|Home | About | Journals | Submit | Contact Us | Français|
B lymphocyte hyperactivation and hypergammaglobulinemia are pathogenic manifestations of HIV-1 infection. Here we provide evidence that these B cell defects are driven by factors produced by HIV-1 infected macrophages and that Nef is necessary for this activity. In vitro, HIV-1 infected macrophages or macrophages expressing Nef promoted B cell activation and differentiation to immunoglobulin-secreting cells. Activation of NF-κB by Nef induced secretion of the acute-phase reactant ferritin and ferritin was necessary and sufficient for these B cell effects. The extent of hypergammaglobulinemia in HIV-1 infected individuals correlated directly with plasma ferritin levels and with viral load. We further demonstrate that induction of ferritin and hypergammaglobulinemia could be recapitulated when Nef was specifically expressed in macrophages and T cells of transgenic mice. Collectively, these results reveal the presence of a pathogenic determinant within the Nef protein of HIV-1 which governs B cell defects in HIV-1 infection.
HIV infection causes a dramatic impairment of T cell-mediated immunity as a consequence of the rapid turnover of CD4+ T lymphocytes (Fauci, 1993). The humoral arm of the immune system frequently sustains further damage and B cells exhibit profound functional deficiencies (Jacobson et al., 1991; Lane et al., 1983). In HIV infected patients, B lymphocyte responses to T cell-independent and common recall antigens are diminished (De Milito et al., 2004; Moir et al., 2001) and cells exhibit increased surface expression of CD70 (De Milito et al., 2004) and CD71 (Martinez-Maza et al., 1987) indicative of hyperactivation. B cells also demonstrate a more differentiated phenotype as evidenced by higher levels of CD38 (Conge et al., 1998) and the sustained expression of CD70 (De Milito, 2004; Martinez-Maza et al., 1987), which is accompanied by spontaneous immunoglobulin secretion in vitro (Amadori et al., 1989; Moir et al., 2001) and hypergammaglobulinemia in vivo (Abelian et al., 2004; De Milito, 2004; Lane et al., 1983). Infected individuals have levels of plasma immunoglobulin up to twice that of normal levels (Lane et al., 1983; Lucey et al., 1992) which is primarily a result of polyclonal activation since plasma immunoglobulin concentrations remain high, despite a decline in HIV-specific responses as disease progress (Amadori et al., 1989; Shirai et al., 1992).
Several co-factors have been proposed as mediators of B cell dysfunction in HIV-1 infection. HIV-1 virions or virion proteins, such as the viral envelope glycoprotein gp120, have been shown to impair B cell function in vitro (Pahwa et al., 1986; Patke and Shearer, 2000). Cytokines induced as a result of perturbed signaling during HIV infection also impact B cell activity in culture (De Milito, 2004). An array of principally pro-inflammatory mediators, TNFα IL-6, IL-10 and IL-15 can account in part for some aspects of B cell dysfunction, however, evidence that these factors promote all facets of HIV-dependent B cell dysfunction in vitro is incomplete and their correlation to patient data is poor (De Milito, 2004; Kacani et al., 1997; Muller et al., 1998; Patke and Shearer, 2000; Rautonen et al., 1991). Cytokine production is significantly modified by the direct infection of permissive target cells with HIV. For example, in macrophages, viral infection induces inflammatory chemokines and release of soluble immunoregulatory surface antigens (Swingler et al., 2003; Swingler et al., 1999). Chemokine production is regulated by the accessory protein Nef and involves signaling through the NF-κB cascade (Swingler et al., 2003).
Here, we present evidence that HIV-1 Nef, through activation of NF-κB, induces the secretion of ferritin from infected macrophages and that this causes B cell activation and hypergammaglobulinemia. Our results reveal the presence of an inherent pathogenicity determinant in HIV-1 Nef that is responsible for B cell defects that are a hallmark of HIV-1 infection.
