Interferon–β Induces Hepatocyte Growth Factor in Monocytes of Multiple Sclerosis Patients

Interferon-β is a first-line therapy used to prevent relapses in relapsing-remitting multiple sclerosis. The clinical benefit of interferon-β in relapsing-remitting multiple sclerosis is attributed to its immunomodulatory effects on inflammatory mediators and T cell reactivity. Here, we evaluated the production of hepatocyte growth factor, a neuroprotective and neuroinflammation-suppressive mediator, by peripheral blood mononuclear cells collected from interferon-β−treated relapsing-remitting multiple sclerosis patients, relapsing remitting multiple sclerosis patients not treated with interferon-β, and healthy volunteers. Using intracellular flow cytometry analysis, increased production of hepatocyte growth factor was observed in circulating CD14+ monocytes from patients undergoing long-term treatment with interferon-β versus untreated patients. Complementary in vitro studies confirmed that treatment with interferon-β induced rapid and transient transcription of the hepatocyte growth factor gene in CD14+ monocytes and that intracellular and secreted monocytic hepatocyte growth factor protein levels were markedly stimulated by interferon-β treatment. Additional exploration revealed that “pro-inflammatory” (CD14+CD16+) monocytes produced similar levels of hepatocyte growth factor in response to interferon-β as “classical” (CD14+CD16−) monocytes, and that CD14+ monocytes but not CD4+ T cells express the hepatocyte growth factor receptor c-Met. Our findings suggest that interferon-β may mediate some of its therapeutic effects in relapsing-remitting multiple sclerosis through the induction of hepatocyte growth factor by blood monocytes by coupling immune regulation and neuroprotection.


Introduction
The recognition of multiple sclerosis (MS) as an inflammatory, demyelinating, and neurodegenerative disease of the CNS [1] emphasizes the necessity for therapeutic strategies to target inflammation and neurodegeneration simultaneously. [2] This can be accomplished with combination therapies or by applying molecules that are capable of targeting both pathogenic processes. In the search for single molecules combining these requirements, recent reports suggest that hepatocyte growth factor (HGF) is one such candidate [3,4].
HGF has strong neuroprotective properties [3,5] reported to enhance the survival and maturation of myelin producing oligodendrocytes. [6,7] HGF also exerts anti-inflammatory effects through T cell bystander deviation and inhibition of antigen-presenting cell (APC) function. [8,9] In animal models of demyelinating diseases, HGF was recently shown to confer protective immunoregulation, and to promote myelin repair in the absence of modulation of the immune system. [3,4] Likewise, data from MS patient suggest that HGF may also potentially contribute to stimulation for remyelination [10].
Interferon-b (IFN-b) is a first-line treatment for relapsingremitting (RR)MS [11,12] shown to exert potent immunoregulatory effects on myeloid cells such as monocytes, [13] and likely to confer neuroprotection through secretion of neuroprotective factors, [14] including HGF by microglia. [6] These data suggest that, by combining immunomodulatory and neuroprotective effects, HGF may be a promising mediator of the clinical benefit of IFN-b treatment in RRMS.
Here we report that monocytes from RRMS patients exhibited a reduced ability to produce HGF when compared with healthy volunteers, and that monocytes from IFN-btreated RRMS patients produced significantly higher levels of HGF. These findings provide new information regarding the mechanisms that mediate the therapeutic effects of IFN-b in RRMS.

