Hyaluronan (or hyaluronic acid; HA), a component of the extracellular matrix (ECM), is composed of repeating polymeric disaccharides of D-glucuronic acid and N-acetyl-D-glucosamine linked by a glucuronide bond. Hyaluronan is abundant in a variety of human tissues, including skin, umbilical cord, synovial fluid, cartilage, and skeletal tissues, and is turned over through the catabolic activity of hyaluronidases [1]. In addition to its structural role in the ECM, both intact high molecular weight (HMW) HA (>1,000 kDa) and its catabolites (low molecular weight (LMW) HA; 25-1,000 kDa or smaller oligo- and disaccharides) exert various biologic effects [2]. A number of biologic activities have been attributed to HA, including stimulation of inflammation and tumor growth [1,3]. Various reports in the literature suggest different roles for HMW and LMW HA in processes such as inflammation and wound healing. For example, antiinflammatory effects have been suggested for HMW-HA based on its inhibition of leukocyte migration in the murine air pouch model [4], reduction in gene expression levels of cytokines and matrix metalloproteases in arthritis models [5,6], and reduced production of interleukins in a model of ultraviolet (UV) light-induced corneal damage [7]. In contrast, small oligosaccharides derived by in vitro digestion of HA using hyaluronidases have been reported to stimulate inflammation in cell culture [8], and it has been suggested that LMW HA fragments which accumulate at sites of tissue injury and repair may stimulate inflammatory gene expression in macrophages [9]. In addition, a recent report suggests that hyaluronidase may increase inflammation and angiogenesis to promote wound healing in an animal model [10]. Taken together, these reports have raised concerns about potential inflammatory adverse effects when hyaluronidases, such as recombinant human PH20 (rHuPH20), are used therapeutically. However, the true effects of LMW HA and hyaluronidases on inflammation are unclear and, in many cases, the reported effects may be attributable (at least in part) to proinflammatory contaminants, particularly endotoxins and peptidoglycans, in many commercially available hyaluronidase reagents which are often used to generate HA fragments in experimental models. Indeed, the available evidence shows that the pro-inflammatory activity of commonly used test reagents is directly correlated with their level of endotoxin and peptidoglycan contamination (Table 1) [2].

Table 1: Endotoxin and peptidoglycan levels in test reagents.

These findings have important implications with respect to the clinical use of rHuPH20. In recent years, the development of rHuPH20 has facilitated subcutaneous administration of monoclonal antibodies such as trastuzumab and rituximab [11-13]. In addition, high levels of HA in tumors is associated with treatment resistance to monoclonal antibodies and poor clinical outcomes. This has led to ongoing research investigating the use of pegylated rHuPH20 (PEGPH20) to temporarily deplete HA from the tumor microenvironment, thereby improving the penetration of systemic anticancer treatments into solid tumors [14]. The hyaluronidase PH20 works exclusively in the extracellular space to progressively degrade HMW HA into LMW HA. We have conducted extensive research to characterize the HA catabolites generated by rHuPH20 and their biologic activity in a murine model of inflammation. Although digestion of HA using PH20 in vitro can result in generation of small oligosaccharides (<5 kDa), this enzyme typically does not generate appreciable levels of oligo- and disaccharides in vivo (Huang Z, et al; unpublished data). We summarize here recent preclinical and clinical data demonstrating that rHuPH20 does not elicit pro-inflammatory reactions.

The pro-inflammatory effects of LMW HA have been proposed to be mediated via engagement of toll-like receptors (TLR)2 and TLR4, and both CD44 dependent and independent mechanisms have been reported [8,9]. However, there is no direct evidence that LMW HA binds to TLR2 or TLR4, and some reports have suggested that if endotoxin (a putative stimulator of inflammation) is reduced to extremely low levels in purified preparations, LMW HA does not trigger normal tissue macrophage-mediated inflammatory reactions [15,16]. Our research clearly demonstrates that endotoxin-free LMW HA fragments (molecular weight range: 5-500 kDa) do not bind either TLR2 or TLR4; nor do these fragments stimulate TLR2- or TLR4- mediated nuclear factor-kappa-B (NF-kB) pathway signaling [3]. We further found that rHuPH20 and endotoxin-free LMW HA do not induce inflammation in established models such as the murine air pouch model [2,3]. However, bovine testicular hyaluronidase (BTH) was found to be pro-inflammatory by stimulating the TLR-mediated NF-kB pathway [3]. Notably, in the murine air pouch model, rHuPH20 mostly generated LMW HA fragments in the molecular weight range of 100 kDa to 1,000 kDa [3]. This is very different from HA oligosaccharides in the 3 5 kDa range generated by BTH in vitro, which some groups have shown are able to promote TLR-dependent NF-kB pathway activation [8].

