Cellular Signalling 26 (2014) 683–690

Contents lists available at ScienceDirect

Cellular Signalling

j ourna l homepage: www.e lsev ie r .com/ locate /ce l l s ig

TNFR2 increases the sensitivity of ligand-induced activation of the p38MAPK and NF-κB pathways and signals TRAF2 protein degradationin macrophages☆

Gerhard Ruspi a, Emily M. Schmidt a, Fiona McCann a, Marc Feldmann a, Richard O. Williams a,A. Allart Stoop b, Jonathan L.E. Dean a,⁎a Kennedy Institute of Rheumatology, Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, Old Road Campus, Roosevelt Drive, Headington,Oxford OX3 7FY, United Kingdomb Innovation Biopharm Discovery Unit, Biopharm R&D, GlaxoSmithKline, Cambridge CB4 0WG, United Kingdom

Abbreviations: TNF, tumour necrosis factor; TNFR, TNFassociated factor 2; MAPK, mitogen-activated protein kinaNF-κB, nuclear factor-κB; LPS, lipopolysaccharide; MCSF,factor; GM-CSF, granulocyte/macrophage colony stimulatBMDM, bone marrow-derived macrophages; PBMC, perip☆ The authors claim no conflict of interest. This study wArthritis Research UK and the Kennedy Institute of Rheumtant for GlaxoSmithKline.⁎ Corresponding author. Tel.: +44 1865 612641.

E-mail address: jonathan.dean@kennedy.ox.ac.uk (J.L.

0898-6568/$ – see front matter © 2013 Elsevier Inc. All rihttp://dx.doi.org/10.1016/j.cellsig.2013.12.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 29 October 2013Received in revised form 19 December 2013Accepted 22 December 2013Available online 27 December 2013

Keywords:TNFTNFRp55p75TRAF2Macrophage

Tumour necrosis factor (p55 or p60) receptor (TNFR) 1 is themajor receptor that activates pro-inflammatory sig-nalling and induces gene expression in response to TNF. Consensus is lacking for the function of (p75 or p80)TNFR2 but experiments in mice have suggested neuro-, cardio- and osteo-protective and anti-inflammatoryroles. It has been shown in various cell types to be specifically required for the induction of TNFR-associatedfactor-2 (TRAF2) degradation and activation of the alternative nuclear factor (NF)-kappaB pathway, and to con-tribute to the activation of mitogen-activated protein kinases (MAPK) and the classical NF-kappaB pathway. Wehave investigated the signalling functions of TNFR2 in primary human andmurine macrophages. We find that inthese cells TNF induces TRAF2 degradation, and this is blocked in TNFR2−/− macrophages. TRAF2 has been pre-viously reported to be required for TNF-induced activation of p38 MAPK. However, TRAF2 degradation does notinhibit TNF-induced tolerance of p38 MAPK activation. Neither TNF, nor lipopolysaccharide treatment, inducedactivation of the alternative NF-kappaB pathway in macrophages. Activation by TNF of the p38 MAPK and NF-kappaB pathways was blocked in TNFR1−/− macrophages. In contrast, although TNFR2−/− macrophagesdisplayed robust p38 MAPK activation and IkappaBα degradation at high concentrations of TNF, at lower dosesthe concentration dependence of signalling was weakened by an order of magnitude. Our results suggest that,in addition to inducing TRAF2 protein degradation, TNFR2 also plays a crucial auxiliary role to TNFR1 insensitising macrophages for the ligand-induced activation of the p38 MAPK and classical NF-kappaB pro-inflammatory signalling pathways.

© 2013 Elsevier Inc. All rights reserved.

1. Introduction

There are two separate TNF receptors: the type I receptor (TNFR1,p55, or p60) and the type II receptor (TNFR2, p75, or p80). Both re-ceptors are widely expressed but TNFR1 is thought to be the majorreceptor required by a wide variety of cells for activation of thepro-inflammatory NF-κB and MAPK signalling pathways which inturn induce the expression of proteins of the inflammatory response(for reviews see [13,17,31]).

receptor; TRAF2, TNF receptor-se; P-p38, phospho-p38MAPK;macrophage colony stimulatinging factor; I-κB, inhibitor of κB;heral blood mononuclear cells.as funded by GlaxoSmithKline,atology Trust. MFwas a consul-

E. Dean).

ghts reserved.

In contrast, TNFR2 displays cell-type specific expression and is themajor TNFR expressed on activated T cells in which it signals the induc-tion of apoptosis [9,35] in a process involving sensitisation to TNF [9].TNFR2 is also constitutively expressed in regulatory T cells and plays arole in their activation, proliferative expansion, and survival [7]. Endo-thelial cells express more equal levels of both receptors [20], but inthese cells, TNFR1 is thought to be themajor receptor involved in the ac-tivation of the NF-κB andMAPK pathways and subsequent induction ofinflammatory response proteins [4,21,24,36].

Immortalised macrophages are reported to express both receptorswith TNFR1 being expressed weakly [23]. Both TNFR1- and TNFR2-deficient macrophage cell lines display reduced activation of the NF-κB and MAPK pathways in response to TNF compared to immortalisedwild-type macrophages [23]. Experiments employing artificial ligandsfor the different receptors, also called muteins, have shown thatTNFR2 cannot significantly activate pro-inflammatory signalling path-ways independently of TNFR1 [6,8].

Physiological signalling functions of TNFR2 that are distinct fromTNFR1 have proved elusive to identify. In cell lines, TNFR2 has been

684 G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

reported to induce TRAF2 [18,33] and ASK1 [34] protein degradation,and to regulate protein kinase B expression [16]. Recently, it has beenshown in primary T cells and in tumour cell lines to activate the alterna-tive NF-κB pathway [26]. Membrane-bound TNF has been shown to be amore potent activator of TNFR2 than the soluble form of the ligand[12,26]. In addition, TNFR2 shedding in response to TNF serves to neu-tralise TNF and inhibit signalling [25].