We previously demonstrated that soluble factors induced from infected macrophage by Nef facilitated the ability of resting CD4+ T-lymphocytes to support HIV-1 replication (Swingler et al., 2003). We initiated the current study when we observed that peripheral blood lymphocytes (PBLs) proliferated when exposed to supernatants from macrophages expressing HIV-1 Nef (Figure 1A). Nef-expressing macrophage supernatants selectively induced the proliferation of B cells but not of T cells (Figure 1A) whilst supernatants from GFP-expressing macrophages or mock-infected macrophages did not cause the proliferation of either lymphocyte subset (Figure 1A). B lymphocytes undergo a pattern of differentiation in the transition from a resting state to antibody producing plasma cells. This is characterized by the time-dependent up-regulation of activation molecules, such as CD38, and ultimately the expression of Syndecan-1 (CD138) which is a proteoglycan indicative of plasma cells (Jego et al., 1999). Following exposure of B cells to supernatants from Nef-expressing macrophages, but not macrophages expressing a control protein (GFP) or untreated macrophages (mock), there was an increase in the percentage of B cells expressing CD38 and CD70 (Figure 1B). These effects were also elicited by Nef when expressed in the context of HIV-1 infection (Figure 1C). Thus, supernatants of wild-type HIV-1-infected macrophages, but not ΔNef HIV-1-infected macrophages obtained at the peak of viral replication, caused a similar change in activation in differentiation status as evidenced by CD38 expression (Figure 1C). FACS analysis of CD38+-gated cells for sIgM and CD138 expression demonstrated that B cells exposed to Nef expressing macrophages or wild-type HIV-1 infected macrophages exhibited increased surface IgM and CD138 expression (Figure 1D). The B cell effects promoted by Nef-expressing macrophage supernatants or by supernatants from wild type HIV-1 infected macrophages were comparable in magnitude to those induced by a classical B cell differentiation stimulus (αIgD/IL-4/rCD40L) (Figure 1D). The secretion of immunoglobulin subtypes, IgA, IgG, and IgM was also induced relative to supernatants from macrophages expressing a control protein (GFP) [Figure 1E] or macrophages infected with a Nef-deleted virus (HIV-1ΔNef) and mock-infected macrophages (Figure 1F). Collectively, these results suggest that Nef expression, or Nef in the context of HIV-1 replication in macrophages, leads to the release of a soluble factor(s) which induces hyperactivation of resting B cells and differentiation to antibody-secreting cells.
In order to characterize the factors responsible for the B cell effects, macrophages were cultured in serum-free medium and infected with a GFP or Nef-expressing adenovirus vector. Supernatants were harvested, passed through a series of defined molecular weight cut-off filters and assayed for ability to induce B cell proliferation. This activity was specific to Nef-expressing macrophage supernatants and was retained by 100kD molecular weight cut-off filters (Figure 2A). In supernatants of Nef-expressing macrophages relative to GFP-expressing macrophages the molecular weight of the B cell stimulatory activity was more precisely estimated by separation of concentrated supernatants under non-denaturing polyacrylamide gel electrophoresis and analyses of factors eluted from gel slices (data not shown). When proteins were eluded from those regions and assayed for the ability to induce B cell proliferation, the majority of the proliferative activity resided in the region around 250K. When a replicate gel was prepared under denaturing conditions proteins in this region, several Nef-specific bands were evident in supernatants of Nef-expressing macrophages relative to GFP-expressing macrophages and these were subject to peptide microsequencing and multiple peptides from this high molecular weight species corresponded exclusively to human ferritin, human ferritin light chain and human ferritin heavy chain (Figure 2B); accession numbers 0901237A, NP000137, PO2792, and AAH69538). The identity of ferritin in macrophage supernatants was confirmed by the use of reducing gels and Western blotting and co-migration with recombinant heavy and light forms of ferritin (Figure 2B lower panel). Macrophages expressing Nef produced elevated levels of ferritin relative to GFP-expressing macrophages or mock-infected macrophages (Figure 2C) and there was a statistical correlation between the presence of Nef and ferritin in eight replicate experiments (p<0.0001) [Figure 2D]. We also examined ferritin induction in macrophages supporting a spreading infection. We have previously demonstrated that wild type and Nef-deleted variants of HIV-1 replicate to similar extents in macrophages (Swingler et al., 1999) and here it was also evident that Nef was required for the induction of ferritin (Figure 2E). In macrophages from five different donors, infection with macrophage-tropic HIV-1 (HIV-1ADA) wild type but not Nef-deleted HIV-1 ADA, was necessary for the release of ferritin (p=0.031) [Figure 2F]. Although ferritin was induced by HIV-1WT infected macrophages, there was no induction of ferritin from HIV-1-infected T lymphocytes purified from peripheral blood (Figure 2G). Therefore, the induction of ferritin by HIV-1 was manifest specifically in infected macrophages.
