¹Department of Microbiology and Immunology, College of Medicine, University of South Florida, Tampa, Florida, USA

Correspondence: Dr DL Lefkowitz, Department of Microbiology and Immunology, College of Medicine, University of South Florida, 12901 Bruce B Downs Blvd, Mdc 4017, Box 10, Tampa, FL 33612-4799, USA. Email: slefkowi@tampabay.rr.com

Received 16 January 2001; Accepted 5 April 2001.

Abstract

Macrophages have been described as 'factories' of pro-inflammatory cytokines. Several years ago the present investigators reported that binding of inactive myeloperoxidase (iMPO) to the macrophage-mannose receptor resulted in the induction of TNF and other cytokines. Also, if endothelial cells were incubated with iMPO, but not enzymatically active myeloperoxidase (MPO), upregulation of cytokine mRNA and cytokines was observed. Taken in their entirety, the data suggest a dichotomy of function for myeloperoxidase; that is, enzymatically active MPO functions primarily in cell killing through the 'cytotoxic triad' and iMPO functions as an immunoregulatory molecule through the induction of numerous cytokines. These studies underscore a previously unrecognized interaction among neutrophils, endothelial cells and macrophages, resulting in the induction of TNF and perpetuation of inflammation. The inflammation induced could be relevant in a number of diseases in which neutrophils play a prominent role. The importance of this interaction in the pathogenesis of rheumatoid arthritis is currently under investigation.

With the discovery of cytokines, immunological cross-talk among cells has become a shouting match. Today we know that these low molecular weight substances determine a myriad of cellular functions. It is now common knowledge that not enough or excessive amounts of cytokines are associated with many diseases resulting in morbidity and even mortality.

In 1995, we proposed a theory for the perpetuation of inflammation that involved the binding of neutrophilic myeloperoxidase (MPO) to the macrophage mannose receptor (MMR).¹ In the present paper, we revisit the paradigm and further expand the model to include the 'cross-talk' between MPO and eosinophil peroxidase (EPO). The former upregulates 'pro-inflammatory cytokines' whereas the latter appears to downregulate various 'pro-inflammatory cytokines'.² Also, because of the newly recognized role of the endothelium in various processes such as inflammation, the updated model includes the effect of either MPO or EPO on endothelial cell (EC) cytokine expression.³

It is estimated that 1–2% of the population in the USA are afflicted with rheumatoid arthritis (RA), with a female to male ratio of 3:1.⁴ Although the aetiology of RA is still unknown, it is readily accepted that there is a plethora of 'pro-inflammatory cytokines' associated with the pathology of this disease.^{5, 6, 7} In addition to various cytokines, there is a cellular infiltrate of macrophages (M), lymphocytes, dendritic cells (DC) and neutrophils into an arthritic joint.^{4, 8} Also, in 40% of patients with severe RA, eosinophilia may be present.⁴ Although investigators do not agree as to which one of the aforementioned cells is pivotal in RA, there seems to be a consensus that TNF- is the pivotal cytokine involved in RA.^{7, 8, 9, 10, 11} It has been reported that 'virtually every step of the inflammatory process could be TNF- dependent'.⁹ Part of the reason for this statement is that this cytokine is capable of initiating the 'pro-inflammatory cytokine cascade'.^{10, 11} However, even with the knowledge of the cells and cytokines that appear to be intimately involved in RA, this knowledge still does not address the actual reasons for the perpetuation of the inflammatory state.

This paper presents concepts, supported by research, which give more credence to our previous hypothesis. This hypothesis states that there is a previously unrecognized cell–cell interaction that is essential for the perpetuation of inflammation associated with various diseases including RA.

As stated earlier, in RA, the extending pannus of an inflamed joint is infiltrated by the M, T cell, B cell, and the neutrophil.⁴ Although many investigators have ascribed to the belief that T cells direct the inflammatory response of RA,^{12, 13} the failure of anti-T cell-based therapy⁹ is causing the focus of the disease process to turn away from this cell. The role of other cells of the pannus such as M, neutrophils, B lymphocytes and DC are once again being scrutinized with respect to their contribution to RA.