These studies further suggest that reagent contaminants, including endotoxin and peptidoglycans, may be responsible for the observed inflammatory effects of commercially available hyaluronidases, particularly BTH, which is a commonly used reagent for in vitro HA research. We also found that BTH types I S and IV S (both of which induce inflammation in the murine air pouch model) each contain more than 60 different protein contaminants with various biologic activities. Some of the key contaminants are shown in Table 2 and a detailed list has been reported previously [3]. In addition, both BTH type I-S and IV-S contained endotoxin (Table 1) [3]. Depletion of endotoxin substantially reduced neutrophil infiltration associated with these BTH preparations in the murine air pouch model [3], suggesting that endotoxin is a major contributor to the observed proinflammatory effects of these reagents. Although BTH type VI-S is free of endotoxin, it also stimulated inflammation in the murine air pouch model [2]. Further analysis of this reagent identified substantial levels of peptidoglycans (Table 1), which activated both TLR2 and TLR4 and induced neutrophil infiltration in the murine air pouch model. In contrast, neither rHuPH20 nor LMW HA fragments, which are both free of endotoxin and peptidoglycan, stimulated neutrophil infiltration or chemokine/cytokine production in that model [2,3].

Table 2: Partial list of protein contaminants in bth type iv-s.

The potential confounding effects of contaminants in hyaluronidase reagents are not limited to studies of inflammation. A recent report suggested that HA fragments generated by PH20 in vitro inhibited differentiation of oligodendrocyte progenitor cells (OPC) and subsequent remyelination of brain lesions in a mouse model [17]. However, the key experiments characterizing the effect of HA fragments in this remyelination model used BTH-digested HA rather than rHuPH20-digested HA or purified LMW HA fragments, again raising the possibility that the observed effect might have been influenced by reagent contamination. We have subsequently found that BTH preparations contain basic fibroblast growth factor (bFGF; Huang Z, et al; unpublished results), which could potentially block OPC differentiation [18]. Taken together, the available evidence underscores the need for appropriate purification of reagents, together with extreme caution and extensive experimental controls when using commercial BTH preparations to study effects on inflammation and other cellular functions in preclinical models.

Clinical studies with rHuPH20 have not shown any safety signals indicative of inflammation-related events. The first clinical study using rHuPH20 with trastuzumab delivered subcutaneously did not have any serious adverse events (SAEs) or severe hypersensitivity reactions [19]. Continuing in the development program, there were no cases of severe hypersensitivity reactions, such as anaphylaxis in the pivotal phase 3 HannaH trial comparing SC versus intravenous (IV) trastuzumab formulations. At 12 months median follow-up, only 11% of patients reported injection-site pain, which was predominantly Grade 1 [12], which was further confirmed with the 20-month median follow-up [13]. The Phase 3 SABRINA trial that assessed SC versus IV rituximab with rHuPH20 in combination with chemotherapy also showed that both formulations had similar tolerability; the proportion of patients with Grade 3 or worse AEs or SAEs did not differ between groups [11].

In conclusion, current data suggest that contaminant-free LMW HA and hyaluronidase PH20 have no pro-inflammatory activity, and that neither rHuPH20 nor PEGPH20 stimulate inflammation. Purity of reagents and use of appropriate experimental controls are critical to determining the effects of hyaluronidases and HA fragments on various biologic processes.

We thank H. Michael Shepard, PhD, Monica Zepada, PhD, Sam Dycther, MD, Douglas Muchmore, MD, Dan Maneval, PhD, Nerash Nayyar, Fred Drake at Halozyme Therapeutics, Inc., for scientific suggestions; and Shalini Murthy, PhD at ProEd Communications, Inc., for medical editorial assistance with this manuscript.

The author is an employee and stockholder of Halozyme Therapeutics