Study Design
Twenty-seven RRMS patients fulfilling the 2005-revised Mc-Donald Criteria [15] and 17 age-and sex-matched healthy individuals were recruited at the University Hospital of Geneva in accordance with institutional guidelines. All patients were classified as RRMS in remission for at least 3 months according to clinical history. Patients were required to be free of corticosteroid medication for at least 3 months before any blood sampling. All treated patients were receiving one single approved preparation of IFN-b-1a 44 mg 3 times/week subcutaneously (RebifH, Merck Serono, Germany). Treated patients had to be on the same treatment regimen for at least 12 months. This study was performed using a single sample of venous blood. To minimize the possible acute effects of IFN-b, all samples were taken in the morning before the next scheduled injection, which was routinely self-administered in the evening. Characteristics of patients and healthy controls are described in Table 1. While pharmacokinetic information of IFN-b-1a on patients with RRMS has not been evaluated, the steady-state concentration of IFN-b-1b is usually between 40 and 80 IU/mL in patients receiving 8610 6 IU subcutaneously three times a week. [16,17] Higher concentrations of IFN-b are known to be necessary to achieve effects on immune parameters in vitro [18].

Standard Protocol Approvals, Registrations, and Patient Consents
This study received approval from the Geneva University Human Studies Committee Institutional Review Board for experiments using human subjects. The participants provided their written informed consent to participate in this study. The ethics committee approved this consent procedure.

Cell Isolation and Culture Conditions
PBMCs were obtained by density gradient centrifugation of human peripheral blood over Ficoll-Paque (Pharmacia). Isolated PBMCs were resuspended in complete culture medium consisting of RPMI 1640 medium supplemented with 10% fetal bovine serum, 4 mM L-glutamine, 25 mM Hepes buffer, 50 U/ml penicillin, and 50 mg/ml streptomycin (all components were purchased from Life Technologies). For certain experiments, CD14 + monocytes and CD4 + T cells were further isolated using negative selection microbeads (EasySep). The purity of CD14 + and CD4 + T cells obtained by negative selection was routinely .90%, as determined by flow cytometry analyses, PBMCs, CD14 + monocytes or CD4 + T cells were cultured for the indicated time with or without IFN-b-1a alone.

Flow Cytometry HGF
For analysis of HGF expression by monocytes from healthy controls and RRMS patients, 32 ml (468 ml) of venous blood was drawn into 3.2% buffered sodium citrate Vacutainer Tubes (BD) and processed within 2 h. Cells were stained with antibodies against CD14 and HGF or its IgG isotypic control, and analyzed on a FACSAria (BD Pharmingen) using CellQuest software (BD Pharmingen). For each sample, the geometric mean fluorescence intensity (Gmean) of specific HGF-positive cells was compared with that of isotype-matched control cells. The ratio was calculated as the Gmean for HGF-positive cells divided by the Gmean for isotype-matched control cells. To determine c-Met expression among different PBMC populations, cells were stained with antibodies against c-Met, CD4, CD8, CD14, CD16, CD19, CD56 and the respective isotype controls from the same manufacturer (BD Pharmingen).

HGF Immunoassay
PBMCs and/or monocytes were activated for 24 h to 72 h with the indicated dose of IFN-b-1a. HGF production was measured in culture supernatants using ELISA (Quantikine, R&D). All experiments were performed with PBMCs or untouched monocytes obtained by negative magnetic-activated cell separation (MACS). Experiments are representative of PBMCs or monocytes isolated from at least three different blood donors. Alternatively, serum HGF level was determined from sodium citrate plasma (BD Vacutainer tubes, for 8 ml of blood). The results for ELISA assays are expressed as an average of triplicate wells6standard error of mean (S.E.M). SOFTmax ELISA plate reader and software (Molecular Devices Corporation) were used for data analysis.

Quantitative Reverse Transcription Polymerase Chain Reaction
Monocytes were activated with IFN-b-1a (200,000 UI/ml) for the indicated times. Total RNA was isolated using the RNeasy micro Kit (Qiagen) and quantitative real-time duplex PCR analysis (ABI 7500; Applied Biosystems) was conducted after reverse transcription with the iScript cDNA Synthesis Kit (Bio-Rad). Primers and probes were obtained from Applied Biosystems. The levels of HGF mRNA expression were normalized with the expression of a simultaneously analyzed housekeeping gene (bactin). All measurements were conducted in triplicate.