TNFR2 has been reported to regulate apoptosis via a mechanism of‘ligand-passing’, shown to involve TNFR2-mediated facilitation of theactivation of TNFR1 [29]. In this mechanism, TNF binds muchmore rap-idly to TNFR2 than TNFR1. When both receptors are in close proximity,the presence of TNFR2 increases the association rate of TNF with TNFR1thereby sensitising the cell to TNFR1-mediated cytotoxicity [29]. How-ever, TNFR2 is able to drive apoptosis independently of TNFR1, and co-operation between TNFR1 and TNFR2 in activating the NF-κB pathwayhas also been found to be additive rather than synergistic [32].

Since the TNFR1 receptor has been more strongly implicated in theactivation of pro-inflammatory signalling pathways than TNFR2, block-ade of TNFR1 appears a logical choice for therapy of chronic inflamma-tory diseases. Indeed, TNFR1 expressed on mesenchymal cells has beenshown to play an important role in arthritis in mice [1]. TNFR2 has alsobeen suggested to be involved in diverse processes that may be of ben-eficial function and thus its blockade could have deleterious implica-tions. These include diverse neuro-, cardio-, and osteo-protectiveeffects of TNFR2 suggested from experiments in TNFR−/− mice. Activa-tion of TNFR2 by TNF inhibits seizures [2] and attenuates cognitivedysfunction following brain injury [19]. TNFR2 appears not to affectmyocardial infarct size, but does promote survival followingmyocardialinfarction in mice [22]. Similarly, TNFR2 has been suggested to protectagainst myocardial ischaemia/reperfusion injury [10], and to reduce re-modelling andhypertrophy following heart failure [14]. Furthermore, inexperimental arthritis, TNFR2 has also been shown to protect againstjoint inflammation and erosive bone destruction through regulation ofosteoclastogenesis [27].

Given the wide range of protective functions of TNFR2, and the lackof a clear consensus on the signalling function of this receptor, wesought to identify unique physiological TNFR2-dependent signallingprocesses that are distinct from TNFR1. We have focused on TRAF2 pro-tein degradation, the TNFR2-dependent signalling event that is moststrongly implicated from previous studies in cell lines. To see if TRAF2degradation occurs in primary cells we have investigated the ability ofTNFR2 to induce this process in primary wild-type and TNFR1- orTNFR2-deficient macrophages. LPS-treated macrophages strongly ex-press TNF, and this allowed us to test the activation of the alternativeNF-κB pathway by endogenous membrane-bound TNF. Since thesecells express both TNF receptors we also investigated if, according tothe ‘ligand-passing’ model, TNFR2 sensitises primary macrophages foractivation of classical pro-inflammatory signalling pathways.

2. Materials and methods

2.1. Materials

Lipopolysaccharide (LPS) (TLR-grade) was purchased from AlexisBiochemicals (Exeter, UK). Macrophage colony stimulating factor(MCSF) and TNF (human cells were treated with human TNF; murinecells were treated with murine TNF) and human granulocyte/macrophage colony stimulating factor (GM-CSF) were from PeproTech(London, UK). Antagonistic anti-murine TNFR1 (55R-170) and anti-murine TNFR2 (TR75-32.4) monoclonal antibodies were from BioLegend(Cambridge, UK). Antibodies for western blotting (phospho(Thr180/Tyr182)-p38 MAPK, p38 MAPK, IκBα, TRAF2, p100/p52) were from CellSignalling (Danvers, Massachusetts, USA) or (tubulin) Sigma-Aldrich(St. Louis, Missouri, USA). Anti-TNFR1 (FAB225P (Clone #16803)) andanti-TNFR2 (FAB226P (Clone #22235)) phycoerythrin-conjugated

antibodies for flow cytometry were from R&D Systems (Abingdon,UK). General lab reagents were from Sigma-Aldrich.

2.2. Mice

Wild-type mice of C57BL/6 background were from Harlan Laborato-ries (Wyton, UK). TNFR1−/− and TNFR2−/− mice (C57BL/6 background)were maintained as heterozygotes and have been previously described[37]. All animal studies were ethically reviewed and carried out inaccordance with Animals (Scientific Procedures) Act 1986 and theGSK Policy on the Care, Welfare and Treatment of Animals.

2.3. Cell culture

Bone marrow-derived macrophages (BMDM) were isolated anddifferentiated as described previously by Hitti et al. [15]. Briefly,bone marrow cells were harvested from the femurs and tibiae of10–12 week-old mice and were incubated for 7 days in the presenceof MCSF (100 ng/ml). BMDMwere cultured in DMEM supplementedwith 10% (v/v) FCS, 2 mM L-glutamine (PAA, Yevil, UK), penicillin andstreptomycin and 0.1% (v/v) 2-mercaptoethanol (Life Technologies,Paisley, UK).

Centrifugal elutriation was used to isolate monocytes from humanperipheral blood mononuclear cells (PBMC). Briefly, human PBMCwere obtained by density centrifugation through ficoll/hypaque of aleukoreduction system chamber of apheresis instruments after routineplatelet collection. The resulting PBMC were centrifugally elutriated in1% (v/v) heat-inactivated FCS in RPMI in a Beckman JE6 elutriator(Beckman, High Wycombe, UK). The monocytes separated by thismethod were assessed for purity (≥75%) by flow cytometry (FACScan,BD, Oxford, UK) analysis of forward scatter and side scatter.

Monocytes were differentiated into macrophages by treatmentwithhuman GM-CSF (50 ng/ml) for 6 days or human MCSF (100 ng/ml) for5 days respectively. All cells were cultured in a humidified atmospherecontaining 5% CO2 at 37 °C.

