Concomitant detection of IFNα signature and activated monocyte/dendritic cell precursors in the peripheral blood of IFNα-treated subjects at early times after repeated local cytokine treatments

Interferons alpha (IFNα) are still the cytokines most widely used in clinical medicine today, with applications both in oncology and in the treatment of certain viral infections [1]. Several decades of research on IFNα have revealed that these cytokines exert immunomodulatory activities possibly involved in their in vivo therapeutic efficacy, spanning from the differentiation of the Th1 subset, the generation of CTL and the promotion of T cell in vivo proliferation and survival [reviewed in ref. [2]]. In particular, IFNα have proved to play an important role in the differentiation of monocytes into dendritic cells (DC) and in enhancing DC activities [3‐8]. It has been suggested that IFNα-mediated DC activation can represent one of the mechanisms underlying the cytokine therapeutic efficacy in vivo[2].

In the attempt to understand in more detail the mechanisms of IFNα in vivo, several studies have recently utilized microarray technologies to detect and analyze an IFNα-specific signature in the peripheral blood cells of IFNα-treated individuals, with particular focus on HCV and melanoma patients [9‐15]. These studies have revealed that many interferon-stimulated genes [16] (ISG), previously known to be induced by this cytokine in other animal or human in vitro settings, can be found up-regulated in the blood of patients treated in vivo with the cytokine. Furthermore, novel and unexpected ISG were added to the list of possible in vivo mediators of IFNα immunomodulatory and/or antitumor activity [9‐15]. Defining with acceptable accuracy the pool of genes considered to be the signature of IFNα in vivo helps to understand the involvement of this cytokine in clinical as well as therapeutic settings [17, 18]. Notably, an IFNα signature has been observed in systemic lupus erythematosus (SLE) patients, suggesting that the overexpression of a specific set of genes can represent the hallmark of in vivo cell exposure to IFNα, which is commonly detected in the sera of these patients [19]. More recently, the presence of a prominent IFNα signature has been reported in patients experiencing a growing list of autoimmune disorders, including psoriasis, multiple sclerosis, rheumatoid arthritis, dermatomyositis, primary biliary cirrhosis and insulin-dependent diabetes mellitus [20]. These data, together with the autoimmune-like phenomena reported in melanoma patients responding to IFNα therapy [21], confirmed the involvement of this cytokine in the delicate balance between immunity and autoimmunity.

Besides helping to gain insight into IFNα mechanisms of action in vivo, identifying a clear-cut IFNα signature ex vivo opens the possibility to define patterns of gene expression profiles significantly associated with IFNα treatment efficacy. In turn, this may also provide insights into candidate predictor biomarkers of response to therapy, and possibly assist in making the appropriate therapeutic decisions when a patient does not present with a favorable response profile. In spite of many efforts performed in this direction, the literature in this field suffers from a lack of consistency among the results obtained from patients suffering from different diseases and receiving different IFNα preparations. The majority of these studies have been performed in patients chronically infected with HCV, while attempting to identify a consensus blood biomarker predictive of IFNα/Ribavirin efficacy in patients blood [9‐12, 15]. Since it is known that the pattern of PBMC gene expression in HCV patients is altered by the infection itself [15], IFNα-induced modulations observed in these patients may be somehow related to the HCV disease, and possible affected by individual-specific variability, thus providing little information on the general mechanisms of action of the cytokine per se.

Despite the accumulating information on the IFNα-induced genes and of their possible in vivo role, little is known about the consistency of the IFNα signature in healthy vs cancer patients. A still elusive area of investigation is the kinetics of gene up-regulation in correlation with the possible appearance of immune cells elicited by IFNα and playing a primary role in the biological responses of IFNα-treated cancer patients. Likewise, no information is currently available on the transient and long-term effect of low doses of IFNα used with modalities typical of a vaccine adjuvant, as IFNα, in spite of their now recognized role as natural links between innate and adaptive immunity [2], have been extensively and generally used in clinics as typical antiviral or antitumor drugs. As a matter of fact, although the more effective and better tolerated pegylated IFNα2b is now widely used for the therapy of HCV infection [22] and in the adjuvant melanoma setting [23], no study is currently available on the clinical use of this molecule administered as vaccine adjuvant.

In the present study, we utilized PBMC derived from melanoma patients and healthy individuals, who had been enrolled in two clinical trials with similar treatment schedule, aimed at assessing the role of IFNα administered as vaccine adjuvant. We exploited microarray technology to evaluate and compare the modulations of PBMC global gene expression profiles induced by IFNα in melanoma and normal donors. The effects of the administration of different doses of IFNα, as well as of repeating the administration of the cytokine in successive treatment cycles, were evaluated. The kinetics and the biological significance of the modulations observed at the transcriptional level were correlated with the phenotypic changes observed in circulating CD14⁺ and CD14⁺/CD16⁺ monocytes. The overall results provide new insights in the identification of specific biomarkers for adjuvant IFNα and in the characterization of new molecular and cellular players mediating the response to this cytokine in patients.

