Osteopontin (OPN)/SPP1: from its biochemistry to biological functions in the innate immune system and the central nervous system (CNS)

Abstract

Osteopontin (OPN) is a multifunctional protein, initially identified in osteosarcoma cells with its role of mediating osteoblast adhesion. Later studies revealed that OPN is associated with many inflammatory conditions caused by infections, allergic responses, autoimmunity and tissue damage. Many cell types in the peripheral immune system express OPN with various functions, which could be beneficial or detrimental. Also, more recent studies demonstrated that OPN is highly expressed in the central nervous system (CNS), particularly in microglia during CNS diseases and development. However, understanding of mechanisms underlying OPN’s functions in the CNS is still limited. In this review, we focus on peripheral myeloid cells and CNS-resident cells to discuss the expression and functions of OPN.

OPN in innate immune cells and CNS-resident cells

Graphical Abstract

Graphical Abstract.

Introduction

Osteopontin (OPN) is a pleiotropic protein encoded by the SPP1 (or Spp1 in rodents) gene and found in various cells, tissues, and fluids, identified as a versatile protein with multiple functions (1–3). Upregulation of Spp1 gene expression is frequently associated with inflammation elicited by events, including but not limited to infections (4–9), allergic responses (10–12), autoimmunity (13–16) and tissue damage (16, 17). OPN plays an important role in immune responses of different immune cell types, including both lymphocytes and myeloid cells.

Initially identified in osteosarcoma cells (18), OPN was reported to play a critical role in bone synthesis and resorption by mediating osteoblasts adhesion (19). Later, OPN was found as early T lymphocyte activation-1 (Eta-1) because activated T cells express copious OPN (20). Of note, the Spp1 gene is not transcribed in naive T cells unless hosts are aged (21, 22). However, macrophages constitutively express OPN (21). More recent studies identified OPN transcript in the central nervous system (CNS), particularly in microglia under pathogenic conditions and during brain development (23–25). A majority of functional studies on OPN has been performed with cells in the peripheral immune system, but the functions of OPN in the CNS have yet to be largely explored. Here, our review starts with the biochemistry of OPN and discusses the biological roles of OPN in peripheral innate immune cells and CNS-resident cells.

Gene and protein structures of OPN

The Human SPP1 gene locus locates in 4q22.1 and spans 7.7-kb with 7 exons. The full length of human OPN (hOPN) contains 314 aa residues. In contrast, the mouse Spp1 gene locates in chromosome 5 with the approximately 6-kb coding region including 8 exons. The full length of mouse OPN (mOPN) protein has 294 aa. Although human and mouse OPN peptide sequences only share 63% of homology, several structures and motifs in OPN are highly conserved across species (26). The molecular weight of hOPN is around 35 kDa, but hOPN protein migrates on SDS-PAGE gels from 40 to 80 kDa due to extensive posttranslational modifications (27, 28). The calculated size of mOPN is 32 kDa, but it also appears between 40 and 80 kDa on SDS-PAGE gels.

OPN has isoforms: alternative splicing isoforms and alternative translation isoforms. Alternative splicing isoforms were studied in hOPN because mOPN does not show splicing isoforms, at least in leukocytes (29). Splicing isoforms are termed OPNa, OPNb, and OPNc. OPNa contains Exons 2–7, including the full-length protein-coding regions. OPNb and OPNc lack Exon 5 and 4, respectively. All the human alternative splicing isoforms retain the signal sequence for protein secretion and other critical OPN domains and sites, such as the calcium-binding domain, heparin-binding domain, cleavage sites by thrombin and matrix metalloproteases (MMPs), and domains with amino acid sequences with SVVYGLR and RGD. These three isoforms have been intensively studied, mainly in cancer settings.

OPN also has alternative translation isoforms. Spp1 alternative translation initiation excludes the signal sequence and prevents OPN secretion, resulting in the generation of intracellular OPN (iOPN) (29). In contrast, the full Spp1 transcript, including the signal sequence, generates secreted OPN (sOPN). Indeed, both mouse and human OPN were identified in the cytoplasm and the nucleus of cells (29–35), although only a limited number of human studies evaluated the impact of the balance between iOPN and sOPN on human diseases (35–37). sOPN and iOPN are considered to have distinct phosphorylation and glycosylation status because of their differential subcellular localization.