2.4. Western blot

In experiments analysing phospho-p38 MAPK, IκBα and p100/p52,cells were lysed for 10 min on icewith≤200 μl of whole cell lysis buffer(50 mM Tris–HCl (pH 7.5), 250 mMNaCl, 3 mMEDTA, 3 mM EGTA, 1%(v/v) Triton X-100, 0.5% (v/v) NP-40, 10% (v/v) glycerol) supplementedwith protease inhibitor cocktail (1% v/v; Sigma-Aldrich), NaF (200 μM),NaVO3 (100 μM), microcystin (1 μM) and DTT (1 mM). Samples wereclarified by centrifugation at 16,000 ×g, for 10 min at 4 °C. For TRAF2degradation experiments cells were lysed on ice for 30 min with≤100 μl of a mild lysis buffer (30 mM Tris–HCl, 120 mM NaCl, 10%(v/v) glycerol, 1% (v/v) Triton X-100) supplemented with proteaseinhibitors (1% v/v) and DTT (1 mM). Samples were clarified by centri-fugation at 14,000 ×g for 30 min at 4 °C. Proteins in lysates (≥20 μg)were separated by electrophoresis through 10% (w/v) polyacrylamide-SDS gels and transferred to a polyvinylidene fluoride membrane(DuPont, Stevenage, UK). Membranes were blocked in 5% (w/v)dried-fat milk powder (Premier International Foods, Spalding, UK)and incubated in primary antibody and secondary antibody coupledto horseradish peroxidase (Dako, Glostrup, Denmark). Protein detectionwas carried out by enhanced chemiluminescence (GE Healthcare,Amersham, UK). Films were scanned with ImageScanner III (GEHealthcare). Densitometry of digital images with background sub-traction was performed using Quantity One software (BioRad) anddata analysis was performed using Excel (Microsoft). Error bars areshown for all data except for normalised values but in some cases aretoo small to be visible.

685G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

2.5. Assessment of TNF receptor expression by flow cytometry

Differentiated humanmacrophages were harvested by cell scraping.Cells were blocked with FcR block (Miltenyi Biotec, Bisley, UK), andtreated isotype control IgGs (BD) or stained with fluorochrome-labelled antibodies for 30 min at room temperature in PBS supplement-ed with 2% (v/v) FCS and 2 mM EDTA (final concentrations). Analysiswas carried out on a FACS Canto II (BD) and using FlowJo™ software(Treestar, Ashland, USA).

3. Results

3.1. Function of the TNFR2 in TNF-induced TRAF2 protein degradation: timeand concentration dependence

Signalling via TNFR2 induced by high concentrations of soluble TNFhas previously been shown to cause TRAF2 protein degradation[18,33], but this has not been reported in macrophages. To determineif TRAF2 degradation occurs in murine macrophages, and to identifywhich receptor is responsible for this with the use of receptor-deficient macrophages, it was firstly necessary to determine the timeand concentration dependence of the process in BMDM. For this cellswere treated with a high dose (100 ng/ml) of TNF and harvested at dif-ferent times. Cells were then lysed using a mild lysis buffer containing alow concentration of sodium chloride and Triton X-100 to isolate theTNF-regulated pool of TRAF2. Cell lysates were analysed for TRAF2 pro-tein bywestern blot. In the blot shown in Fig. 1A TRAF2degradationwas

0 1 2 4 6 8 16TNF (h)

TRAF2

tubulin

A

0 1 3 10 30 100TNF (ng/ml)

TRAF2

tubulin

B

TNF - + - + - +

WT R1 -/- R2 -/-

TRAF2

tubulin

C

kD

50

50

kDa

50

50

Fig. 1. Timecourse and TNF concentration and TNFR dependence of TRAF2 protein degradation ior B, at the concentrations indicated for 6 h. Cells were lysed with mild lysis buffer (See Mate(50 kDa). Dotted line indicates lanes removed from original blot for clarity of presentation. Reanalysis show mean TRAF2/tubulin protein ± SD for two independent experiments normadifferentiated BMDM were treated with TNF (30 ng/ml) for 4 h or left untreated and westernR2−/− cells. Data for R1−/− and R2−/− cells are indicated as hashed or black bars respecti

detected in BMDM at 1 h and was sustained until 16 h. TRAF2 degrada-tionwas observed up to 4 h TNF treatment in two separate experiments(as shown graphically in Fig. 1A). Next, we examined TRAF2 degrada-tion at 6 h over a range of TNF concentrations. The effect of TNF treat-ment was maximal at 30 and 100 ng/ml TNF (Fig. 1B).

3.2. TRAF2 protein degradation in BMDM is TNFR2-dependent

To investigate which receptor is required for TRAF2 degradation in-duced by TNF, wild-type, TNFR1−/− and TNFR2−/−, BMDMwere treat-ed with 30 ng/ml TNF for 4 h and TRAF2 degradation was analysed bywestern blot, as before. TNF treatment resulted in TRAF2 degradationin wild-type and TNFR1−/− BMDM but not in TNFR2−/− cells(Fig. 1C). Thus in macrophages, TNFR2, but not TNFR1, is necessary forligand-induced TRAF2 degradation, as reported previously in other celltypes.

In some blots, such as the ones shown in Fig. 1A and C, anti-TRAF2antibody detected TRAF2 protein as a doublet. The precise reason forthis is unclear but the data suggest that under certain gel running con-ditions two different forms of TRAF2 protein are resolved.