We have previously demonstrated that Nef regulates chemokine production (MIP-1α and MIP-1β) by macrophages and that this induction is NF-κB –dependent (Swingler et al., 2003). Consistent with this, mRNAs for IκBα and NF-κB-p105, the precursor of the NF-κB p50 subunit and cytoplasmic anchor, were induced in wild type HIV-1 infected macrophages relative to HIV-1ΔNEF or mock-infected macrophages (Figure 3A and 3B) [also examined at the peak of HIV-1 replication]. In addition, when HIV-1-infected macrophages were examined at the peak of HIV-1 replication (Figure 3C), IκBα was phosphorylated on a specific regulatory serine residue (S32) demonstrating NF-κB activation in HIV-1WT infected macrophages relative to HIV-1 ΔNef-infected or mock-infected macrophages (Figure 3D). These effects were recapitulated in macrophages expressing Nef. Therefore, there was increased turnover of mRNAs encoding IκBα and NF-κB-p105 in macrophages expressing Nef consistent with activation of NF-κB (Figure 3E). This was further evidenced in immunoprecipitation/in vitro kinase assays of upstream IKKα/β using recombinant IκBα as substrate, Nef induced a strong activation of the IKK kinases in line with that seen with TNFα treatment (Figure 2F). To determine whether the induction of ferritin was NF-κB dependent, macrophages were co-infected with adenoviruses expressing Nef and a dominant negative form of IκBα (IκBαSR) (Brown et al., 1995). Since the efficiency of adenoviral transduction was equivalent when macrophages were infected with the adenovirus vectors singly or in combination (data not shown) co-infection experiments were performed. In the presence of the IκBαSR suppressor, the induction of ferritin by Nef was reduced to background (mock) levels (Figure 3G) whereas ferritin production was unaffected by co-infection with an adenovirus vector expressing LacZ (Figure 3G). The ability of the IκBαSR to inhibit Nef-mediated ferritin induction was not due to effects of this suppressor on Nef expression since levels of Nef were similar in the presence and absence of the IκBαSR (Figure 3H). In addition, the IκBαSR impaired CD40-dependent chemokine induction in primary macrophages (Figure 3I) consistent with our previous observations (Swingler et al., 2003). The ability of the IκBαSR to almost completely negate the ability of Nef to induce ferritin suggests a principal role for the canonical NFκB pathway in the induction of ferritin by Nef. Proinflammatory signals have been shown to directly regulate ferritin at the transcriptional level. Likewise, both Nef expression and TNFα substantially increased the level of mRNA for both ferritin light chain and ferritin heavy chain (Figure 2J). These results suggest that Nef exerts a direct effect on ferritin transcription more similar to a proinflammatory stimuli than the enhancement of mRNA translation mediated by Iron Regulatory Proteins (Fahmy and Young, 1993; Pham et al., 2004).
To determine whether the effects of Nef on B cell function were mediated by ferritin, we examined the impact of ferritin immunodepletion on the induction of B cell proliferation and immunoglobulin secretion by Nef-expressing macrophage supernatants. Ferritin was quantitatively removed by immunodepletion with a ferritin-specific antibody but not with an isotype antibody (Figure 4A). While Nef-expressing macrophage supernatants immunodepleted with an isotype antibody induced B cell proliferation, when ferritin was immunodepleted, B cell proliferation was reduced to background (GFP-expressing) levels (Figure 4A). Similarly, supernatants of Nef-expressing macrophages did not induce immunoglobulin secretion from B cells when ferritin was removed from these supernatants (Figure 4B). Previous studies have indicated that recombinant Nef protein can induce the release of inflammatory factors from macrophages and suppress class switching in B cells (Mangino et al., 2007; Olivetta et al., 2003; Qiao et al., 2006). Therefore, one possibility is that extracellular Nef released from transduced macrophages were mediating the effects of these macrophages supernatants. However, immunodepletion of Nef from supernatants of Nef-transduced macrophages had no effect on the ability of these supernatants to induce B cell proliferation (Figure 4C). Since the effects of recombinant Nef were observed at high concentrations of the protein, it was possible that supernatants of Nef-transduced macrophages contain insufficient quantities of Nef to effect B cell function. However, recombinant Nef at concentrations up to 1,000 ng.ml−1 did not induce B cell proliferation (Figure 4D). This suggests that the induction of B cell proliferation and immunoglobulin release by Nef-expressing macrophage supernatants requires virus infection and de novo expression of Nef.
To determine if B cell differentiation and immunoglobulin release could be recapitulated by ferritin alone, purified human liver ferritin (PF), consisting of broadly equivalent amounts of heavy and light chains (data not shown), recombinant light chain ferritin (FTL) and recombinant heavy chain ferritin (FTH) were examined for their ability to induce B cell proliferation. At concentrations observed in macrophage supernatants (200 ng.ml−1), all ferritins produced significant B lymphocyte proliferation when compared to a potent activation stimuli (αIgD/IL-4/rCD40L) [Figure 5A]. Since PF was most representative of Nef-induced, macrophage-derived ferritin, its ability to induce B cell activation and differentiation was further studied by flow cytometry. PF stimulation caused an increase in expression of the markers CD38 and CD70 on populations of B cells (Figure 5B) and surface IgM+ B cells co-expressing the plasma cell marker CD138 where evident consistently two days later on CD38+ B cells (Figure 5C). Consequently we next examined whether PF, FTL or FTH could induce immunoglobulin secretion. We observed a dose-dependent increase in the production of IgA, IgG and IgM following addition each of the three ferritins to resting B cells (Figure 5D).
