Progranulin (PGRN) is a pleiotropic growth factor with immunosuppressive properties. Recently, it was reported that PGRN was an antagonist of tumour necrosis factor (TNF) receptors, preferentially for TNFR2. However, we and others showed that TNF–TNFR2 interaction was critical for the activation and expansion of functional CD4+ Foxp3+ regulatory T (Treg) cells. We therefore examined the effect of PGRN on the proliferation of naturally occurring murine suppressive Treg cells induced by TNF. Consistent with our previous reports, TNF overcame the hyporesponsiveness of highly purified Treg cells to T-cell receptor stimulation. Furthermore, in the presence of interleukin-2, TNF preferentially stimulated proliferation of Treg cells contained in unfractionated CD4 cells. These effects of TNF on suppressive Treg cells were markedly increased by exogenous PGRN. TNF and TNFR2 interactions are required for this effect of PGRN, because the PGRN by itself did not stimulate Treg cell proliferation. The effect of PGRN on Treg cells was abrogated by antibody against TNFR2, and Treg cells deficient in TNFR2 also failed to respond to PGRN. Furthermore, PGRN also enhanced the proliferative responses of effector T cells to TNF, but to a lesser extent than that of Treg cells, presumably caused by the different levels of TNFR2 expression on these two subsets of CD4 cells. Hence, our data clearly show that PGRN promotes, rather than inhibits, the functional consequence of TNF–TNFR2 interaction on Treg cells.
Progranulin (PGRN) – also known as granulin/epithelin precursor, proepithelin, acrogranin, or prostate cancer cell-derived growth factor – is a 70 000-molecular weight secreted glycosylated protein. As a pleiotropic growth factor, PGRN has been shown to play a critical role in regeneration, tumorigenesis, inflammation, wound healing, insulin resistance and neurodegeneration.[1, 2] Recently, the effect of PGRN on immune responses and inflammation has been extensively studied.
There is considerable evidence that PGRN has immunosuppressive and anti-inflammatory properties. For example, PGRN-deficient mice are more susceptible to neuroinflammation.[4-6] In contrast, over-expression of PGRN in mice showed a neuroprotective effect against inflammation, and glial cells from PGRN transgenic mice produced lower levels of pro-inflammatory cytokines such as interleukin-1β (IL-1β), IL-6 and tumour necrosis factor (TNF), and increased levels of the anti-inflammatory cytokine IL-10 upon lipopolysaccharide stimulation. In patients with frontotemporal lobar degeneration, loss of function mutations of the PGRN gene are related to a marked increase of IL-6 serum levels and to the prevalence of autoimmune diseases. However, a positive feedback loop has been reported because granulins (GRN, also known as epithelins), which are 6000 molecular weight peptide fragments generated from proteolytically degraded PGRN, function as a pro-inflammatory factor. Nevertheless the immunosuppressive effects of the intact PGRN appear to prevail.
Although the anti-inflammatory and immunosuppressive activity of PGRN is compelling, the cellular receptor mediating this effect remained unknown until Tang et al. reported that PGRN bound to the extracellular domain of TNF receptors, especially TNFR2, with an affinity even higher than that of TNF. They showed that PGRN inhibited TNF binding to TNFR1 and TNFR2, and consequently acted as a physiological antagonist of TNF signalling. However, this observation was not confirmed by studies that failed to show the binding of PGRN to the membrane-associated TNFRs and failed to show the capacity of PGRN to inhibit TNF–TNFRs signalling.[11, 12]
We and others have reported that TNF–TNFR2 interaction can preferentially activate and expand CD4+ Foxp3+ regulatory T (Treg) cells, and promote the phenotypic and functional stability of Treg cells in the inflammatory environment.[13-15] As Treg cells are critical for the maintenance of immunological homeostasis and prevention of autoimmune responses,[16, 17] blockade of TNF–TNFR2 pathways may result in an inflammatory effect by eliminating Treg cell activity. We therefore investigated the effect of PGRN on TNF/TNFR2-mediated proliferation of functional mouse Treg cells. Surprisingly, we found that PGRN, at physiologically relevant nanomolar concentrations, actually promoted the proliferative response of naturally occurring suppressive Treg cells, and this effect of PGRN is dependent on the TNF–TNFR2 pathway.