3.3. TNF induces TRAF2 degradation in human macrophages but notmonocytes

To investigate whether TRAF2 degradation occurs in human macro-phages in response to TNF, monocytes were isolated by elutriation anddifferentiated into M1- or M2-like macrophages using GM-CSF or

- + - + - + TNF

WT

0

0.2

0.4

0.6

0.8

1

1.2

0 1 2 4

TR

AF

2 / t

ubul

in

TNF (h)

0

0.2

0.4

0.6

0.8

1

1.2

TR

AF

2 / t

ubul

in

kDa

a

50

50

R1 -/- R2 -/-

00.20.40.60.8

11.2

0

30

100

TR

AF

2 / t

ubul

in

TNF (ng/ml)

nBMDM. A,Wild-type (WT) BMDMwere treatedwith TNF (100 ng/ml) for different timesrials and methods) and lysates analysed by western blot for TRAF2 (56 kDa) and tubulinlative molecular masses are shown in kilodaltons (kDa). Plots of data from densitometriclised for untreated cells. C, WT, TNFR1−/− (R1−/−) and TNFR2−/− (R2−/−) MCSF-blots performed as above. Plots as for (A) but normalised for untreated WT, R1−/− andvely.

686 G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

MCSF, respectively. Cells were then treatedwith a high dose human TNF(30 ng/ml) for 4 h, or left untreated. Undifferentiated monocytes weretreated in the same way. Both M1- and M2-like human macrophagesdisplayed TRAF2 degradation in response to TNF, but humanmonocytesdid not (Fig. 2A). Given that in murine macrophages TNFR2 is requiredfor TRAF2 degradation (Fig. 1C), we decided to assess the expression ofthe two TNF receptors on human macrophages and monocytes. Cellswere analysed by flow cytometry using antibodies to TNFR1 or TNFR2,and isotype control IgGs. This showed that monocytes differentiatedinto macrophages using either GM-CSF, or MCSF, express similar levelsof both TNF receptors, but humanmonocytes display greater TNFR1 ex-pression than macrophages (Fig. 2B). The predominance of TNFR1 asopposed to TNFR2 expression on human monocytes compared to mac-rophages (Fig. 2B) correlates with the apparent lack of TNF-inducedTRAF2 degradation in these cells (Fig. 2A).

3.4. Role of TNFR2 in TNF-induced tolerance

Since signalling by TNFR2 induces TRAF2 degradation, and becauseTRAF2 is required for the activation of MAPKs and the NF-κB pathway,we theorised that the phenomenon of tolerance of TNF signalling in-duced by TNF pre-treatment might involve TRAF2 protein degradation.Thismay identify TRAF2 degradation as a bonafide signalling eventwith

A

TRAF2

tubulin

GM-CSF MCSFMonocytes

TNF - + - + - +

BGM-CSF

M

M M

5

5

kD

Fig. 2. Induction of TRAF2 protein degradation by TNF in human monocytes differentiated intHuman monocytes were left untreated or treated with TNF (30 ng/ml) for 4 h. Monocytes wabove. As described in Fig. 1, cells were lysed and lysates were analysed by western blot anwere either left untreated, or differentiated into macrophages using GM-CSF or MCSF and TNTNFR2 antibodies, or using isotype control IgGs. Similar results were obtained for (A) and (B)

functional consequences. To test this, wild-type and TNFR2−/− BMDMwere treated with TNF (30 ng/ml) according to the regimens shownin Fig. 3A. Cells were left untreated at t = 0 and then treated with thesame dose of TNF at 6 h to induce signalling, or cells were treatedwith TNF at t = 0 to induce tolerance and, either left untreated, or treat-ed with TNF at 6 h. Some cells were left untreated and served as a con-trol. All cells were harvested at 6 h 15 min. Treatment of wild-type orTNFR2−/− BMDM with a single high dose of TNF (30 ng/ml) resultedin strong p38 MAPK phosphorylation (Fig. 3B; lane 3), consistent withits activation, but this was inhibited by a prior 6 h TNF treatment(Fig. 3B; lane 4). The ability of the initial TNF treatment to block signal-ling was not impaired in TNFR2−/− BMDM (Fig. 3B; compare lanes 4and 8). In both wild-type and TNFR2−/− BMDM treated with a singledose of TNF at t = 0, there was no detectable p38 MAPK activation fol-lowing 6 h 15 min of treatment (Fig. 3B; lanes 2 and 6), consistent withresolution of signalling. TRAF2 degradation occurred in wild-type cellsthat were treated with TNF at t = 0 (Fig. 3B; lanes 2 and 4), but not inTNFR2−/− cells treated in the same way (Fig. 3B; lanes 6 and 8), aspredicted.

In order to rule out the involvement of TRAF2 degradation in TNF-induced tolerance, similar experiments were performed in wild-typecells using a low concentration of TNF which fails to induce TRAF2 deg-radation. TNF-induced tolerance of p38 MAPK activation was also

0

0.2

0.4

0.6

0.8

1

1.2T

RA

F2

/ tub

ulin

- + - + - + TNF

GM-CSF MCSF Monocytes

MCSFM

Monocytes

M M

0

0

a

o macrophages using GM-CSF or MCSF but not in undifferentiated human monocytes. A,ere also differentiated into macrophages (MΦ) using GM-CSF or MCSF and treated as

d data for normalised values for untreated cells for each cell type plotted. B, MonocytesFR1 and TNFR2 protein expressions assessed by FACS analysis with anti-TNFR1 or anti-using cells from two separate donors.

