these parameters and contributed to understanding the polymer’s role in normal physiology and development, as well as during disease.