Materials and methods
Mice and reagents
Normal C57BL/6 mice and TNFR2−/− mice were provided by the Animal Production Area of the National Cancer Institute (Frederick, MD). Frederick National Laboratory for Cancer Research is accredited by AAALAC International and follows the Public Health Service Policy for the Care and Use of Laboratory Animals. Animal care was provided in accordance with the procedures outlined in the Guide for Care and Use of Laboratory Animals (National Research Council; 1996; National Academy Press; Washington, D.C.). Anti-mouse CD4 (GK1.5), CD25 (PC61) and TNFR2 (CD120b, TR75-89) antibodies were purchased from BD Biosciences (San Diego, CA). Functional grade purified anti-mouse CD3e (eBio500A2) and CD28 (37.51) antibodies, and Foxp3 Staining Set (FJK-16s) were purchased from eBioscience (San Diego, CA). Murine IL-2 and TNF were purchased from PeproTech (Rocky Hill, NJ). Functional grade anti-mouse TNFR1 (55R-170) and TNFR2 (TR75-32.4) were purchased from Biolegend (San Diego, CA). Murine PGRN was purchased from Adipogen (San Diego, CA)
To prepare a single-cell suspension, spleens and lymph nodes (inguinal, axillary and mesenteric regions) were gently mashed and passed through a 70-μm mesh (BD Labware, San Jose, CA). CD4+ T cells were purified using magnetic beads coated with anti-CD4 antibody (clone L3T4) according to the manufacturer's instructions (Miltenyi Biotec Inc., Auburn, CA). Subsequently, the CD4+ cells were stained with anti-CD4, anti-CD25 antibodies and sorted into naive CD4+ CD25− T cells and CD4+ CD25+ Treg cells (> 92% of Foxp3+cells).
In vitro proliferation of T cells
Flow-sorted CD4+ CD25+ Treg cells or CD4+ CD25− effector T (Teff) cells from wild-type C57BL/6 mice or TNFR2−/− mice were seeded at 1·25 × 104 to 2·5 × 104 cells/well in a U-bottomed 96-well plate. The cells were stimulated with 2 × 105 cells/well antigen-presenting cells (APCs) (CD4-depleted splenic cells, 3000 rad-irradiated) and functional grade anti CD3e antibody (2 μg/ml), with or without TNF (20–50 ng/ml), in the presence of medium or increasing concentration of PGRN (2–200 ng/ml). Cells were pulsed with 1 μCi [3H]thymidine (Perkin Elmer Life Sciences, Boston, MA) per well for the last 6 hr of the culture period. In some experiments, CFSE-labelled unfractionated CD4 cells (1 × 105/wells) were cultured with IL-2 (20 ng/ml) with or without TNF (20 ng/ml), in the absence or in the presence of increasing concentrations of PGRN (1–200 ng/ml). After incubation for 72 hr, CFSE dilution was determined by FACS, by gating on Foxp3+ or Foxp3− T cells. In some experiments, flow-sorted Treg cells were stimulated with plate-bound anti-CD3e antibody (5 μg/ml) and soluble anti-CD28 antibody (2 μg/ml) for 3 days, and expression of Foxp3 was analysed by FACS. RPMI-1640 (Lonza BioWhittaker, Walkersville, MD) was used in all other cultures. The medium was supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) containing 2 mm glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin, 10 mm HEPES, 1 mm sodium pyruvate, 0·1 mm non-essential amino acids and 50 μm 2-mercaptoethanol.
After blocking FcR, cells were incubated with appropriately diluted antibodies. Appropriate species-matched antibodies served as isotype controls. For detection of Foxp3, cells were fixed and permeabilized using the anti-mouse Foxp3 staining kit (FJk-16S, eBioscience). Acquisition was performed using an LSRII (BD Biosciences, Mountain View, CA) and data analysis was conducted using FlowJo software (Tree Star Inc., Ashland, OR). FACS analysis was gated on the live cells only by using a LIVE/DEAD Fixable Dead Cell Stain Kit (Life technologies™, Grand Island, NY).
Data were analysed by one-way analysis of variance test using Graphpad Prism 6.0 (Graphpad Software, Inc., La Jolla, CA). Differences were considered statistically significant when P was < 0·05.