A

B (30 ng/ml TNF)

C (1 ng/ml TNF)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

TR

AF

2 / t

ubul

in

0

0.2

0.4

0.6

0.8

1

1.2

1.4

P-p

38 /

tubu

lin

0

0.5

1

1.5

TR

AF

2 / t

ubul

in

0

0.2

0.4

0.6

0.8

1

1.2

P-p

38 /

tubu

lin

Time (h) - 0 6 0,6 - 0 6 0,6

WT

WT

R2 -/-

P-p38

tubulin

TRAF2

P-p38

tubulin

Time (h) - 0 6 0,6

Time (h) - 0 6 0,6

- 0 6 0,6

WT R2 -/-

- 0 6 0,6

- 0 6 0,6

WT R2 -/-

- 0 6 0,6

WT

- 0 6 0,6

WT

TRAF2

- 37

- 50

- 50

- 50

- 50

- 37

kDa

not stimulated cells lysed at 6h 15m

1, 5

2, 6

3, 7

4, 8

TNF added at 0 cells lysed at 6h 15m

cells lysed at 6h 15mTNF added at 6h

cells lysed at 6h 15mTNF added at 6hTNF added at 0

Fig. 3. TNFR2 is not required for tolerance of p38 MAPK activation induced by TNF in BMDM. A, Schematic of treatment protocol: Murine wild-type (WT) and TNFR2−/− (R2−/−) mac-rophageswere left untreated (Lanes 1, 5), treatedwith TNF (30 ng/ml) at t = 0 (Lanes 2, 6), and then further treatedwith TNF at t = 6 h (lanes 3, 7) or, treatedwith TNF at both 0 and6 h(Lanes 4, 8). Cells were not washed and culture medium was not changed in-between treatments. Cells were harvested and lysed at t = 6 h, 15 min. B, Wild-type and TNFR2−/− cellswere treated with TNF according to lane numbers as indicated in (A). C, Wild-type BMDM cells were treated according to lane numbers with 1 ng/ml TNF as in (A) or left untreated. In(A) and (B) cells were lysed with whole cell lysis buffer (see Materials and methods) and lysates were analysed by western blot for phospho-p38 MAPK (P-p38); (41 kDa), TRAF2 andtubulin. Plots show mean phospho-p38/tubulin protein ± SD and mean TRAF2/tubulin protein ± SD normalised to cells treated with TNF for 6 h, or untreated cells, respectively fromtwo independent experiments for (B) and two independent experiments, one of which was performed in duplicate for (C).

687G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

observed at a low concentration of TNF (1 ng/ml) (Fig. 3C; comparelanes 3 and 4) although in this case the inhibition of p38 MAPK activa-tion induced by TNF pre-treatment was slightly less than that observedusing 30 ng/ml TNF (Fig. 3B; lane 4). Western blot for TRAF2 proteinconfirmed that treatment of cells with 1 ng/ml TNF did not causeTRAF2 degradation (Fig. 3C; lanes 2 and 4). Overall the data suggestthat TNF-induced tolerance of p38 MAPK activation occurs indepen-dently of TRAF2 degradation.

3.5. Neither TNF, nor LPS activate the alternative NF-κB pathway in BMDM

We next sought to test the involvement of TNFR2 in another signal-ling event which has been previously reported to be specifically activat-ed by this TNF receptor, namely the activation of the alternative NF-κBpathway. Ligand-induced activation of TNFR2 has previously beenshown to activate the alternative NF-κB pathway by the processing ofp100 to p52 [26].

To test whether TNF induces p100 processing in BMDM, cells weretreated with a high concentration of TNF (20 ng/ml) for the timesshown and lysates were analysed for p100 and p52 protein. No

reduction in p100 protein or accumulation of p52 protein could be de-tected for TNF treatment between 1 and 4 h (Fig. 4A), indicative of aninability of TNF to induce p100 processing in murine macrophages.Since membrane-bound TNF has been suggested to be a more potentstimulus for the TNFR2 receptor than soluble TNF [12,26], cells werealso treated with LPS for 1–4 h to induce TNF expression and p100 pro-cessing was analysed as before. As for TNF, LPS treatment failed to in-duce activation of the alternative NF-κB pathway in BMDM (Fig. 4B).However, the antibody used to detect p100 and the cleaved product,p52, also detected a weakly stained band (indicated by *) representinga 40 kDa protein which was induced by both TNF and LPS (Fig. 4A andB). The identity of this protein is unclear but it may represent a cleavedspecies of p100, or p52, which is weakly induced following cellstimulation.

3.6. TNFR2 sensitises macrophages for ligand-induced activation of the p38MAPK and NF-κB pathways

Wewere also interested in examining the function of the TNFR2 re-ceptor in activating pro-inflammatory signalling pathways at different

A

BLPS (h) 0

p100

p52

tubulin

1 2 4

0

p100

p52

tubulin

TNF (h) 1 2 4- 150

- 75

- 37

- 50

- 50

kDa

- 150

- 75

- 37

- 50

- 50

kDa

*

*

Fig. 4. Neither TNF, nor LPS treatment, results in activation of the alternative NF-κB path-way in BMDM. A, MCSF-differentiated BMDM were treated with TNF (20 ng/ml) for thetimes shown, cells were lysed as in Fig. 1 and lysates analysed for p100 processing usingan antibody that recognises both p100 (97 kDa) and the processed form p52 (50 kDa).An additional band representing a 40 kDa species (indicated by *) was also detectedwhich was induced following cell stimulation. B, As for (A) but cells were treated withLPS (100 ng/ml).

688 G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

concentrations of TNF as this would allow any sensitisation of signallingdue to ligand-passing to TNFR1 to be detected.We focused on theNF-κBand p38MAPK pathways since these are both strongly implicated in theexpression of proteins of the inflammatory response. To investigate sig-nalling in primarymacrophages itwasfirstly necessary to check that thep38 MAPK and NF-κB pathways are activated at low concentrations aswell as at high concentrations of TNF. To test this, wild-type BMDMwere treated with different concentrations of TNF for 15 min and p38MAPK activation was assessed by western blot as before. Treatment ofwild-type BMDMwith concentrations of TNF as lowas 0.3 ng/ml resultedin activation of phosphorylation of p38 MAPK (Fig. 5).