Conserved domains and motifs

Both human and mouse OPN contains the classical RGD motif that binds αvβ1, αvβ3, αvβ5, αvβ6, α5β1, and α8β1 integrins (38–40). An additional integrin-binding motif is SVVYGLR¹⁶⁸ for hOPN or SLAYGLR¹⁵³ for mOPN. The 7-aa motifs are located immediately downstream of RGD and are called OPN ‘cryptic epitopes’ because of their exposure upon thrombin cleavage between R¹⁶⁸ and S¹⁶⁹ residues in hOPN and between R¹⁵³ and S¹⁵⁴ in mOPN (39, 41). In addition to thrombin, OPN is also cleaved by MMP-3 and MMP-7 between the G¹⁶⁶ and L¹⁶⁷ residues in hOPN and between G¹⁵¹ and L¹⁵² in mOPN (42–47), which are within the SVVYGLR and SLAYGLR cryptic epitopes. MMP cleavage disrupts the cryptic epitopes and prevents α5β1 binding to RGD, although RGD-binding αv integrin recognition is not affected (41). Heparin-binding domain (HBD) YGLRSKSKKF in hOPN or YGLRSKSRSF in mOPN overlaps the SVVYGLR or SLAYGLR cryptic epitopes, respectively. There are other biologically functional domains in OPN, such as the calcium-binding domain. These are summarized in excellent reviews (48, 49).

Enzymes that act on OPN

We mentioned above that OPN is a substrate of thrombin and MMP-3/7. Studies also showed that OPN is a substrate of other enzymes. One of those is transglutaminase 2, which covalently crosslinks OPN residues to form an OPN polymer (polyOPN). The OPN conformation change by polymerization makes polyOPN bind integrins, especially α₉β₁, through the exposure of the cryptic SVVYGLR epitope (50–52). Another enzyme, phosphate-regulating endopeptidase homolog X-linked (PHEX), digests OPN (53). Accumulation of OPN and other SIBLING proteins is caused by loss-of-function mutation of PHEX, resulting in the genetic disease X-linked hypophosphatemia (54, 55).

Posttranslational modifications of OPN

A comprehensive study of posttranslational modifications identified 34 phospho-serines, 2 phospho-threonines, 5 O-glycosylated threonines and no apparent N-glycosylation in milk hOPN (28), although different cell types generate various degrees of OPN phosphorylation (56). Another study using milk hOPN showed that O-glycosylation at the 5 threonine residues protects the integrin binding motifs from pepsin digestion, thus retaining the ability to mediate cell adhesion after pepsin digestion (57). For OPN phosphorylation, Golgi-casein kinases, such as FAM20C and VLK, are involved (58). Phosphorylation and glycosylation can regulate the OPN-integrin interactions if the modification sites are within or close to integrin binding motifs. For example, serine phosphorylation in SVVYGLR significantly reduces the α_vβ₃ integrin binding affinity to RGD (58, 59). These studies suggested that posttranslational modifications of OPN can alter the functions of OPN.

Receptors detecting sOPN

As we mentioned, various integrins bind the RGD motif of sOPN. The OPN cryptic epitope is recognized by α4β1 and α4β7, while α9β1 recognizes the cryptic epitope only after thrombin cleavage (40). The second type of sOPN receptor is the splicing variants of CD44 containing exon v6 or v7 (CD44v6 or CD44v7) (60–62). The binding of CD44v6 or CD44v7 to OPN requires the presence of β1 integrin but is RGD-independent. Thrombin cleaved OPN N-terminal and C-terminal fragments have similar affinity to the CD44v6 or CD44v7, suggesting multiple epitopes are involved in the sOPN-CD44 interaction (61). Recently, a cell surface receptor, inducible T-cell costimulator ligand (ICOSL), was also discovered to bind sOPN in several types of tumor cells (63). The OPN-ICOSL interaction promotes tumor metastasis and angiogenesis (64).