Appropriate signaling through the B cell receptor promotes B-cell proliferation and differentiation via activation of cell cycle promoting, anti-apoptotic genes (Niiro and Clark, 2002) and changes in cellular immunology. We compared the expression of host genes involved in the signal transduction, transcription, B cell activation and differentiation by targeted gene arrays on resting B cells and those stimulated with PF or the B cell stimulus, αIgD/IL-4/rCD40L. The responses of resting B cells to PF and the B cell differentiation stimulus were highly comparable and revealed an activation of the MAPK and NF-κB pathways (Figure 5E and 5F). The ability of ferritin to induce B cell proliferation was significantly impaired in the presence of the NFκB inhibitor CAPE and an inhibitor of MEK-dependent ERK activation (PD98059), (Figure 5G) suggesting that ferritin mediated its effect on B cell function via the NF-κB and MEK/ERK pathways. B cells exposed to either PF or αIgD/IL-4/rCD40L showed elevated expression of mRNAs for activation antigens, plasma cell marker CD138 and immunoglobulin genes (Figure 5H). In contrast, analysis of mRNAs encoding an array of transcription factors suggested some differences in transcription profiles in B cells stimulated by PF as opposed to αIgD/IL-4/rCD40L (Figure 5E and 5H). Collectively, these results indicate that ferritin promotes changes in B cell gene expression that are mostly reminiscent of those elicited by a classical B cell stimulus.
In order to evaluate whether ferritin plays a role in B cell dysfunction during HIV-1 infection, we examined whether there was a correlation between plasma ferritin concentration, plasma viral RNA load and immunoglobulin levels in a cohort of 83 HIV-1-infected individuals possessing a wide variation in viral load. As suggested by a previous study where the reduction in patient viral load due to the application of HAART resulted in a corresponding decrease in plasma ferritin levels (Boom et al., 2007), we observed a similar relationship between viral RNA and ferritin in HIV-1 infected patients (Figure 6A). Here, a statistically significant relationship was evident between viral load and plasma ferritin (p<0.0001) with a strong degree of correlation between the two parameters (r=0.79) [Figure 6A]. There was also a statistically significant correlation between plasma viral RNA load and hypergammaglobulinemia in HIV-1-infected individuals, with plasma IgA (p<0.0001; r=0.70), IgG (p<0.0001; r=0.84) and IgM levels (p<0.0001; r=0.73) exhibiting a significant and strong relationship to the extent of viral burden (Figure 6A). Our data also demonstrated that a highly significant correlation existed between plasma ferritin and the levels of IgA, IgG and IgM in HIV-1 infected patients (Figure 6B). There was a strong relationship between plasma ferritin and plasma IgA (p<0.0001; r=0.48), plasma ferritin and plasma IgG (p<0.0001; r=0.54), and plasma ferritin and IgM (p<0.0001; r=0.57) [Figure 6B]. This indicated that the extent of viral replication, and hence ferritin, in HIV-1 infected patients were intrinsically linked to the genesis of hypergammaglobulinemia in agreement with our in vitro findings. In contrast, no correlation was observed between plasma viral RNA levels and cytokines known to affect B cell function including B-Lys (B Lymphocyte Stimulator) or Granulocyte-Colony Stimulating Factor (data not shown).
Ferritin is an acute-phase reactant and is elevated as a consequence of inflammatory disease, infection and certain neoplasms (Olive and Junca, 1996). Other than the cellular activation evident in HIV-infected macrophages that results in the aberrant production of ferritin (Figure 2), the cohort of infected patients plasma ferritin did not correlate with systemic immune activation as evidenced by examination of plasma levels of soluble TNF receptor (sTNFRII) [p=0.4911; r=0.12] or β-2 microglobulin [p=0.2395; r=0.01] (Figure 6C). To further confirm the relationship of HIV infection with hyperferritinemia and hypergammaglobulinemia, ferritin and immunoglobulin levels were determined in patients with similar viral loads (~104–105 copies per ml) versus normal healthy control subjects (Figure 6D). HIV-infected individuals demonstrated a statistically significant elevation of plasma ferritin (p=0.0019), plasma IgA (p=0.0037), plasma IgG (p=0.0432) and plasma IgM (p=0.0026) [Figure 6D]. Collectively, this indicates that ferritin is a correlate of viral RNA and hypergammaglobulinemia but not of immune activation in HIV-1-infected individuals.