To investigate the involvement of either TNFR1 or TNFR2, BMDMfrom TNFR1- and TNFR2-deficientmicewere used. TNF-induced activa-tion of signalling at lowdoses of TNF (≤3 ng/ml) and at doseswhich arereported to activate TNFR2most strongly (10 and 30 ng/ml) was exam-ined. Little activation of p38 MAPK occurred in TNFR1−/− BMDM(Fig. 5). Surprisingly, at low doses, TNF-induced signalling was also im-paired in TNFR2−/− cells inwhich the ligand concentration dependenceof the activation of p38 MAPK was increased ~10-fold compared towild-type cells (Fig. 5). The TNF concentration dependence of IκBα deg-radation in both wild-type and knockout cells was identical to that ob-served for p38 MAPK activation (Fig. 5). Thus at high doses of solubleTNF, TNFR1 is the main receptor required for activation of signallingand the presence of TNFR2 is not required for activation. However, atlower concentrations of TNF, TNFR2 is essential for full activation ofsignalling.

A possible caveat in the above interpretation is that knockout cellsmay adapt in response to receptor deficiency and the results could bemisleading. To address this we also investigated the signalling functionof TNFR1 and TNFR2 using receptor-specific monoclonal antibodies.Wild-type BMDM were pre-treated with an IgG control, anti-TNFR1mAb, anti-TNFR2 mAb and a murine form of soluble TNFR2 coupled toan Fc region (mTNFR2-Fc) as a control to block signalling by both

receptors. Cells were then stimulated with a low concentration of TNF(0.3 ng/ml) or left untreated. Anti-TNFR1 mAb strongly inhibited TNF-induced p38 MAPK signalling and IκBα degradation (Fig. 6A). In agree-ment with the result in TNFR2−/− cells, the anti-TNFR2 mAb alsostrongly inhibited signalling at a low concentration of TNF (Fig. 6A).Similar experiments were performed at a high concentration of TNF(10 ng/ml), and consistent with the results from receptor-deficientmacrophages, the anti-TNFR1 mAb and mTNFR2-Fc blocked signalling,albeit rather weakly, but anti-TNFR2 mAb did not (Fig. 6B). Thus athigh TNF concentrations, TNFR1 is the dominant receptor which acti-vates signalling in the absence of TNFR2, whereas both receptors are es-sential for the activation of signalling at lower concentrations of TNF.

4. Discussion

TNFR1 is known to activate the NF-κB and MAPK pathways in re-sponse to TNF in a wide range of cell types. In this study, induction byTNF of p38MAPK phosphorylation and IκBα degradationwas completelyblocked in TNFR1−/− macrophages at all concentrations of TNF used,underscoring a dominant role for this receptor and showing that TNFR2cannot significantly activate these pathways independently of TNFR1.The literature suggests that the role of TNFR2 in signalling is more com-plex. In macrophages, we find that at high concentrations of its ligand,TNFR2 is needed for TRAF2 protein degradation, whereas at low ligandconcentrations, TNFR2 serves to sensitise the cells for activation of pro-inflammatory signalling. Our findings in primary cells differ from thosein immortalised macrophage lines in which TNFR2 deficiency almost en-tirely blocked signalling induced by a wide range of TNF concentrationsthat encompassed the range used in this study [23]. A possible explana-tion is that, unlike in primary TNFR2-deficient macrophages, TNFR1 ex-pression was reduced in the TNFR2-deficient line following extendedpassaging. Our observation that even sub-ng/ml concentrations of TNFare sufficient to strongly activate the p38 MAPK and NF-κB pathways incells in culture suggests that our findings are very relevant to any situa-tion in vivo where small amounts of TNF are present.

The majority of previous studies reporting TNFR2-mediated TRAF2degradation have utilised cell lines and receptor overexpression, al-though during the course of this work it has also been reported tooccur in primary CD8+ T cells [30]. Treatment of both human andmurine primary macrophages with high concentrations of TNF causeda reduction in TRAF2 protein. The loss of TRAF2 protein in murinemacrophages was exclusively TNFR2-dependent as the TNF-inducedreduction in TRAF2 protein was blocked by TNFR2, and not by TNFR1deficiency. TRAF2 degradation was shown to occur in human macro-phages, but the very high concentrations of soluble TNF needed formaximal TRAF2 degradation (30 ng/ml) in BMDM suggest that thisprocess may be most relevant to physiological settings where localisedTNF expression is extremely high or where presence of membrane TNFwould favour TNFR2 signalling.

TRAF2 plays an important role for the activation of themajor signal-ling pathways downstreamof TNF [3] including the p38MAPK pathway[5]. The lack of a requirement for TNFR2-mediated TRAF2 degradation inTNF-induced tolerance of activation of p38MAPK suggests that, in mac-rophages, TRAF2 degradation may serve some other purpose, such asthe previously reported inhibition of the NF-κB pathway [11]. We failedto detect any evidence for activation of the alternativeNF-κB pathway inresponse to soluble TNF or LPS in macrophages. The results suggest thatunder the conditions used either LPS fails to induce a significant amountof membrane-bound TNF in macrophages, or endogenous membrane-bound TNF does not activate the alternative NF-κB pathway inmacrophages.

To investigate TNF-induced gene expression, the cytokine is fre-quently reported to be used to treat cells at a concentration of≥10 ng/mlwhich in BMDM lieswell above the concentration thatmax-imally activates signalling. Thus the important contribution of TNFR2 tosignalling at sub-ng/ml concentrations of TNF suggests that many

0.3

1TNF (ng/ml) 10 303

tubulin

P-p38

WT R1 -/- R2 -/-

0

I B

0.3

1 10 3030 0.3

1 10 3030

0

0.2

0.4

0.6

0.8

1

1.2

1.4P

-p38

/ tu

bulin

0.3

1TNF (ng/ml) 10 303

WT R1 -/- R2 -/-

0 0.3

1 10 3030 0.3

1 10 3030

0

0.5

1

1.5

2

IB

/ tub

ulin

0.3

1TNF (ng/ml) 10 303

WT R1 -/- R2 -/-

0 0.3

1 10 3030 0.3

1 10 3030

- 37

- 50

- 37

kDa

Fig. 5. TNFR1 is the major receptor needed for TNF-induced p38 MAPK and NF-κB pathway activation in BMDM but TNFR2 is also required at a low concentrations of TNF. Murine wild-type, TNFR1−/− and TNFR2−/−BMDMwere treatedwith the concentrations of TNF indicated for 15 min. Similar results were obtained in two separate experiments.Western blot analysisof BMDM lysates (prepared as in Fig. 3) for phospho-p38 MAPK, IκBα (35 kDa) and tubulin is shown. Plots show mean phospho-p38/tubulin ± SD and mean IκBα/tubulin ± SD fromduplicate blots normalised to values for 30 ng/ml TNF-treated wild-type cells, or untreated wild-type cells, respectively.