Involvement of iOPN in signal transduction in the cytoplasm

As an intracellular molecule, iOPN is physically segregated from all the cell surface receptors, exerting distinctive functions from those of sOPN. Previous studies demonstrated that iOPN is an adaptor protein, without enzymatic functions, in cytoplasmic signal transduction (9, 32, 34, 65–67). The first report of such a function of iOPN indicated that iOPN promotes IRF7 activation to induce type-1 interferon (IFN-I) gene expression downstream of TLR7 and TLR9 in plasmacytoid dendritic cells (pDCs) (9). iOPN also associates with MyD88, TRAF2, IRAK1 and Syk, in mononuclear phagocytes (MNPs) and controls TLR4 and CD40 dual signaling, as well as ‘tethering’ multiple cell surface receptors (TLR2, dectin-1, mannose receptor) to cluster receptors to synergize signal transduction triggered by the receptors (32, 33). Another study also suggested that tumor necrosis factor (TNF) receptor-associated factor 3 (TRAF3) associates with iOPN through an OPN C-terminal region in macrophages to stabilize TRAF3, promoting anti-viral responses through upregulating IFNβ production (68). iOPN also interacts with vimentin, a structural protein of intermediate filaments and stabilizes vimentin by extending the vimentin half-life. Increased vimentin is considered a hallmark of cancer metastasis (69).

iOPN in the nucleus

iOPN also localizes in the nucleus. An earlier study identified nuclear iOPN in a human cell line, HEK294, and demonstrated nuclear translocation of iOPN correlating to chromatin condensation during cell cycles (35). Another study on hypoxia/reoxygenation showed that iOPN is cleaved by caspase-9, and its C-terminal fragment translocates into the nucleus, leading to upregulating the transcriptional activity of a pro-apoptotic p53 molecule and apoptosis of HeLa cells (70). In mouse T follicular helper (Tfh) cells, a molecule p85α chaperones iOPN for its nuclear entry (67). Nuclear iOPN promotes functional Tfh differentiation by interacting and stabilizing Bcl-6 (67). Another study showed that nuclear iOPN in human cell lines interacts with HIF2α to eventually impact lung metastasis in patients (36).

OPN in peripheral innate immunity

Currently, many articles describing OPN expression by all kinds of leukocytes are available, suggesting the critical involvement of OPN in various immune responses (Fig. 1 focuses on myeloid cells and CNS-resident cells only). OPN was studied in autoimmune diseases [such as systemic lupus erythematosus (SLE), rheumatoid arthritis (RA) and multiple sclerosis (MS)], various types of infections, cancer and other pathological conditions. OPN in innate immunity is critically involved in pathogenesis and health.

Fig. 1.

OPN expression and its outcomes in various cell types. OPN receptors, integrins and CD44, are commonly expressed on the surface of cell types described here. sOPN has been extensively studied, particularly in peripheral immunity, and generally promotes cell migration and effector functions, such as opsonization and cell proliferation. iOPN has been studied in the immune system but not in the CNS-resident cells. For some cells, mainly CNS-resident cells other than microglia, OPN expression was reported but not to the extent of sOPN and/or iOPN. For astrocytes and neurons, we interpreted that sOPN and/or iOPN may be produced based on data from published articles (indicated as ‘sOPN?’ or ‘iOPN?’). Oligodendrocytes and OPC detect sOPN, but the source of sOPN is unclear in vivo. Degenerated neurons are considered to be opsonized by sOPN for phagocytosis by microglia.

Macrophages/monocytes

Among cell types in the innate immune system, OPN may have been most intensively studied in macrophages. Many early studies already showed that sOPN plays critical and pleiotropic roles in mediating macrophage functions, including cell migration, proliferation, phagocytosis and MMP production (71–75). Furthermore, exposure of monocytes to sOPN promotes their differentiation to macrophages through a sOPN-integrin interaction (76). There are many studies of sOPN and macrophages, and we would like to leave further discussions about sOPN and macrophages to outstanding review articles (77–79).

iOPN has also been reported to be involved in immune regulation. iOPN supports receptor clustering of dectin-1, TLR2 and mannose receptor on the surface of macrophages to detect a fungus, Pneumocystis murina, and enhance collaboration of dectin-1/TLR2 signaling pathways to elicit strong host cell responses (33). In peritoneal macrophages, iOPN also participates in the innate anti-viral response by inhibiting ubiquitination and degradation of TRAF3 (68). iOPN in macrophages also negatively controls TLR4 signaling through CD40 ligated by CD40L on T cells (32). Here, iOPN plays a role as an adaptor molecule to couple IRAK1 and TRAF2 to promote IRAK1 sumoylation (32). The outcome of the iOPN’s function during CD40L/TLR4 co-stimulation is promoting IL-10 expression and downregulation of TNFα to protect hosts from LPS endotoxemia (32).