Preparation of Cell Extracts
Monocytes were activated with 50,000 to 500,000 IU/ml of IFN-b-1a. At the indicated time, the activation was stopped and total cell lysates were prepared and subjected to Western blot analysis. Total protein content was determined using the Bio-Rad DC protein assay (Bio-Rad).

Western Blot Analysis
Samples were loaded onto a 12% sodium dodecyl sulfate polyacrylamide gel (20 mg protein per lane). Proteins were transferred onto a nitrocellulose membrane (Bio-Rad). After blocking, the membrane was probed with a polyclonal anti-HGF antibody (1 mg/ml) (Santa Cruz) and further incubated with a secondary HRP-conjugated antibody (Amersham) in the blocking buffer. Immunoreactive HGF was detected using an ECL kit (Amersham). To re-probe the blot for the housekeeping control, the membrane was stripped and incubated with an anti-GAPDH antibody (Santa Cruz).

Statistical Analysis
Mean and median fluorescence intensities were recorded for each subset of cells, as defined by regions in two-color scatter plots. The level of HGF expression was determined from frequency histograms, as the difference between the median channels of the specific antibody sample and the isotype control (specific median, sMed). Differences between medians of controls and patients were statistically analyzed by Mann-Whitney U test. Prism version 5.0c was used for all statistical procedures. Data are given as mean6SD and SEM.

In Vitro Induction of HGF Release by Human PBMCs in Response to IFN-b
We first evaluated the potential of PBMCs, particularly T cells, to secrete HGF in response to IFN-b. IFN-b induced a dose-and time-dependent increase in PBMC-secreted HGF, with the maximal effect obtained with 200,000 IU/ml of IFN-b at 72 h post-stimulation (Fig. 1A). In contrast, IFN-b did not induce release (Fig. 1B) or production (Fig. 1C) of HGF by CD4 + T cells, an important source of other neurotrophic factors (e.g., brainderived neurotrophic factor, BDNF) in MS patients during immunomodulatory drug therapy.

IFN-b Induces the Production and Release of Biologically Active HGF Protein by Monocytes
We next investigated the role of IFN-b in HGF synthesis by monocytes, a primary cell target for immunomodulatory drugs in CNS autoimmunity. [19] Stimulation of monocytes with IFN-b resulted in a time-dependent release of HGF by the cells in the media, as determined by ELISA. A significant increase in HGF levels was observed as early as 24 h post-treatment, and was greatest at 72 h post-stimulation ( Fig. 2A) (p,0.0001). Because this assay recognized both pro-HGF (inactive) and mature HGF, we next evaluated the ability of IFN-b to induce the production of biologically active (mature) HGF by monocytes. Western-blot analysis of the cell-associated fraction of IFN-b-treated monocytes demonstrated that HGF detected in monocyte supernatants was mainly in the cleaved mature form (presence of the 69-KDa a-chain) (Fig. 2B, C). Moreover, IFN-b induced HGF production in monocytes in a time-and dose-dependent manner (Fig. 2B, C). Using real-time PCR, we confirmed rapid gene expression of HGF by monocytes in response to IFN-b treatment, indicating that IFN-b induces HGF by activating its transcription (Fig. 2D).
HGF Levels are Similar between CD14 + CD16 2 and CD14 + CD16 + Monocytes With the use of anti-CD14 and anti-CD16 antibodies, human monocytes can be divided into two populations. Most cells are CD14 strongly positive (CD14 ++ ) and CD16 negative, representing what had previously been referred to as monocytes. Cells expressing CD14 at low levels together with the CD16 molecule (i.e., CD14 + CD16 + monocytes) usually comprise 10% of all monocytes. With regard to their inflammatory phenotypes and functional properties in inflammation, CD14 + CD16 + monocytes have been labeled proinflammatory. [20] Therefore, we compared HGF production by both monocyte subsets in response to IFN-b treatment by intracellular cytokine flow cytometry. IFN-b induced HGF expression in both monocyte subsets in a time-and dosedependent manner with similar potency (Fig. 3A, B).