689G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

previous studies may have underestimated the importance of TNFR2 inthe induction and regulation of gene expression in response to TNF. Thismay be particularly relevant for cell types such as endothelial cellswhich express both TNF receptors and in which TNF strongly inducesthe expression of a wide range of proteins, including inflammatorycytokines and adhesion molecules. It is noted that TNFR2 has beenpreviously reported to play an important role in tissue factor expressionin endothelial cells treated with low but not high concentrations of TNF[8].

Our data also suggest that under situations where the concentrationof TNF is high, such as in inflammation, TNFR2may have little role in theactivation of pro-inflammatory signalling. However, where TNF levelsare low, TNFR2 may be important for the initiation of inflammationand its subsequent autoregulated resolution. It is tempting to speculatethat as opposed to chronic inflammation in which resolution may bedysregulated, TNFR2-dependent signalling induced by small amountsof TNF may activate signalling, its normal resolution, and the associatedanti-inflammatory effects, thus providing a additional explanationfor the previously reported anti-inflammatory functions of TNFR2in mice. It has been reported recently that TNFR2-mediated TRAF2

degradation is needed for resolution of NF-κB pathway activation in293 cells [28].

5. Conclusions

In macrophages high concentrations of soluble TNF induce TRAF2protein degradation via TNFR2. TRAF2 degradation does not appear tobe required for TNF-induced tolerance of p38 MAPK activation andTNF does not appear to induce TRAF2 degradation in humanmonocytes.However, in addition to its previously reported anti-inflammatory func-tions inmice, inmacrophages TNFR2may also act as an auxiliary recep-tor to TNFR1 to enable cells to activate pro-inflammatory signallingpathways in response to low levels of TNF which are likely to be withinthe range found in vivo.

Acknowledgements

Weare grateful to GSK, Arthritis ResearchUK, and theKennedy Insti-tute of Rheumatology Trust for funding and members of the Joint

0

0.2

0.4

0.6

0.8

1

1.2

1.4

P-p

38 /

tubu

lin

0

0.5

1

1.5

IB

/ tub

ulin

A (0.3 ng/ml TNF)

0

0.2

0.4

0.6

0.8

1

1.2

P-p

38 /

tubu

lin

0

0.5

1

1.5

2

IB

/ tub

ulin

B (10 ng/ml TNF)

TNF

tubulin

P-p38

- +

-- IgG

I B

+ ++ +

mA

bR

1

mA

b-R

2

mT

NF

R2-

Fc

tubulin

P-p38

I B

TNF - +

-- IgG

+ ++ +

mA

b-R

1

mA

bR

2

mT

NF

R2-

Fc

TNF - +

IgG

+ ++ +

mA

bα

R1

mA

bα

R2

mT

NF

R2-

Fc--Ab

TNF - +

IgG

+ ++ +

mA

bα

R1

mA

bα

R2

mT

NF

R2-

Fc--Ab

TNF - +

IgG

+ ++ +

mA

bα

R1

mA

bα

R2

mT

NF

R2-

Fc--Ab

bAbA

TNF - +

IgG

+ ++ +

mA

bα

R1

mA

bα

R2

mT

NF

R2-

Fc--Ab

- 37

- 50

- 37

kDa

- 37

- 50

- 37

kDa

Fig. 6. Effect of TNFR1 and TNFR2 blockade by receptor-specific monoclonal antibodies on p38 MAPK and NF-κB signalling at low and high concentrations of TNF. A, Wild-type macro-phages were pre-treated for 1 h with isotype control IgG (10 μg/ml), anti-TNFR1 mAb (10 μg/ml), anti-TNFR2 mAb (10 μg/ml) or mTNFR2-Fc (1 nM) as a control to block signallingby both receptors, and then further treated with TNF (0.3 ng/ml) for 15 min. B, As for (A) but using 10 ng/ml TNF. Western blots are shown together with plots for mean phospho-p38/tubulin ± SD and mean IκBα/tubulin ± SD from two independent experiments normalised to values for cells treated with TNF only, or untreated cells, respectively.

690 G. Ruspi et al. / Cellular Signalling 26 (2014) 683–690

Research Committee for useful discussions throughout the course of thisproject. We thank Simon Arthur for reading the manuscript.

References

[1] M. Armaka, M. Apostolaki, P. Jacques, D.L. Kontoyiannis, D. Elewaut, G. Kollias, J. Exp.Med. 205 (2008) 331–337.

[2] S. Balosso, T. Ravizza, C. Perego, J. Peschon, I.L. Campbell, M.G. De Simoni, A. Vezzani,Ann. Neurol. 57 (2005) 804–812.

[3] V. Baud, M. Karin, Trends Cell Biol. 11 (2001) 372–377.[4] S. Bertok, M.R. Wilson, A.D. Dorr, J.O. Dokpesi, K.P. O'Dea, N. Marczin, M. Takata, Am.