Dendritic cells have recently become a point of focus because these cells can serve as a link between the innate and the specific immune response.¹⁴ Their potential for presentation of self-antigen and their ability to activate certain T cells, without Th cell function,^{14, 15} could make DC vital players in RA. Another cell, the M, can also serve as a link between the two arms of the immune response; however, in addition to this function, M serve as a major source of 'pro-inflammatory cytokines'.⁶ Moreover, investigators have reported that the intensity of the M infiltrate into an RA joint correlates with the clinical activity of this disease.⁶ In addition, other investigators have reported that the neutrophil participates in joint destruction in RA.^{16, 17, 18, 19, 20} The neutrophil's importance is evident by the fact that this cell is present in abundance in the synovial fluid throughout all stages of RA.^{16, 17} Besides the numerous 'pro-inflammatory cytokines' neutrophils can secrete, this cell also ensures that there is a continuous supply of the enzyme MPO. When MPO is released from the neutrophil, the enzyme is rapidly inactivated in the microenvironment. Thus, at a site of inflammation one would have both enzymatically active MPO and enzymatically inactive MPO (iMPO) present.^{21, 22, 23}

Although there are numerous studies that have addressed the function of MPO as a participant in the cytotoxic triad, few studies have addressed the pro-inflammatory aspects of this enzyme once it is in the milieu of the microenvironment. Other investigators have reported that MPO is taken up by cells (e.g. M) by the MMR.²⁴ Binding of either MPO or iMPO (both highly mannosylated substances) to certain cell surface receptors induce the secretion of: (i) reactive oxygen intermediates (ROI)^{25, 26, 27} and (ii) TNF-, IL-1, IL-6, IL-8 and granulocyte–macrophage CSF (GM-CSF).^{3, 28} Reactive oxygen intermediates can cause direct tissue damage in addition to enhanced secretion of TNF-.^{29, 30}

Tumour necrosis factor- is, indeed, pivotal for the perpetuation of inflammation. This cytokine functions in an autocrine manner, resulting in the secretion of more TNF-. It also functions in a paracrine manner because it induces secretion of other cytokines from various cells. Moreover, TNF- upregulates cellular adhesion molecules (CAM).⁷ Granulocyte–macrophage CSF, another 'pro-inflammatory cytokine', also upregulates ICAM-1, which aids in neutrophil extravasation.³¹ The upregulated expression of CAM on the endothelium of the vasculature, as well as CAM expression on the neutrophil, are a prerequisite for extravasation of the neutrophil into an area of inflammation. Moreover, MPO is not a passive participant in the process of cell adhesion. It has been reported that MPO itself induces integrins that would aid in leucocyte extravasation.³²

Neutrophils, eosinophils and, to a lesser extent, T cells are drawn to an area of inflammation by M secreting IL-8, the potent neutrophil chemotactant.³³ Moreover, IL-8, TNF-, and GM-CSF secreted by the M after MPO binds to the MMR, induces neutrophils to degranulate and release MPO into the joint microenvironment (DL Lefkowitz and SS Lefkowitz, unpubl. data, 1999).^{34, 35, 36, 37} Once in the joint microenvironment, both MPO and iMPO will tend to remain in the joint. Studies by others have indicated that highly cationic molecules, such as MPO, do not easily leave the joint environment; that is, charge is important for MPO–cell interactions.^{38, 39}

The cytokine cascade initiated by TNF- induces the production of IL-6 which, in turn: (i) induces the acute phase response; and (ii) is used by B cells for maximum antibody secretion. Rheumatoid factors (usually IgM) are present in approximately 75% of all patients with RA.⁴ Therefore, the result of M–MPO interactions would be an enhancement of the production of rheumatoid factors. It is known that patients with RA express a form of IgG in which the terminal galactose is missing and N-acetylglucosamine is exposed.^{40, 41} N-acetylglucosamine binds with high affinity to the MMR and to the mannose-binding lectin (MBL).^{42, 43} Mannose-binding lectin ligand–receptor interaction with N-acetylglucosamine would result in the activation of the lectin complement pathway.⁴² Therefore, in addition to MMR–MPO or MMR–iMPO binding, the availability of yet another MMR high-affinity ligand would enhance the inflammatory process. It should also be noted that fucose, another high-affinity ligand of the MMR, is markedly increased in arthritic patients compared to control subjects.⁴⁴ Hence, this sacharride could also augment the inflammatory response via binding to the MMR. In summary, the common denominator is ligation of the MMR on the M cell surface, which results in cytokine secretion by these cells.

Experiments we have completed in the laboratory have shown that iMPO induces the secretion of higher concentrations of cytokines than enzymatically active MPO for both M and EC.^{3, 23} For several of these cytokines an RNA protection assay was done, which confirmed the fact that iMPO upregulated cytokine genes more than MPO.³ A possible explanation for this is as follows: iMPO induces a small increase in ROI, whereas MPO induces a much greater secretion of ROI.^{25, 26} The resultant effect would be that MPO could oxidize cell surface receptors⁴⁵ that could prevent cell signalling for enhanced cytokine secretion. Conversely, iMPO would bind to the MMR resulting in minimal production of ROI.²⁵ With low levels of ROI, receptors would not be oxidized, enabling enhanced cytokine secretion. Considering the above, one might think of a dichotomy of function for MPO and iMPO; that is, MPO functions in a cytotoxic triad resulting in cell death and iMPO functions as a pro-inflammatory molecule.