We next examined the role of HIV-1 Nef to directly induce the production of ferritin and associated B cell defects in the CD4C/HIVNef transgenic murine model of AIDS where only Nef among the genes of HIV-1 is expressed in mature and immature CD4+ T cells and in cells of the myeloid lineage, including macrophages (Hanna et al., 1998). The ability of HIV-1 Nef to induce ferritin production in mouse cells was first confirmed by transient transfection of the murine macrophage cell line RAW 264.7. In these cells, Nef promoted the release of high levels of ferritin (Figure 7A) and there was a statistically significant relationship between the introduction of Nef and ferritin (p<0.0001; n=12) [Figure 7B]. In serum samples derived from CD4C/HIVNef transgenic mice, a significant hypergammaglobulinemia; IgA [(p=0.0147). IgG (p=0.0068) and IgM (p=0.0115)] relative to that in their normal control non-Tg littermates (n=20), could be detected (Figure 7C), a result consistent with previous results (Hanna et al., 1998). In addition, these Tg mice showed enhanced ferritin levels (p=0.0003) in their serum, to an extent similar to HIV-infected patients (Figure 7C). However, Tg serum sTNFRII levels showed no indication of systemic immune activation (data not shown). Although Nef is expressed in both CD4+ T cells and macrophages in these transgenic animals, HIV-1 infection with macrophages but not T cells led to induction of ferritin (Figure 2G). This suggests that macrophages are the principal source of ferritin in Nef-transgenic mice. Thus the Nef-dependent release of ferritin from macrophages in vitro relates directly to the occurrence of increased circulating ferritin and immunoglobulin levels in vivo and requires only the Nef gene of HIV-1.
Our studies reveal a pathogenicity determinant within Nef that is responsible for a central feature of HIV-1 immunopathogenicity. The inappropriate activation and differentiation of B cells during HIV-1 infection impairs the development and maintenance of a normal humoral immune response (De Milito et al., 2004) and whilst the mechanisms responsible for collateral B cell dysfunction are complex, soluble factors regulated by Nef in HIV-1-infected macrophages, are a significant factor. Elevated ferritin levels in HIV-1 infection have been reported (Gordeuk et al., 2001) but the pathophysiological relevance of this has not previously been ascertained. We report that HIV-1 infected macrophages release ferritin due to Nef expression and that this induced the proliferation of resting B lymphocytes and their differentiation in to immunoglobulin-producing plasma cells. The accessory protein Nef is necessary for viral replication in vivo and has a role in subverting the humoral immune response. We present evidence that Nef activates ferritin production in macrophages via the NF-κB pathway. Motifs in Nef that are responsible for NF-κB activation remain to be identified. Previous studies show that myristoylation and proline-repeat mutants of Nef failed to potentiate T-cell receptor activation and subsequent NF-κB signaling because they were mis-localized in the cell (Fenard et al., 2005). In addition, studies have suggested that EE155 and DD174 motifs in Nef were required for the ability of extracellular, recombinant Nef to activate NF-κB in macrophages (Mangino et al., 2007; Olivetta et al., 2003). However, Nef deletion mutants lacking these motifs were still capable of inducing chemokines in HIV-1-infected macrophages (L. Dai, unpublished data). Since we have previously demonstrated that chemokine induction by Nef is NF-κB dependent (Swingler et al., 2003), these domains may be dispensable for NF-κB activation in the context of Nef synthesized de novo during viral infection of macrophage. Preliminary analysis indicates that the ability of Nef to activate NF-κB signaling can genetically be uncoupled from other reported activities of Nef including CD4 and MHCI downregulation (Dai et al, unpublished observations).
Previous reports indicate that ferritins have immunomodulatory functions that extend beyond their accepted role in iron metabolism. Ferritin promotes the growth of B-leukemic cell lines (Kikyo et al., 1995) and Swingler et al., data not shown). Modification of ferritin levels in HeLa cells by RNA interference does not influence iron availability but positively affected cell proliferation rate in an iron-independent manner (Cozzi et al., 2004). Ferritin-binding activity has been observed on a variety of cell types, including B cells and T cells (Anderson et al., 1989; Fargion et al., 1991) but the identity of the receptor for ferritin is unclear, as well as proximal membrane events that accompany ferritin binding. However, recently it has become evident that ferritin co-operates with a number of known cellular receptors for G-CSF and SDF-1 (Li et al., 2006; Yuan et al., 2004). The interaction of ferritin with G-CSF-R transduces a signal and likely explains the previously described importance of ferritin in myelopoesis whereas binding of ferritin to CXCR4 modulated chemokine receptor signaling and chemotaxis (Li et al., 2006). These observations underscore the emerging importance of ferritins in regulating lymphocyte function and our studies indicate that ferritins induced by HIV-1 Nef account for B cell abnormalities observed in HIV-1 infection. In the context of viral infection, another study has demonstrated that HIV-1 Nef manipulates cellular iron homeostasis by down regulating the hemochromatosis protein HFE (Drakesmith et al., 2005). As a result, Nef–induced, HFE-dependent ferritin iron accumulation in macrophages, which has also been previously reported in HIV-1 infected microglia (Yoshioka et al., 1992), is consistent with our study demonstrating that ferritin is a target of HIV-1 Nef. Modulation of HFE likely also serves to aid virus-infected cells evade immune recognition through the associated down-regulation of class I MHC, and thus ferritin induction by Nef may contribute to a comprehensive strategy enabling enhanced viral growth and persistence in the face of a hostile immune environment (Ben-Arieh et al., 2001; Drakesmith et al., 2005).