HGF Receptor (c-Met) Is Expressed by CD14 + Monocytes and B Cells, but not T Cells
For HGF production by monocytes to be of biological significance in immune reactions, the cells engaged in immune responses would need to express the HGF receptor. We used flow cytometry to test whether T and B lymphocytes expressed the high-affinity HGF receptor c-Met. c-Met is expressed by CD19 + B lymphocytes, but not CD4 + or CD8 + T cells, creating the potential for a functional interaction between monocytes and B lymphoid cells (Fig. 4). These studies confirm previous publications with other lymphoid tissues revealing strong c-Met expression by B cells, but not T cells. [21] Additional experiments revealed c-Met expression by both classical and   pro-inflammatory CD14 + monocyte fractions of PBMCs. Finally, c-MET was expressed by CD56 + CD16 + natural killer cells (Fig. 4).

Monocytes from IFN-b2Treated MS Patients Exhibit Increased HGF Production
To study the effect of in vivo IFN-b therapy on monocytic HGF secretion in RRMS patients, we compared cell-associated HGF levels in CD14 + monocytes from RRMS patients treated with IFN-b to those of age-and sex-matched untreated patients (see Table 1). Monocytes from IFN-b2treated RRMS patients (n = 12) expressed significantly higher levels of HGF compared to those of untreated patients (n = 15) (Fig. 5A, B). We also evaluated the monocytic expression of cell-associated HGF in healthy controls, untreated RRMS patients, and RRMS patients treated with IFN-b (RRMS-IFN-b patients). Monocytes from healthy controls and RRMS-IFN-b patients exhibited significantly higher levels of HGF than monocytes from untreated RRMS patients (Fig. 5B). HGF was measured in the sera of IFN-b2treated and untreated patients, as well as healthy controls (Fig. 5C). Median serum HGF levels were not statistically different among the three groups (615 ng/ml (controls) vs. 705 ng/ml (RRMS) vs. 534 ng/ml (RRMS-IFNb), p = ns).

Discussion
In this study, we addressed the question of whether IFN-b treatment could induce the synthesis and secretion of HGF, a protein with immunoregulatory and neuroprotective properties, by PBMCs. To this purpose, HGF protein expression was analyzed in separate PBMC populations. We observed that IFN-b treatment induced strong HGF secretion by PBMCs, in particular by monocytes. Consistent with this observation, monocytes from untreated RRMS patients exhibit reduced HGF production compared with monocytes from IFN-b2treated RRMS patients. Flow cytometry for HGF was performed on CD14 + monocytes from the three groups (healthy control, untreated RRMS patients, and IFN-b2treated RRMS patients). Monocytes were stained for surface CD14 antigen and cell-associated HGF. Data show representative histogram overlays of isotype (filled histogram) and HGFstained cells (unfilled histogram). (B) Cell-associated HGF levels in CD14 + cells were higher in healthy controls and IFN-b-treated RRMS patients. Surface expression was measured by flow cytometry and calculated as the mean corrected fluorescence index (MFI) ratio. Background HGF expression was assessed by measuring the fluorescence of cells incubated with a nonspecific isotype control antibody similarly labeled. The MFI for control anti-HGF antibody isotype staining was divided with the HGF MFI of monocytes. (C) Median serum HGF levels were similar in all three groups. ELISA for HGF was performed on sera from the three groups (healthy control, untreated RRMS patients, and IFN-b2treated RRMS patients). For both monocyte and serum HGF level analysis, each circle represents a single individual and the lines show the medians. Difference in median levels between groups was examined by Kruskal-Wallis test followed by Mann-Whitney U test due to a non-Gaussian distribution of values. *, p,0.05 and ***, p,0.001. doi:10.1371/journal.pone.0049882.g005