It has long been established that the body prefers a homeostatic state and, therefore, would pursue a path to downregulate the inflammatory process. From studies completed in the laboratory, it was observed that when enzymatically inactive EPO (iEPO) isolated from patients with eosinophilia was added to M cultures, the secretion of TNF-, IL-1 and IL-6 was inhibited.² Interestingly, iEPO did not affect the secretion of GM-CSF. The mRNA levels mimicked the secreted product profile. In terms of EPO, the intact enzyme had no effect on cytokine secretion.²

From all of the aforementioned, the following paradigm for the perpetuation of inflammation is proposed.

After an event that initiates inflammation, neutrophils are among the first cells to arrive at an inflamed site (Figure 1).⁴⁶ Once there, either through frustrated phagocytosis or phagocytosis of foreign substances, neutrophils degranulate releasing MPO into the microenvironment.²² Once in the microenvironment, MPO is rapidly inactivated.²² Both MPO and iMPO will bind to the MMR of the M.^{24, 25, 26} After MPO binds to the M, it induces a respiratory burst that can cause destruction of tissue as well as the secretion of TNF-.^{28, 29, 36, 37} Tumour necrosis factor- functions in an autocrine manner and, along with iMPO, initiates a cytokine cascade (IL-1, IL-6, IL-8, GM-CSF). With regard to the perpetuation of the inflammatory process, two cytokines are of particular importance: (i) IL-8 attracts more neutrophils; and (ii) IL-8 and GM-CSF cause neutrophils to degranulate and release MPO. With the release of additional MPO, the cycle starts once again.

Figure 1.

The cellular communication network during an inflammatory response.

Full figure and legend (131K)

At the same time, the endothelium, which is no longer considered a passive barrier for cells to migrate across, would produce radicals in response to iMPO but not MPO. Again, radicals of the respiratory burst could cause direct tissue damage in the inflamed joint and TNF- from the M would upregulate the expression of EC CAMS. Cellular adhesion molecules, as stated earlier, would be needed for neutrophils to emigrate from the blood into the tissues. Also, in response to iMPO, but not MPO, EC would secrete the 'pro- inflammatory cytokines' IL-6, IL-8 and GM-CSF.³ As for the M, cytokine gene mRNA levels mimicked the secreted product profile.³ Also, the upregulation of the monocyte-chemotactic protein-1 (MCP-1) gene was noted.³ Monocyte-chemotactic protein-1 would be chemotactic for monocytes, T cells and eosinophils.

As the inflammatory process continued, at some point the body would attempt to return to the pre-inflammatory state. Eosinophils would arrive at a site because of the chemotactic properties of IL-8. Once there, these cells would release EPO into the microenvironment. This enzyme could be inactivated by local proteases as well as other mechanisms. Binding of the iEPO to the MMR of the M would result in the diminished secretion of TNF-, IL-1 and IL-6 and, therefore, an anti-inflammatory response would occur. The implications of these various interactions are that the active and inactive form of two different peroxidases (MPO and EPO) could modulate an inflammatory response (Figure 1).

If our hypothesis is correct, then a substance that would prevent MMR–ligand interaction should prevent an arthritic flare. Mannans is known to bind to the MMR.⁴⁷ When we injected iMPO and mannans into the joints of rats primed for RA, the experimental arthritis flare was completely ablated.²³ In other studies, when rats were gavaged with mannans, there was a markedly diminished arthritic flare (D Kinney et al., unpubl. data, 2000). Again, these results implicate the M–MMR interaction with respect to the perpetuation of inflammation. Also, it is known that agalactosylated IgG can function as a ligand of the MMR. Investigators have reported that mannans prevents the uptake of agalactosyl IgG, which is a form of IgG that is present in patients with RA, by the M MMR.⁴¹

Finally, the interaction between the MMR and either MPO or iMPO may not be exclusive to RA. Tumour necrosis factor- appears to be involved in a number of other inflammatory diseases including Crohn's and multiple sclerosis.^{48, 49, 50} Because other investigators have reported that alveolar M ingested MPO in the lungs of animals with pulmonary inflammation,⁵¹ studies completed by us using alveolar M and either MPO or iMPO, have also indicated that iMPO induces more cytokine secretion than MPO. These findings support the concept of the participation of enzyme–M interaction in lung disease pathology.

Taken in its entirety, the aforementioned indicate that there is a previously unrecognized interaction between M and neutrophilic myeloperoxidase. In addition, there is an interaction between endothelial cells and myeloperoxidase. In both cases, iMPO, rather than MPO, induced both enhanced cytokine gene expression and cytokine secretion. This interaction can result in the perpetuation or chronicity of inflammation, such as that associated with RA.