Ferritin, whether derived from Nef-expressing macrophages, HIV-1 infected macrophages, or in purified and recombinant form, was sufficient to induce a pattern of B cell hyperactivation and elevated immunoglobulin synthesis mirroring B cell dysfunction observed in vivo (Conge et al., 1998; De Milito, 2004; De Milito et al., 2004). In vitro, ferritin alone recapitulated the activated phenotype of B lymphocytes derived from HIV-infected patients. This was evident by up-regulation of the activation marker CD38 (Conge et al., 1998), the expression of CD70; which has been shown to be aberrantly displayed by naïve B cells from HIV-infected individuals (De Milito, 2004), and the increase in CD138+ immunoglobulin-secreting plasma cells. The antigen non-specific activation of resting B cells by ferritin released by infected macrophages is likely a consequence of chronic HIV infection and immune activation. In this scenario, persistent antigen exposure depletes the memory B cell pool and leads to the development of naïve, activated B lymphocytes that undergo activation and differentiation in response to abnormal stimuli (De Milito, 2004). Such a hypothesis was supported from flow cytometry analyses which indicated that immunoglobulin secreting CD138+ cells remained highly CD19 positive (data not shown).
The positive correlations between plasma viral RNA load, ferritin and immunoglobulin levels seen in HIV-1 infected individuals were similarly found in mice transgenic for HIV-1 Nef and paralleled the ability of Nef to induce ferritin from macrophage cultures in vitro. Both HIV-infected patients and mice transgenic for Nef exhibit B cell dysfunction. These results suggest that the elevated levels of ferritin in HIV-1 infected individuals are a direct result of viral replication and resultant Nef expression. Furthermore, the observation that ferritin correlated strongly with plasma viral RNA and immunoglobulin levels but not with markers of immune activation (sTNFRII; β-2 microglobulin) or markers of immune function such as Neopterin and Serum Albumin (data not shown) indicated that hyperferritinemia was not a consequence of generalized immune activation. In HIV-1-infected individuals, elevated ferritin production may be restricted to the lymphoid tissues where HIV-1-infected macrophages are abundant (Embretson et al., 1993). In HIV-1 infected individuals, intestinal mucosa, which is a major reservoir of infected macrophages, contain 100-fold higher frequencies of antibody-secreting cells than blood (Eriksson et al., 1995). Therefore, B cells within the lymphoid tissue may be predisposed to the localized effects of soluble factors that are released from infected macrophages.
We have previously shown that HIV-1 Nef intersects the CD40 signaling pathway in macrophages and, as a result, induces the release of soluble factors which enhance susceptibility of CD4+ T cells to HIV-1 infection (Swingler et al., 2003). These soluble proteins do not act directly on T cells. Rather, they stimulate the expression of co-stimulatory molecules on B cells which, in turn, engage corresponding ligands on T cells and render them permissive to HIV-1 infection (Swingler et al., 2003). The impact of Nef on T cell permissivity is mechanistically distinct from its effect on ferritin production and B cell proliferation. Nef intersects the CD40 signaling pathway in order to affect T cell permissivity (Swingler et al., 2003). However, we observed that induction of ferritin by Nef was CD40-independent and immunodepletion of ferritin from Nef-expressing macrophage supernatants did not affect the ability of those supernatants to promote T cell permissivity (data not shown). Thus, Nef induces the release of soluble factors from macrophages that both promote conditions for viral spread (in the case of T cell permissivity) and drive dysfunction in the humoral immune system (in the case of B cell activation).