In the presence of TNF, PGRN promotes proliferative responses of purified Treg cells to T-cell receptor stimulation
First we examined the effect of PGRN on the in vitro proliferative response of Treg cells to T-cell receptor (TCR) stimulation. Previously we showed that the profound hyporesponsiveness of Treg cells to in vitro TCR stimulation was overcome by exogenous TNF. We now observed that this effect of TNF on the proliferative response of highly purified Treg cells to APCs and anti-CD3 antibody stimulation was markedly increased by exogenous PGRN (P < 0·01, Fig. 1a). The proliferation of Treg cells was increased by 147% and 166% by 2 and 20 ng/ml of PGRN respectively (P < 0·05 to P < 0·01, Fig. 1b). This effect of PGRN was limited to the low nanomolar doses, and then largely disappeared at concentrations > 200 ng/ml (Fig. 1). Interestingly, PGRN only enhanced proliferation of Treg cells in the presence of TNF, but by itself PGRN did not stimulate the proliferation of Treg cells under these experimental conditions.
Effect of PGRN on Treg cells is dependent on the TNF–TNFR2 pathway
It is known that the Treg-activating effect of TNF was mediated by TNFR2,[13-15] we therefore examined whether TNFR2 was required for PGRN to further promote proliferation of Treg cells induced by TNF. As shown in Fig. 2(a), the proliferation of highly purified Treg cells in response to exogenous TNF was further enhanced by PGRN (P < 0·05). This effect of PGRN was not blocked by antibody against TNFR1. In contrast, anti-TNFR2 antibody blocked TNF-induced proliferation of Treg cells and also completely blocked the effect of PGRN on Treg cells. For further verification, Treg cells purified from wild-type mice and TNFR2−/− mice were stimulated with APCs plus anti-CD3 antibody, with or without TNF, in the presence of increasing concentrations of PGRN. As shown in Fig. 2(b), unlike wild-type Treg cells, Treg cells deficient in TNFR2 did not respond to TCR stimulation even in the presence of TNF. Furthermore, the effect of PGRN on Treg cells was also absent if Treg cells were deficient in TNFR2. Therefore, our data clearly show that TNF–TNFR2 interactions are required for PGRN to stimulate the proliferation of Treg cells.
Effect of PGRN on the expression of Foxp3 and TNFR2 of Treg cells
It was reported and we tried to confirm that TNF down-regulated FoxP3 expression by Treg cells and PGRN abolished this effect of TNF. To this end, we stimulated flow-sorted Treg cells (> 92% of Foxp3+ cells) with plate-bound anti-CD3 antibody and soluble anti-CD28 antibody for 3 days. In this experiment system, expression of Foxp3 by Treg cells was markedly reduced by the potent TCR stimulation. However, we observed that TNF markedly increased the proportion of Foxp3-expressing cells (P < 0·01), which was not further enhanced by PGRN (Fig. 3a).
One possible mechanistic basis for the observed effect of PGRN on Treg cells was that it may up-regulate surface expression of TNFR2 on Treg cells, and therefore enhance signalling from TNF–TNFR2 interaction. In agreement with our previous report, in the presence of IL-2, TNF did up-regulate surface expression of TNFR2 on Treg cells (P < 0·01), however, PGRN did not further enhance this effect of TNF (Fig. 3b). Hence, up-regulation of TNFR2 appears not to be responsible for the stimulatory effect of PGRN on Tregs.
The anergic state and suppressive function of Tregs was proposed to be linked. So, any agent that has the capacity to promote proliferation of Treg cells may hamper their immunosuppressive function. We therefore also examined the effect of PGRN on the suppressive function of Treg cells with a standard in vitro Treg function assay. The result showed that PGRN over a concentration range of 0·1–10 ng/ml did not interfere with Treg function (data not shown). Therefore, in addition to promoting proliferation of Treg cells, PGRN does not interfere with the highly suppressive properties of Treg cells on a per cell basis. However, as the number of Treg cells was increased, the overall level of suppression was also greater when co-stimulated by PGRN.
PGRN also stimulates the proliferative responses of Teff cells to TCR stimulation
To examine if the effect of PGRN was limited to Treg cells, we also observed its effect on the proliferative responses of Teff cells to TCR stimulation. As shown in Fig. 4(a), TNF also enhanced the proliferation of Teff cells in response to APCs and anti-CD3, although this effect of TNF was markedly less potent compared with its effect on Treg cells (Fig. 1). Exogenous PGRN (20 ng/ml) also markedly enhanced the proliferative response of Teff cells to TNF (P < 0·01), although PGRN by itself had no effect on the proliferation of Teff cells (Fig. 4a–b). The addition of PGRN (20 ng/ml) resulted in a 131% increase of proliferation of TNF-stimulated Teff cells, which is markedly lower compared with the 166% increase of Treg cell proliferation in response to PGRN and TNF (Fig. 1b, P < 0·05). Hence, in the presence of PGRN, the proliferative responses of Teff cells to TNF and TCR stimulation was enhanced to a lesser degree compared with that of Treg cells.