J. Physiol. Lung Cell. Mol. Physiol. 300 (2011) L781–L789.[5] J.L. Cannons, Y. Choi, T.H. Watts, J. Immunol. 165 (2000) 6193–6204.[6] G.B. Chainy, S. Singh, U. Raju, B.B. Aggarwal, J. Immunol. 157 (1996) 2410–2417.[7] X. Chen, J.J. Oppenheim, Immunology 133 (2011) 426–433.[8] M. Clauss, M. Grell, C. Fangmann, W. Fiers, P. Scheurich, W. Risau, FEBS Lett. 390

(1996) 334–338.[9] S.L. Erickson, F.J. de Sauvage, K. Kikly, K. Carver-Moore, S. Pitts-Meek, N. Gillett,

K.C. Sheehan, R.D. Schreiber, D.V. Goeddel, M.W. Moore, Nature 372 (1994)560–563.

[10] M.P. Flaherty, Y. Guo, S. Tiwari, A. Rezazadeh, G. Hunt, S.K. Sanganalmath, X.L. Tang,R. Bolli, B. Dawn, J. Mol. Cell. Cardiol. 45 (2008) 735–741.

[11] M. Fotin-Mleczek, F. Henkler, D. Samel, M. Reichwein, A. Hausser, I. Parmryd, P.Scheurich, J.A. Schmid, H. Wajant, J. Cell Sci. 115 (2002) 2757–2770.

[12] M. Grell, E. Douni, H. Wajant, M. Lohden, M. Clauss, B. Maxeiner, S. Georgopoulos,W.Lesslauer, G. Kollias, K. Pfizenmaier, et al., Cell 83 (1995) 793–802.

[13] H. Hacker, M. Karin, Sci. STKE 2006 (2006) re13.[14] T. Hamid, Y. Gu, R.V. Ortines, C. Bhattacharya, G.Wang, Y.T. Xuan, S.D. Prabhu, Circu-

lation 119 (2009) 1386–1397.[15] E. Hitti, T. Iakovleva, M. Brook, S. Deppenmeier, A.D. Gruber, D. Radzioch, A.R. Clark,

P.J. Blackshear, A. Kotlyarov, M. Gaestel, Mol. Cell. Biol. 26 (2006) 2399–2407.[16] E.Y. Kim, H.S. Teh, J. Immunol. 173 (2004) 4500–4509.

[17] J.M. Kyriakis, J. Avruch, Physiol. Rev. 81 (2001) 807–869.[18] X. Li, Y. Yang, J.D. Ashwell, Nature 416 (2002) 345–347.[19] L. Longhi, F. Ortolano, E.R. Zanier, C. Perego, N. Stocchetti, M.G. De Simoni, Acta

Neurochir. Suppl. 102 (2008) 409–413.[20] F. Mackay, H. Loetscher, D. Stueber, G. Gehr, W. Lesslauer, J. Exp. Med. 177 (1993)

1277–1286.[21] S.M. McFarlane, G. Pashmi, M.C. Connell, A.F. Littlejohn, S.J. Tucker, P. Vandenabeele,

D.J. MacEwan, FEBS Lett. 515 (2002) 119–126.[22] Y. Monden, T. Kubota, T. Inoue, T. Tsutsumi, S. Kawano, T. Ide, H. Tsutsui, K.

Sunagawa, Am. J. Physiol. Heart Circ. Physiol. 293 (2007) H743–H753.[23] A. Mukhopadhyay, J. Suttles, R.D. Stout, B.B. Aggarwal, J. Biol. Chem. 276 (2001)

31906–31912.[24] E.M. Paleolog, S.A. Delasalle,W.A. Buurman,M. Feldmann, Blood 84 (1994) 2578–2590.[25] F. Porteu, C. Hieblot, J. Biol. Chem. 269 (1994) 2834–2840.[26] H. Rauert, A. Wicovsky, N. Muller, D. Siegmund, V. Spindler, J. Waschke, C. Kneitz, H.

Wajant, J. Biol. Chem. 285 (2010) 7394–7404.[27] K. Redlich, B. Gortz, S. Hayer, J. Zwerina, N. Doerr, P. Kostenuik, H. Bergmeister, G.

Kollias, G. Steiner, J.S. Smolen, et al., Am. J. Pathol. 164 (2004) 543–555.[28] M. Rodriguez, L. Cabal-Hierro, M.T. Carcedo, J.M. Iglesias, N. Artime, B.G. Darnay, P.S.

Lazo, J. Biol. Chem. 286 (2011) 22814–22824.[29] L.A. Tartaglia, D. Pennica, D.V. Goeddel, J. Biol. Chem. 268 (1993) 18542–18548.[30] Y.C. Twu, M.R. Gold, H.S. Teh, Eur. J. Immunol. 41 (2011) 335–344.[31] H. Wajant, P. Scheurich, FEBS J. 278 (2011) 862–876.[32] T. Weiss, M. Grell, B. Hessabi, S. Bourteele, G. Muller, P. Scheurich, H. Wajant,

J. Immunol. 158 (1997) 2398–2404.[33] C.J. Wu, D.B. Conze, X. Li, S.X. Ying, J.A. Hanover, J.D. Ashwell, EMBO J. 24 (2005)

1886–1898.[34] Y. Zhao, D.B. Conze, J.A. Hanover, J.D. Ashwell, J. Biol. Chem. 282 (2007) 7777–7782.[35] L. Zheng, G. Fisher, R.E. Miller, J. Peschon, D.H. Lynch, M.J. Lenardo, Nature 377

(1995) 348–351.[36] Z. Zhou, M.C. Connell, D.J. MacEwan, Cell. Signal. 19 (2007) 1238–1248.[37] J.J. Peschon, D.S. Torrance, K.L. Stocking, M.B. Glaccum, C. Otten, C.R. Willis, K.

Charrier, P.J. Morrissey, C.B. Ware, K.M. Mohler, J Immunol. 160 (1998) 943–952.