In the study presented herein, we applied microarray technology to profile the gene expression in human PBMC treated in vivo with IFNα, administered at low dose as vaccine adjuvant in the context of two separate clinical trials, performed on melanoma patients and healthy subjects, following a similar treatment schedule. A clear-cut signature of IFNα in vivo could be observed in human PBMC 24 hours after the cytokine administration in both clinical studies (Figure 1). Interestingly, the modulation of PBMC global gene expression profile was consistently induced after each repeated administration of the cytokine, suggesting that, at least at the transcriptional level, the extent of the modulations induced by the cytokine is mainly transient, and does not reach a steady state level refractory to further stimulations (Figure 2). In general, the transcriptional modulations observed appear quite homogeneous among the different subjects analyzed, and no major differences between groups of subjects receiving two different doses of the cytokine were observed (Figure 3).

The results of our transcriptional profiling provided the molecular basis supporting a predominant immunomodulatory role of IFNα when administered as vaccine adjuvant. According to the gene ontology analysis (Figure 4), the immunological pathways influenced by IFNα in vivo recapitulate the progression of the main steps for the generation of a specific immune response, from the early non specific antiviral defense (OAS, MX), to inflammation (TLR7, NMI, CXCL10, MYD88), recruitment of immune cells (CXCL10, C3AR1, CX3CR1), antigen processing and presentation (PSMB9, HLA-DOA) and finally to the effectors specific immune response: (SERPING1, C2, BST2, MYD88, TNFSF13B/BAFF).

Since PBMC is a heterogeneous population consisting of various subsets of cells that may experience different responses to IFNα, changes at the transcript level observed in total PBMC specimens cannot be ascribed to a specific immune effect. However, when we compared human PBMC isolated after the in vivo administration of the cytokine, to PBMC or purified monocytes isolated from healthy donors and exposed in vitro to the cytokine, we found a significant correlation among the IFNα-up-modulated genes in the various group, so that we were able to define a "core" IFNα signature consistently observed in all the in vivo and in vitro settings (Figure 5). Interestingly, among the genes consistently up-regulated by IFNα associated with inflammation, we found the metalloprotease MMP-1 not previously reported to be an ISG (to the best of our knowledge), and involved in extra cellular matrix degradation for cells migration and tissue remodeling, during physiological and pathological conditions [35]. Of interest, MMP-1 can be released by macrophages, monocytes [35] and monocyte-derived DC [36], and alteration of its expression has been recently associated with autoimmunity phenomena [36, 37]. The "core" signature of IFNα identified in our in vitro and in vivo experiments also included BAFF, a gene showing a crucial role in B cell maturation and activity, reported to be involved in the pathogenesis of autoimmune diseases, such as Rheumatoid arthritis, SLE and Progressive Systemic Sclerosis in mouse models as well as in humans [38]. Moreover, our list included at least 4 genes belonging to TLR7 pathway (Myd88, IRF7, CXCL10 and STAT1), a system responsible for the activation of the innate immune response in response to RNA viruses, but also implicated in IFNα-related autoimmune phenomena, mainly through plasmacytoid DC [39]. Interestingly, TLR7 have been reported to be expressed by IFN-DC, which could also secrete IFNα following viral stimulation or TLR7-specific stimulation, thus confirming the critical role of this cytokine at the early steps of immune response to pathogens or in autoimmune diseases [8].

Of interest, although a rigorous comparison among the results of different microarray studies is impaired by the bias possibly induced by different platforms and statistical approaches, the core IFNα signature identified by us in subjects receiving the cytokine as vaccine adjuvant is not considerably different, in terms of modulated genes and Gene Ontology categories, from the one reported by studies investigating the same issue in HCV-infected patients treated with IFN (IFNα2b or PEG-IFNα2b) and Ribavirin [9‐16]. Moreover, our data on the transcriptional modulations observed in PBMC treated in vitro with IFNα2b are concordant with data reported by others on the effects on the same cells of the pegylated form of the cytokine administered in association with Ribavirin [40]. Overall, these observations strongly suggest that a similar signature occurs both in vivo and in vitro (at least in PBMC), regardless of the dose or type of IFNα used or even of the condition of the subjects receiving the cytokine (healthy donors and HCV-infected or cancer patients).

The proteomic analysis of the supernatants of monocytes exposed in vitro to IFNα (Additional data file 3) confirmed at the protein level the effect of this cytokine on chemoattraction and inflammation observed at the transcription level in vitro and in vivo, corroborating the results of the gene ontology analysis on the immunomodulatory role of IFNα in vivo. Although further studies on specific cell subsets isolated ex vivo from subjects receiving IFNα are needed to define the role of monocytes in the cytokine activity in vivo, our results suggest that monocytes contribute to the transcriptional modulation seen on total PBMC, in line with previous observations from our group and others on IFNα linking innate and adaptive immunity by affecting monocytes differentiation into DC [reviewed in ref. [2]].

