Induction of PNAd and N-acetylglucosamine 6-O-sulfotransferases 1 and 2 in mouse collagen-induced arthritis

Chronic inflammatory diseases such as rheumatoid arthritis (RA), asthma, inflammatory bowel disease (IBD), and multiple sclerosis still pose a large unmet medical need despite recent therapeutic advances such as inhaled steroids (asthma) or TNFα antagonists (RA and IBD). Thus, significant subpopulations of patients, in particular those with severe disease, respond only poorly to these treatments [1, 2]. Furthermore, patients treated with TNFα antagonists are at risk for serious infections [3]. Therefore, modulation of leukocyte-endothelial adhesion, an obligatory step in the recruitment of inflammatory cells to lesions, has been widely considered as an alternative and perhaps complementary approach for therapy of chronic inflammation [4, 5]. One of the molecules involved in leukocyte trafficking is L-selectin, a member of the selectin family of cell adhesion molecules, which is expressed on leukocytes [6]. During the process of lymphocyte homing to lymph nodes, L-selectin mediates rolling of lymphocytes on high endothelial venules (HEV). This is the first step in a cascade of adhesion and signaling events that culminate in the recruitment of both naïve and central memory lymphocytes into lymph nodes [7]. The major class of ligands recognized by L-selectin consists of a family of sialomucins defined by the adhesion-blocking antibody known as MECA-79. Collectively these ligands are termed peripheral node vascular addressin (PNAd) [8] or sulfoadhesin [9]. One of the shared features of these ligands is 6-O-sulfated N-acetylglucosamine, which is essential for antibody as well as L-selectin binding [10–12]. This modification is found on 6-sulfo sLe^x which is a minimal recognition determinant for L-selectin [13, 14]. The 6-O-sulfation of N-acetylglucosamine of PNAd components occurs in the Golgi compartment and is catalyzed by Golgi-associated N-acetylglucosamine 6-O-sulfotransferases (GlcNAc6STs) [15–17]. Using gene deletion by homologous recombination in mice, we have shown that GlcNAc6ST-2, the high endothelial cell restricted N-acetylglucosamine 6-O-sulfotransferase also known as HEC-GlcNAc6ST, LSST, or GST-3 (gene name chst4 in mouse) is largely responsible for the GlcNAc-6-sulfation of PNAd and contributes substantially to L-selectin ligand activity and MECA-79 reactivity [18, 19]. A related enzyme known as GlcNAc6ST-1 or GST-2 [20] (gene name chst2) also contributes to sulfoadhesin biosynthesis but to a lesser degree [17, 21, 22].

While being constitutively expressed in the HEV of lymph nodes and other secondary lymphoid organs, the induction of PNAd has been reported in activated vessels in synovial biopsies from RA patients [23–26], in a model of Lyme disease arthritis in severe combined immunodeficient (SCID) mice infected with Borrelia burgdorferi (B. burgdorferi) [27], as well as in many other inflammatory lesions [28]. In addition, extralymphoid induction of PNAd in inflammatory lesions was shown to correlate with the de novo expression of GlcNAcST-2 in human RA [25] as well as animal models of autoimmunity [16, 29]. These findings suggested, that blockade of PNAd, either directly, or indirectly through inhibition of the responsible sulfotransferase(s), might be efficacious for anti-inflammatory therapy [30]. As any drug discovery effort relies on robust and predictive animal models, we have studied the expression of sulfoadhesin and GlcNAc6ST-1 and -2 in murine collagen-induced arthritis (CIA), a widely used animal model which is predictive for therapeutic benefit in human rheumatoid arthritis [31–33]. Our data show that PNAd is expressed in this model in arthritic but not in healthy synovial tissue, and that GlcNAc6ST-1 and 2 are induced in arthritic synovium at the transcript level for both enzymes and at the protein level for at least GlcNAc6ST-2.

PNAd-expressing vessels, which often feature HEV morphology, have been reported to occur at sites of chronic inflammation (reviewed in references [6] and [28]), where they are believed to facilitate L-selectin-mediated recruitment of naïve and central memory lymphocytes and possibly other leukocytes to the lesions. Significantly, the presence of PNAd has been shown in synovial vessels in human RA [23–26] as well as in a model of Lyme disease arthritis in SCID mice [27]; however, its role in disease processes is unknown. Other than correlative histological evidence, there has been one report implicating L-selectin in the development of arthritis in an animal model [39]. However, a recombinant P-selectin antagonist, which can also bind to L-selectin, was shown to provide protection in the mouse CIA model [40].

Towards the objective of elucidating novel targets for RA therapies, the present study examined the expression of PNAd, and the sulfotransferases, which are involved in its elaboration, in a model of CIA. This model has proven to be predictive for therapeutic benefit in human disease [32].