Lymphocytes and monocytes were obtained by leukapheresis from normal donors seronegative for HIV-1 and Hepatitis B. Populations of T and B lymphocytes were obtained by additional purification with antibody coated magnetic beads (Dynal-Invitrogen, Carlsbad CA) according to manufacturer’s instructions. Cell purity was determined by flow cytometric staining with fluorochrome antibodies to CD3 (T cell), CD19 (B cell), and CD45 (Leukocyte) [BD Pharmingen, San Diego CA]. Monocytes were further separated by counter-current centrifugal elutriation (Gendelman et al., 1988) (and elutriated monocytes differentiated to macrophages by culture for 4 days in medium containing 1000 U.ml−1 M-CSF (R & D Systems, Minneapolis, MN) and for a further 3 days in medium lacking M-CSF. Macrophages were then used for virus infections within 1–5 days. R5 tropic HIV-1ADA viruses were prepared by transient transfection of HeLa cells and all virus stocks were standardized by Reverse Transcriptase assay (Swingler et al., 1999). The production of recombinant adenoviruses, transduction of macrophages and the preparation of macrophage supernatants and immune-depletion has been described previously (Swingler et al., 1999).
Recombinant human ferritin Heavy and Light Chain proteins, purified human Liver ferritin, Caffeic Acid Phenethyl Ester (CAPE) and PD98059 were obtained from Calbiochem (EMD Biosciences, La Jolla CA). Recombinant human IL-4, TNFα and murine IFNγ were supplied by R & D Systems and human recombinant soluble CD40L by Alexis Corporation (San Diego CA). Recombinant human IκBα was from Santa Cruz Biotechnology (Santa Cruz CA) and recombinant Nef was supplied by Immunodiagnostics (Woburn MA). Antigen capture ELISAs for ferritin and immunoglobulins were constructed with antibody coated plates and HRP-conjugated secondary antibody-specific antibodies were obtained from Dako Corporation (Carpintera CA), ICN (Costa Mesa CA) and Fitzgerald Industries International Inc (Concord MA). Antibody specific to human and murine immunoglobulin subtypes IgD, IgA, IgG, IgM, control normal immunoglobulins and LPS were obtained from Sigma (St. Louis MO). ELISA kits for human and murine soluble TNFRII and β-2 microglobulin were supplied by R & D Systems. ELISA kits specific for murine ferritin was from Panapharm Laboratories (Kunamoto Japan). Antibodies to IKKα and IKKβ were obtained from Santa Cruz Biotechnology and for Western blotting and immuno-depletion of HIV-1 Nef antibodies were purchased from ABI (Foster City CA).
For induction of lymphocyte DNA synthesis, T or B cells were incubated with macrophage supernatants or ferritins for 3 days and DNA synthesis (3H-Thymidine incorporation) measured 16 hrs after the addition of 1 μCi 3H-Thymidine. At 5 and 7 days post stimulation, B cell differentiation to Plasma cells was analyzed by flow cytometry using fluorochrome-conjugated antibodies (BD Pharmingen) to CD19, CD38, CD70 or surface IgM and CD138, respectively. Secretion of immunoglobulins in to cell supernatants was determined at the same time by ELISA. For both assays, B cells stimulated for 16 hrs with Anti-IgD (25 μg ml−1), IL-4 (5 ng.ml−1) and rCD40L (10 ng.ml−1) served as a positive control. To assess the effects of HIV-1 Nef-expressing or HIV-1-infected macrophage supernatants on B cell differentiation, purified B cells were incubated with macrophage supernatants and flow cytometric analysis for CD19, CD38 and CD70 expression was performed after 5 days while CD138 and sIgM was analyzed after 7 days. FACS scatter plots of CD138 and sIgM expression was performed on CD38+-gated B cells. Supernatants from HIV-1-infected macrophages were obtained at the peak of viral replication as measured by reverse transcriptase activity in culture supernatants.
The molecular weight of the B cell stimulatory activity released from macrophages in response to Nef expression was determined by separation through Centricon filters (Millipore, Billerica MA) of increasing molecular size. Subsequently, Serum-free supernatant (50 ml) from adenovirus-transduced macrophages was concentrated 250 fold by centrifugation through a series of 100kD NMWL Centricon filters and resolved by gel electrophoresis. Following staining with Coomassie Blue, candidate protein bands were excised, digested with trypsin and subjected to micro-sequencing and mass spectroscopy as detailed elsewhere (Guilherme et al., 2000). Western blotting was performed with antibodies reactive to human ferritin (ICN). In earlier experiments, serum-free supernatants derived from macrophages expressing Nef of control GFP 16 hours after adenoviral transduction were concentrated approximately 250-fold by centrifugation through 10kD NMWL Centricon filters and resolved by non-denaturing protein gel electrophoresis. Protein was determined by co-migration of non-denatured molecular weight markers and coomassie blue staining of a parallel gel. Following electrophoresis, equal 1 cm slices of the gel were excised and proteins eluted in 0.5 ml Tris buffered saline, agitated overnight at 4°C. Eluates were subsequently sterilized by filtration and the approximate molecular weight of the B cell stimulatory activity determined as described in lymphocyte assays.