PGRN preferentially promotes proliferation of Treg cells driven by TNF without TCR stimulation
Since TCR signalling plays a major role in APCs plus anti-CD3-mediated proliferation in the aforementioned experiments and enhancement of TCR signalling may be attributable to the effect of PGRN. To further clarify the roles of TNF and TNFR2 in the effects of PGRN on Treg and Teff cells, we analysed the proliferation of Treg and Teff cells contained in unfractionated CD4 cells stimulated by TNF in the presence of IL-2. To this end, MACS-purified CD4 cells were labelled with CFSE and cultured with medium containing IL-2 to maintain their survival in vitro, with or without TNF, in the presence of increasing concentrations of PGRN (1–100 ng/ml). After 72 hr, the proliferation of Treg and Teff cells was analysed based on dilution of CFSE signalling, by gating on Foxp3+ cells and Foxp3− cells, respectively. In this experimental system, TNF markedly increased the proliferation of Treg cells and proliferating Treg cells increased from 16% in medium alone to 46·8% (Fig. 5a, P < 0·01). In contrast, TNF treatment had almost no impact on the proliferation of Teff cells (Fig. 5a). Furthermore, PGRN at 20 ng/ml was able to enhance TNF-mediated proliferation of Treg cells (P < 0·05, proliferating Treg cells was 56·6%), although PGRN by itself did not promote proliferation of PGRN in the presence of IL-2 (Fig. 5a). As a result of proliferative expansion of Treg cells in the CD4 cells, the proportion of Foxp3+ Treg cells was markedly increased by TNF (P < 0·05), and further increased by TNF plus PGRN (P < 0·05, compared with TNF alone, Fig. 5b). These data further confirm that the effect of PGRN on Treg cells is mediated in conjunction with TNF, and PGRN preferentially promotes proliferation of Treg cells, even in the absence of TCR stimulation.
The immunosuppressive activity of PGRN was reportedly associated with its effect on Treg cells. It has been reported that the number of Treg cells was reduced in PGRN knockout mice upon induction of experimental dermatitis and that PGRN promoted the differentiation of induced Treg cells. Further, Tang et al. also reported that PGRN protected naturally occurring Treg (nTreg) cells from negative regulation by TNF, since TNF in their report down-regulated Foxp3 expression and inhibited the suppressive function of nTreg cells. Our data agree that PGRN has the capacity to promote nTreg activity, but we found it to be based on a completely different mechanism. We show that PGRN further enhanced the positive stimulatory effect of TNF on functional Treg cells through TNFR2. In the study by Tang et al., the suppressive function of Treg cells was measured by inhibition of interferon-γ (IFN-γ) production from co-cultured Teff cells. The same study also showed that PGRN by itself inhibited production of IFN-γ from Teff cells, hence it is most likely that their observed result actually reflects the suppressive activity of PGRN on IFN-γ production by Teff cells in the co-cultures, rather than blockade of TNF's effect. Interestingly, Tang et al. found that TNFR2−/− collagen-induced arthritis mice were less sensitive to PGRN-derived Atsttrin treatment, and it is known that the number of Treg cells is reduced in TNFR2−/− mice. Hence this evidence is actually compatible with our observation that TNF-TNFR2 was required for PGRN to expand Treg cells. We were able to observe that PGRN modestly but significantly inhibited IFN-γ expression by T helper type 1 (Th1) -polarized CD4 cells (data not shown), while enhancing IL-13 expression by Th2-polarized CD4 cells (data not shown), which was consistent with previous studies.[3, 20] Hence, multiple mechanisms such as expansion of nTreg cells and inhibition of Th1 responses may contribute to the anti-inflammatory activity of PGRN.
TNFR1 accounts for the majority of the pro-inflammatory, cytotoxic and apoptotic effects classically attributed to TNF,[21, 22] although there is considerable evidence that TNFR2 mediates the immunosuppressive action of TNF.[23, 24] Hence, it is possible that PGRN behaves as a partial antagonist of TNFR1 and partial agonist of TNFR2. PGRN potently inhibited TNF-mediated inflammatory effects and increased the survival of actinomycin D-treated mouse WEHI 13v fibrosarcoma cells (our unpublished data). Hence, PGRN may have the characteristics of an antagonist of TNFR1. However, this possibility is not supported by a recent report that both human and mouse PGRN did not inhibit TNFR1-mediated activation of nuclear factor-κB and mitogen-activated protein kinases.