Using the sulfation-dependent specific antibody MECA-79 in IF, we found PNAd expressed on a large number of synovial vessels in arthritic joints at three evenly spaced time points during the 15 day model but never on synovial vessels from healthy control joints. Furthermore, most of the PNAd⁺-positive vessels demonstrated a punctuate staining pattern for GlcNAc6ST-2 protein, consistent with a Golgi-localization.

In addition, we showed by quantitative PCR that transcripts encoding GlcNAc6ST-2, as well as for the related sulfotransferase GlcNAc6ST-1, were significantly upregulated in rheumatoid synovium as compared to healthy control tissue at all time points examined.

Our data demonstrate for the first time that de novo expression of PNAd together with induction of the endothelial sulfotransferases GlcNAc6ST-1 and -2 are associated with disease in the collagen-induced arthritis model, an established animal model of rheumatoid arthritis widely used in preclinical development of disease-modifying anti-rheumatic drugs (DMARDs). While potential roles of PNAd, as well as GlcNAc6ST-2, in RA have been considered before [25], it remains unclear, whether de novo expression of MECA-79-reactive ligands is causal rather than corollary to the disease process in RA. Significantly, Salmi and colleagues observed MECA-79⁺ vessels in human arthritic joints; yet, because of the antibody's inability to significantly inhibit binding of activated mucosal immunoblasts to these vessels, the authors concluded that MECA-79-reactive ligands contributed little to disease-associated leukocyte recruitment [26]. In arthritic joints of SCID mice infected with B. burgdorferi, Schaible and coworkers observed the appearance of PNAd only late during infection following early upregulation of other adhesion molecules such as P-selectin, VCAM-1 and ICAM-1 [27]. Therefore these authors hypothesize that migration of leukocytes to inflamed joints in this model depends primarily on P-selectin-mediated rolling with subsequent engagement of α4β1/VCAM-1 and LFA-1/ICAM-1. However, we have recently demonstrated the pathophysiological significance of PNAd⁺ vessels in a sheep model of bronchial asthma. Intravenous injection of MECA-79 prior to allergen challenge substantially blocked increases in airway resistance and airway hyperresponsiveness in allergic animals [41]. Use of the MECA-79 antibody in mouse in vivo models would be hampered by the very poor half-life of this rat IgM in the mouse circulation (S. Hemmerich, unpublished observations). Nevertheless, the observed induction of PNAd in the mouse CIA model should motivate alternative approaches to investigate the pathophysiological role of PNAd in RA. Recently, mice have become available which have the GlcNAc6ST-1 gene, the GlcNAc6ST-2 gene, or both inactivated [18, 21, 22]. The double-null mouse exhibits a complete loss of PNAd from lymph node HEV and a marked but not complete reduction of L-selectin ligand activity. Breeding of this mouse line onto a CIA-susceptible background such as B10RIII or DBA/J [33], should permit investigation of the contribution of PNAd to disease. Alternatively, the availability of an IgG-based antibody with a specificity related to MECA-79 but having improved pharmacokinetics would allow an antibody intervention approach to CIA in the mouse or other species. We and others have recently embarked on discovery of GlcNAc6ST and other sulfotransferase inhibitors by high throughput screening and rational drug design (reviewed in references [30], [42] and [43]). As the CIA model has become the most widely used preclinical animal model for DMARD development [31], positive proof of concept achieved in the target-validation studies discussed above might motivate the development of GlcNAc6ST inhibitors for disease-modifying treatment of RA in humans.

MECA-79 binding glycoproteins defined as PNAd serve as ligands for L- selectin during the process of lymphocyte homing to lymph nodes. This class of ligands is also critical for inflammatory leukocyte trafficking and the associated disease, as recently demonstrated in a sheep model of asthma. Previous studies in mice have shown the requirement for the sulfotransferases GlcNAc6ST-1 and -2 in generating the MECA-79 epitope and the associated L-selectin ligand activity in HEV and HEV-like vessels. Immunohistochemical analyses have demonstrated ectopic expression of PNAd and GlcNAc6ST-2 in human rheumatoid arthritis. Further investigation into the functional role of PNAd and the requisite sulfotransferases in RA require a suitable animal model. Here we show, that PNAd is expressed de novo in synovial vessels in arthritic joints but not healthy control joints in the CIA model in mice. We further demonstrate that GlcNAc6ST-1 and GlcNAc6ST-2 transcripts are strongly upregulated in diseased tissue in this model and GlcNAc6ST-2 protein is present in a Golgi-associated pattern in the same vessels that show ectopic PNAd expression. Our data suggest the CIA model in mice for future studies into the functional role of PNAd-mediated leukocyte recruitment in RA and motivate the evaluation of PNAd antagonists or GlcNAc6ST inhibitors for therapeutic benefit in this widely used model of human RA.