The levels of mRNAs were assessed on total RNA from 1.4 × 106 macrophages 4 hrs following adenoviral transduction or TNFa stimulation (2 hrs after stimulation with 1000 U.ml−1 TNFα for 2 hrs). Total RNA was prepared by Trizol (Invitrogen, Carlsbad CA) and analyzed by SyBr Green real time RT-PCR (Quantitect SyBr Green kit; Qiagen, Valencia CA) using gene-specific primers for FTL and FTH (primer sequences given in the Supplementary Methods). B lymphocyte gene expression was analyzed after 4 days stimulation with ferritin or Anti-IgD/IL-4/rCD40L using pathway and cell-specific cDNA micro-arrays (Superarray, Frederick MD) as described previously (Swingler et al., 2003). For ferritin mRNA analysis quantitative RT-PCR was performed using the following primers to specifically amplify nucleotide sequences corresponding to FTL ([s] 5-AGAATTATTCCACCG ACGTGG-3′; [as] 3′-GGTAGAGACACTGAAGGACCT-5′) and FTH ([s] 5′-CGACCGCGTCC ACCTCGCAGG-3′; [as] 3′-GGAAGTCCTATAGTTCTTTGG-5′).
To assess NF-κB induction in macrophages, the activity of the upstream kinases Iκκa/b was determined by immunoprecipitation and in vitro assay. Briefly, 9×106 cells transduced with adenoviruses, mock infected or stimulated with 1000 U.ml−1 TNFα were lysed in RIPA buffer after incubation for 2 hrs. Cellular IKKα/β was precipitated by specific antibodies and immune complexes subject to in vitro kinase assay with 32P-γ-ATP and 2.5 μg recombinant IκBα. Kinase reactions were resolved by SDS-PAGE and proteins transferred to PVDF for Phosphor Imaging or Western blotting. For the analysis of ERK3, 5×106 B cells stimulated as indicated for 2 hrs were lysed and similarly immunoprecipitated and assayed for kinase activity with 2.5μg Myelin Basic Protein (Sigma) as an appropriate substrate. Phosphorylation of IκBα on Serine 32 was determined by Western blotting using specific anti-sera from Cell Signaling Technologies (Danvers MA). The activation of NF-κB by HIV-1 Nef was further demonstrated by co-transduction of macrophages with equal amounts of recombinant adenoviruses expressing a dominant negative IκBα, or a LacZ control. The dominant negative IκBα protein contained Serine to Alanine mutation at positions 32 and 36 thus preventing its phosphorylation and subsequent release of NF-κB (Iimuro et al., 1998). Cellular turnover of mRNAs for IκBα and NF-κB-p105 was performed as described except analyses were performed using cDNA micro-gene arrays (Superarray) and Ribonuclease Protection Assays (Swingler et al., 2003).
Blood samples were obtained from HIV-1 positive individuals the UMass Memorial Medical Center outpatient clinics. Patients samples were selected for this study on the basis of providing a broad range of CD4 T cell counts (%CD4=5–53%; absolute CD4 T cell numbers were 20–1660 cell.ml−1) and plasma viral RNA levels were from <50 to >900,000 copies.ml−1. Samples were obtained prior to or during periods of cessation from anti-retroviral treatment. Informed consent was obtained from all HIV-1 individuals and the studies were approved by the UMMS Human Subjects Committee. The CD4C/HIVMutG (designated here CD4C/HIVNef) Tg mice have been described previously (Hanna et al., 1998). Serum samples from animals (3–7 months old) confirmed to be transgenic for Nef and from their normal non-Tg control litter mates were used for the analysis of ferritin and immunoglobulin levels.
The statistical significance of data, where indicated, was determined by ANOVA or two-tailed t-test. Mean values +/− SEM are shown graphically and annotations indicate the confidence level (p value) and number of replicates (n). Viral load, ferritin and immunoglobulin levels were analyzed by linear regression where r described the strength of the relationship and two-tailed t-test described significance. In all analyses, p values < or = 0.05 were considered significant. Statistical calculations were performed using Prism 5 (GraphPad Software).
We thank Katherine Luzuriaga M.D. and John Sullivan M.D. for providing clinical samples from HIV-1 infected individuals and Ronald C. Desrosiers Ph.D. for advice and critical reading of the manuscript. Bruce Blais for FACS analysis and the core facilities within the University of Massachusetts Center for AIDS Research (P30-AI42845) for reagents and cells. Viral clones were obtained from the AIDS research and reference reagent program, NIAID, NIH. This work was supported by grant 106520-35-RGRL from the American Foundation for AIDS Research (amfAR) to S. Swingler, from The Canadian Institute of Health Research, HIV/AIDS Research Program to P. Jolicoeur and from NIH grants RR11589 and MH64411 to M. Stevenson.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.