It was reported that PGRN exhibits ~ 600-fold higher binding affinity to TNFR2 than TNF, and PGRN-induced transcription of target genes in chondrocytes was mediated by TNFR2, hence PGRN may be an agonist of TNFR2. However, the failure of PGRN by itself to stimulate TNFR2 on Treg cells directly is difficult to explain. It has been shown that binding by PGRN may have dramatic consequences, as shown by its ability to bind CpG, as well as proteins, and markedly augment the activities of CpG. Furthermore, PGRN is also known to bind to other receptors such as sortilin,[1, 2] and the signalling triggered by such PGRN receptor(s) may enhance the signalling and functional consequences of TNF/TNFR2. In fact, Chen et al. observed that PGRN potentiated the signalling of TNF, which did not attribute to the PGRN cleavage. The molecular basis of this effect of PGRN merits future study.
Both human and mouse Foxp3+ Treg cells expressed markedly higher levels of TNFR2 compared with their Foxp3− counterparts. For example, the majority or all of mouse and human thymic Treg cells, but not Teff cells, constitutively express TNFR2.[27, 28] In normal human peripheral blood mononuclear cells, almost all CD4+ FoxP3+ Treg cells regardless of their CD25 expression are TNFR2-expressing cells, whereas CD4+ FoxP3− Teff cells expressed markedly lower or no TNFR2 expression.[28, 29] In the peripheral lymphoid tissues of normal mice, TNFR2 is constitutively expressed on 30–40% of resting Treg cells, which can be further enhanced by stimulation of TCR[13, 30] or TNF. In contrast, mouse peripheral Teff cells did not constitutively express TNFR2.[13, 30] Although TCR stimulation up-regulated TNFR2 expression on mouse Teff cells, its expression levels were still markedly lower than that on Treg cells activated in the same manner.[13, 30] Unlike TNFR1, TNFR2 lacks an intracellular death domain and mediates signals promoting activation and proliferation of lymphocytes.[31, 32] Interestingly, PGRN appears to be more potent in enhancing TNF-mediated proliferation of Treg cells than Teff cells. This is consistent with the previous observations that TNF preferentially co-stimulates Treg cells, due to the markedly higher TNFR2 expression by Treg cells in both resting and activated state.[13, 18, 28] Treg cells are primed by self-antigen when generated in the thymus, so they are naturally occurring activated T cells. Therefore, our results suggest that PGRN may also preferentially promote the proliferation of antigen-experienced, pre-activated T cells in vivo.
Our data do not support the findings described by Tang et al. that PGRN blocked TNF–TNFRs interaction. As 10–250 ng/ml of PGRN was used in Treg-related studies by Tang et al., which was in the same concentration ranges as our study (1–200 ng/ml), the discrepancy is not caused by the different doses of PGRN. In fact, we failed to observe the TNFR2 blocking effect of even higher concentrations of PGRN (1000 ng/ml, data not shown). Tang et al. also reported that the ‘Atsttrin’, a recombinant molecule synthesized from three non-contiguous granulin domains, shared the similar ‘TNF antagonistic’ properties. Therefore, it is possible that the reported antagonism of TNFR2 of PGRN is actually due to the effect of granulins generated by degradation of PGRN under their culture conditions. These possibilities need to be investigated in future studies.
Taken together, we present novel data in this report that PGRN has the capacity to promote the proliferative expansion of nTreg cells and to a lesser extent of Teff cells, in a TNF- and TNFR2-dependent manner. This effect of PGRN may contribute to its immunosuppressive function. Further understanding the molecular basis of this effect of PGRN may help in the design of new therapeutics in the treatment of autoimmune disorders.
This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under contract HHSN26120080001E. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. This Research was supported (in part) by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. The authors thank Drs O.M. Zack Howard and Qiong Zhou, Ms Trivett L. Anna and Ms Czarra T. Kelli for their help in this study. We thank the NCI-Frederick Cancer Inflammation Program Fluorescence Cytometry core for expert technical assistance with flow cytometry.
Conflict of interest
The authors declare that there are no conflicts of interest.