Dendritic cells

One of the initial studies on OPN and dendritic cells (DCs) demonstrated that sOPN enhances the production of TNFα and IL-12 by human DCs (80), resulting in enhanced stimulation of allogeneic T cells and Th1 polarization (80). Another study showed that sOPN secreted by CD103⁻ DCs mediated exacerbation of TNBS-induced experimental colitis by promoting Th17 and Th1 polarization in mesenteric lymph nodes (81). The study also demonstrated that the SLAYGLR motif of sOPN ligates α9 integrins to elicit the pathogenic role of sOPN in colitis (81). iOPN expressed in conventional DCs (cDCs) promotes Th17 responses in a model of MS by suppressing IL-27 expression, suggesting the pro-inflammatory role of sOPN (14). Yet, iOPN expression in cDCs is negatively controlled by type-1 interferon receptor (IFNAR) (14). Here, another subset of DCs, pDCs, potently produce IFN-I; and iOPN promotes the IFN-I production downstream of TLR9 by interacting with MyD88 to promote IRF7 activation (9). Thus, iOPN and IFN-I are mutually controlling their expression in cDCs and pDCs. A different study on cancer showed that a hypoxic environment within tumors promotes OPN expression by DCs, and DC-derived sOPN promotes tumor cell migration in a tissue culture setting (82). OPN also modulates a cross-talk between DCs and mesenchymal stem cells (MSCs) (83). Resting MSCs promote OPN expression by DCs in their co-culture, but activated MSCs suppress OPN production by DCs in the presence of pro-inflammatory cytokines (IL-1β, IL-6 and TNFα) (83). This suggests the DC-derived sOPN production is modulated by cytokine milieu. Many other studies demonstrated DCs as an OPN producer and an OPN target, and we leave further discussion on this another excellent review article (84) focusing on that aspect.

Neutrophils

Neutrophils are the most abundant blood-circulating leukocytes, and their destructive function against pathogenic microbes is occasionally detrimental to host cells. Mouse neutrophils express OPN, but the expression was 75-fold less than the RAW264.7 macrophage cell line and 25-fold less than peritoneal macrophages (85), suggesting that neutrophils may not be as potent OPN producers as macrophages. Instead, sOPN is detected by neutrophils and enhances neutrophil migration in a CD44-independent fashion (86). Yet, OPN-deficient neutrophils are not impaired with phagocytosis, generation of reactive oxygen species, MMP-9 production and cytokine production, such as IFN-γ, TNF-α, IL-1β and IL-4 (85). Interestingly, OPN-mediated enhancement of neutrophil migration could favor pathogen dissemination. For example, OPN promotes the migration of neutrophils, infected by West Nile virus (WNV), into the CNS and causes encephalitis and meningitis (87). Indeed, Spp1^−/− mice showed an improved survival rate after WNV infection, and the WNV viral burden in the brain was significantly decreased (87). Thus, sOPN could also promote host-detrimental roles of neutrophils in microbial infections.

Eosinophils

Eosinophils express OPN. Studies indicated that OPN and eosinophils are a detrimental combination in type-2 diseases. Resting eosinophils secrete OPN, which promotes airway angiogenesis in asthma (88). Another study also suggested the association of OPN with severe chronic rhinosinusitis (89). A mouse lung fibrosis model using house dust mite indicated OPN expression triggered airway fibrosis through Th2 responses (90). In patients with asthma, a positive correlation between the sputum eosinophil frequency and the sputum sOPN levels was found (91), suggesting that sOPN in the sputum might be a diagnostic marker. Antibody neutralization of sOPN limited ex vivo migration of human eosinophils by blocking OPN’s interactions with integrin α4 and attenuated eosinophilic inflammation in a mouse asthma model (91). Thus, eosinophils can be an sOPN source and responders to OPN with detrimental impacts on pulmonary diseases. The impact of iOPN in eosinophils is not known yet.