To gain more insight into the specific effects of IFNα on the several monocytes blood populations, we analyzed the immunophenotype changes observed in PBMC isolated from healthy donors before and after IFNα administration, with particular focus on CD14⁺ cells. Of note, at the same time of detection of the typical IFNα signature in PBMC, we also observed an increase in the percentage of CD14⁺ and CD14⁺/CD16⁺ monocytes, and both cell populations proved to express high levels of costimulatory molecules and HLA-DR (Figure 6). Notably, such increase was transient, similarly to the appearance of the IFNα signature, and additional rounds of increase were observed at 24 hr after the subsequent IFNα treatments, in parallel with the de novo detection of an up-regulated expression of the typical IFNα-induced genes.

CD14⁺/CD16⁺ monocytes, coexpressing CD16 and low levels of CD14, were first characterized by Ziegler-Heitbrock and colleagues in 1988 [41], and their number and phenotype/function have been reported to be altered in patients with cancer, infectious diseases or inflammatory disorders [42‐45]. Of note, an increase of CD14⁺/CD16⁺ monocytes was observed in patients infected with pathogens triggering IFNα production, such as certain bacteria and HIV [45]. In general, this cell subset has been indicated as a transitional stage of development of monocytes to macrophages, originating from CD14^highCD16⁺, or DC, derived from CD14^dimCD16⁺ cells [46], and has been shown to exhibit special capabilities to migrate [47], to stimulate CD4⁺ T cells [48] and to produce proinflammatory cytokines [45]. Moreover, CD16⁺ monocytes can differentiate in CD1b⁺ DC endowed with high APC capacity after a short time exposure to TLR2 ligands [49], supporting the concept that these cells may represent natural precursors of DC in response to danger signals. In the light of all this, it is possible that the transient up-regulation of costimulatory molecules and HLA-DR in CD16⁺ monocytes, occurring at the time of the appearance of a PBMC IFNα molecular signature, characterized by enhanced expression of immune-related cytokines/chemokines, can represent a reliable marker of the biologic response to a local IFNα treatment, which may result in the generation of active DC, resembling those naturally generated from this monocyte subset in response to infections and danger signals. Notably, an ensemble of studies from our group and from other laboratories have demonstrated that IFNα can induce a very rapid differentiation of highly active DC from monocytes [50] and these DC (IFN-DC) are characterized by a special signature [51], which partially overlaps with the IFNα-signature described in the present study. In this regard, it is worth mentioning recent results indicating that spontaneous regression of highly immunogenic Molluscum contagiosum virus-induced skin lesions is associated with the infiltration of DC strongly resembling IFN-DC [52], supporting the concept that IFN-DC can indeed represent naturally occurring DC promptly generated in vivo during the response to type I IFN induced by viruses and other natural danger signals.

Of interest, a recent study by Mohty and colleagues [53] has shown the increase of CXCL-10 plasma levels in melanoma patients treated with relatively low doses of IFNα, which also parallels a trend towards an increase of CD16⁺ monocytes. CXCL10 is an IFNα-induced chemokine, which binds and activates the seven transmenbrane G-protein-coupled receptor CXCR3, and is expressed especially in activated Th1 cells, B cells, NK cells and DC, thus suggesting that CXCL10 release can represent a primary event in the amplification of the IFNα response. The results of Mohty and coworkers [53] are consistent with our data showing an up-regulation of CXCL10 expression after local low-dose IFNα injection, as revealed by both genomic and proteomic analysis ex vi vo and in vitro. Of note, the up-regulation of CXCL-10 has been reported to occur also in HCV-infected patients shortly after the administration of PEG-IFNα2b [8], so that it has been suggested that CXCL-10 can represent a marker predictive of the final treatment outcome [54].

Overall, the results presented herein show that: i) the production of CXCL10 and a specific IFNα-signature are observed in PBMC as early as 24 hr after cytokine injection in healthy donors and melanoma patients; ii) such response is transient, does not reach a steady-state level of refractoriness and can occur after inoculation of as little as 1 millions of units of IFNα. iii) The observed molecular signature is paralleled by the raise in percentage and expression of costimulatory molecules of CD14⁺/CD16⁺ peripheral blood cells, reported to be precursors of DC in response to danger signals. These results shed a new light on the immune mechanisms of action of IFNα and, in particular, on the role of CXCL10 and early effects of IFNα on monocyte/DC precursors (such as CD14⁺/CD16⁺ cells) as primary players in the IFNα response and stimulate further studies for identifying molecular and biological markers capable of predicting the clinical response to IFNα.