Mast cells

Mast cells are found in mucosal sites, including the gastrointestinal tract, the lung epithelium, and other vascularized mucosal tissues. Mast cells carry large granules containing inflammatory mediators, such as histamine, heparin, proteases and pro-inflammatory cytokines (92). A study demonstrated that a basal level of OPN is produced by murine fetal skin-derived cultured mast cells (FSMCs) and enhances IgE-mediated degranulation in FSMCs (93). In addition, migration of FSMCs is also enhanced by sOPN by its interaction with αv integrins (93). Another study, using IgE-activated human cultured mast cells (HCMCs) derived from peripheral blood CD34⁺ progenitors, showed that sOPN enhanced HCMC adhesion and suppressed cytokine productions such as IL-5, IL-8 and TNFα (94). Therefore, CD44 and integrins on mast cells are expected to mediate the impacts of sOPN on mast cell function and migration, but the roles of iOPN are still unclear.

OPN in the CNS

Recent single-cell RNA sequence studies indicated upregulation of Spp1 mRNA, as well as OPN expression, during CNS development and various CNS disease models (24, 25, 95), which indicates the roles of OPN in CNS development, pathology, and health. However, OPN functions in the CNS are still largely unexplored. In this section, we will focus on the expression and known functions of OPN in CNS-resident cells (Fig. 1).

Microglia

Microglia may be the most intensively studied cell type of CNS-resident cells regarding OPN. Microglia are macrophage-like cells derived from primitive macrophage progenitors in the yolk sac at the early stage of embryonic development (96). Upregulation of Spp1 and its product, OPN, was observed in microglia in both humans and mice with neurodegenerative diseases, such as AD and MS, and in a cuprizone-induced demyelination animal model (24, 95, 97). Elevated Spp1 expression in microglia was also identified in the early postnatal mouse brain, in which microglia are crucial for regulating the growth and pruning of synapses under a homeostatic condition (25, 98, 99). Spp1 upregulation in microglia also appears to be age-related, on the basis of the increase of Spp1 transcripts in microglia from human individuals over 50 years old (100). Of note, age-related Spp1 expression was previously reported in T cells too (22). Studies showed that microglial Spp1 expression is upregulated in microglia subsets (98, 100), which are described as disease-associated microglia (DAM) in AD (24), microglial neurodegenerative (MGnD)-neurodegenerative microglia highly expressing Apoe (95), CD11c⁺ microglia providing survival factors to support myelin development (101) and the axonal tract-associated and proliferative region-associate microglia (ATM and PAM, respectively) that appear in the early postnatal development (25, 99). These microglia subsets described in different studies appear to have some overlap in gene expression patterns and possibly in their functions.

Microglia express OPN receptors, such as CD44 and integrins; thus, an autocrine impact of sOPN is likely. A study on ischemic brain injury demonstrated detailed imaging of Spp1 mRNA and OPN protein in the brain, showing clear upregulation of Spp1 mRNA expression in Iba-1⁺ cells (i.e. macrophages and/or microglia) 3 days after ischemia induction at latest and lasting until the last day of evaluation on 28 days (102). OPN protein was detected in the extracellular matrix (ECM) of ischemic cores, but cytoplasmic OPN was apparent on day 28 (102), suggesting that sOPN is released at the early stage and iOPN starts to accumulate in the cytoplasm later. The study also suggested by using electron microscopy that sOPN accumulates at the surface of degenerated neurons, and microglia and/or macrophages engulf fragments of degenerated neurons (102). Indeed, OPN’s role as an opsonin for microglia was suggested (103). Recombinant OPN (rOPN; i.e. sOPN) promotes the proliferation of porcine microglia ex vivo, and sOPN neutralization using antibody abolishes the proliferation (104), suggesting sOPN’s role in microglia proliferation. However, rOPN treatment of LPS-stimulated mouse primary microglia does not enhance microglia proliferation; instead, cells treated with rOPN show an anti-inflammatory phenotype with reduced iNOS expression and decreased pro-inflammatory cytokine expression, such as IL-6 and TNFα (105). These results suggest that the role of OPN on microglia is complex, and the outcomes appear to be context-dependent. In fact, the complexity of OPN-mediated outcomes is not specific to microglia, but other immune cells exhibit the multi-faceted roles of OPN in immunological responses. To clarify the complexity, dissecting the differential role of OPN isoforms may be an important next step to be elucidated.

Astrocytes

Primary astrocyte culture showed expression of OPN (106). OPN expression in astrocytes was also detected in demyelinated brain areas in a cuprizone demyelination model (97). Elevated OPN expression was also demonstrated in intratumoral astrocytes (107). Fluorescent microscope images from these studies suggested that OPN is localized in the cytoplasm because of its colocalization with glial fibrillary acidic protein (GFAP) (97, 107), but this point is not discussed in the articles. rOPN treatment of astrocytes in tissue culture showed enhanced proliferation and survival particularly with pre-treatment with pro-inflammatory cytokines (TNFα, IL-1β and IFN-γ), although the impact of sOPN on proliferation and survival was modest without the cytokine pre-treatment, despite a significant increase in the astrocyte migration (108). OPN-deficient astrocytes reduced expression of GFAP, when stimulated with LPS ex vivo and OPN-deficient mice had reduced reactive astrocytes (106), suggesting the involvement of OPN in promoting astrogliosis. OPN-deficient mice with ischemic stroke demonstrated that astrocytes failed to extend their processes (109). The same study also showed that sOPN promoted the re-establishment of the blood-brain barrier (BBB) after ischemic stroke. The study concluded that the source of OPN was macrophages using a clodronate liposome approach (109). (Here, cell depletion with clodronate liposome is not specific; thus, identification of OPN producers requires more precise approaches, such as mouse genetic models.) Yet, these studies strongly suggested the impacts of sOPN on astrocytes. It is still unknown which mechanism enhances OPN expression in astrocytes, which sOPN receptors on astrocytes exert the impacts on astrocytes, and if iOPN also plays some roles in astrocytes.

Neurons

Although microglia and astrocytes can be sources of OPN in the CNS, other CNS resident cells, such as neurons, particularly motor neurons (MNs), are also capable of expressing OPN (110). OPN is detected in the cytoplasm (i.e. iOPN) in a majority of alpha MNs in the spinal cord of SOD^G93A amyotrophic lateral sclerosis (ALS) model mice and also in extracellular spaces as granular deposits (i.e. sOPN) (103). sOPN accumulated in ECM is detected by the αvβ3 integrin on MNs and upregulates MMP-9 expression by the cells, resulting in second-wave neurodegeneration in the ALS mouse model (103). Another study using a rat acute striatal injury model suggested that OPN granular deposits are from degenerated neurites and are engulfed by astrocytes (and microglia) in astroglial scar (111). In contrast, a study using macaques showed OPN localization in the neuron cell bodies but neurons may not secrete detectable levels of OPN (112).

Oligodendrocyte and oligodendrocyte precursor cells

In a cuprizone demyelination model, OPN was not identified in mature oligodendrocytes and oligodendrocyte precursor cells (OPCs), although OPN is produced by microglia and astrocytes in the corpus callosum in the model (97). Yet, rOPN induced cell proliferation of OPC-like cell lines ex vivo, and stimulated myelin basic protein (MBP) synthesis and myelin sheath formation (97). Thus, OPN may protect myelin, and the OPN-mediated outcomes may involve oligodendrocytes.

Conclusions

Because OPN and its transcript (Spp1) are often highly expressed and detected in many pathogenic conditions, OPN has been generally considered to be a pro-inflammatory and pathogenic molecule. However, recent studies indicate that OPN is multi-faceted. Indeed, multiple studies using OPN knockout mice showed that OPN could be pro- and anti-inflammatory, as well as detrimental and beneficial even in similar disease settings (74). As discussed in this review, OPN has diverse functions not only because of its involvement in numerous biological settings but also because of its various forms as mature proteins. For example, alternative intron splicing and translation initiation generate multiple isoforms. OPN also gains functions when cleaved by various proteases. In addition, various levels of posttranslational modifications of OPN in different cell types contribute to altering OPN’s function. Therefore, it is essential to understand how and when the impact of OPN becomes host beneficial or harmful by considering various forms of OPN. This aspect is critically important, especially because OPN is a potential pharmaceutical target in many diseases.

Functions of OPN in the CNS have not been largely explored. In addition to considering various forms of OPN to understand its function, it is critical to identify specific roles of OPN to various CNS cell types.

Funding

This work was supported by National Institutes of Health (R01-AI088100, R01-NS120417, and R01-AI160737) to M.L.S.

Author contributions

E.Y.-H.L., W.X. and M.L.S. wrote the manuscript, and N.A. edited the manuscript.

Conflicts of interest statement: the authors declare no conflicts of interest.