LPS antagonism of TGF-β signaling results in prolonged survival and activation of rat primary microglia


  • Kendall Mitchell,

    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    Search for more papers by this author
  • Jill P. Shah,

    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    Search for more papers by this author
  • Lyubov V. Tsytsikova,

    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    Search for more papers by this author
  • Ashley M. Campbell,

    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    Search for more papers by this author
  • Kwame Affram,

    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    2. Program in Neuroscience, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    Search for more papers by this author
  • Aviva J. Symes

    Corresponding author
    1. Department of Pharmacology, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    2. Program in Neuroscience, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA
    • Address correspondence and reprint requests to Aviva J. Symes, Department of Pharmacology, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Road, Bethesda, MD 20814, USA. E-mail: aviva.symes@usuhs.edu

    Search for more papers by this author


Accumulating evidence indicates that activated microglia contribute to the neuropathology involved in many neurodegenerative diseases and after traumatic injury to the CNS. The cytokine transforming growth factor-beta 1 (TGF-β1), a potent deactivator of microglia, should have the potential to reduce microglial-mediated neurodegeneration. It is therefore perplexing that high levels of TGF-β1 are found in conditions where microglia are chronically activated. We hypothesized that TGF-β1 signaling is suppressed in activated microglia. We therefore activated primary rat microglia with lipopolysaccharide (LPS) and determined the expression of proteins important to TGF-β1 signaling. We found that LPS treatment decreased the expression of the TGF-β receptors, TβR1 and TβR2, and reduced protein levels of Smad2, a key mediator of TGF-β signaling. LPS treatment also antagonized the ability of TGF-β to suppress expression of pro-inflammatory cytokines and to induce microglial cell death. LPS treatment similarly inhibited the ability of the TGF-β related cytokine, Activin-A, to down-regulate expression of pro-inflammatory cytokines and to induce microglial cell death. Together, these data suggest that microglial activators may oppose the actions of TGF-β1, ensuring continued microglial activation and survival that eventually may contribute to the neurodegeneration prevalent in chronic neuroinflammatory conditions.


We propose that LPS/TLR4 signaling interferes with key components in the TGF-β1 signaling pathway in primary microglia. As a result, the ability of TGF- β1 to exert anti-inflammatory effects is significantly reduced leading to prolonged survival of classically activated microglia. Identifying the mechanisms by which TGF- β1 signaling is targeted during microglia activation may yield novel strategies for deactivating chronically activated microglia.

Abbreviations used

bone morphogenetic protein-2


central nervous system


damage associated molecular patterns


interleukin 1 beta




pathogen associated molecular patterns


quantitative polymerase chain reaction


real-time polymerase chain reaction


transforming growth factor-beta 1


toll like receptor 4


tumor necrosis factor alpha


transforming growth factor beta receptor-1


transforming growth factor beta receptor-2

Microglia, the resident macrophages of the central nervous system (CNS), perform critical roles in maintaining CNS homeostasis (Kettenmann et al. 2011). In the normal brain, microglia contain ramified processes that are used for communicating and sensing danger signals. Microglia become activated once membrane and cytosolic receptors detect damage or pathogen-associated molecular patterns (Akira et al. 2006; Gordon 2007; Takeuchi and Akira 2010). Receptor activation prompts the release of toxic pro-inflammatory molecules and the phagocytosis of invading pathogens, foreign material and debris. Importantly, activated microglia and other affected cells in the CNS also release a variety of anti-inflammatory molecules (Colton 2009), which can alter the activation states of microglia. Indeed, peripheral macrophages, can be converted from classically activated states (M1; also referred as pro-inflammatory) to alternative activated (M2) or acquired deactivated (M3) states depending on molecules released in the microenvironment (Stout and Suttles 2004; Mosser and Edwards 2008). M2 and M3 macrophages/microglia produce higher levels of anti-inflammatory molecules (Brodie et al. 1998; van Rossum et al. 2008) and have significant roles in repair and resolution of inflammation (Mosser and Edwards 2008; Colton 2009; Kigerl et al. 2009). Great interest centers on understanding the mechanisms required for converting microglial activation from M1 to M2/M3 with the hope that this would bring therapeutic advantage. Conversely, perturbations in this conversion are predicted to result in prolonged release of toxic molecules from M1 microglia thereby contributing to neurodegeneration.

Transforming growth factor-beta 1 (TGF-β1) is a cytokine which serves pleotropic roles in various physiological and pathological settings (Blobe et al. 2000). In microglia, TGF-β1 is thought to act as an anti-inflammatory cytokine. In combination with IL4 and IL10, TGF-β1 drives macrophages/microglia into M2 and M3 states, respectively (Mosser and Edwards 2008; Zhou et al. 2012). Furthermore, TGF-β1 down-regulates cytokine production stimulated by microglial activators such as lipopolysaccharide (LPS) (Lodge and Sriram 1996; Kim et al. 2004). TGF-β1 signaling is mediated by transmembrane serine/threonine protein kinases (Heldin et al. 1997; Massague and Wotton 2000). Canonical TGF-β signaling consists of ligand binding to the TGF-β receptor TβR2, which recruits and phosphorylates TβR1. TGF-β receptor phosphorylation leads to the phosphorylation and subsequent translocation of Smad2 and Smad3, together with the co-Smad Smad4 to the nucleus to regulate gene expression. Importantly, TGF-β1 activates other pathways including Erk, c-Jun N-terminal kinase (JNK), p38, and PI3 kinase pathways and can also induce Nuclear Factor-κB (NF-κB) and Rho GTPase signaling in a Smad-independent manner (Moustakas and Heldin 2005; Zhang 2009; Zhang et al. 2009). Little is known about the signaling mechanisms through which TGF-β1 regulates microglial activity.

In vitro studies have demonstrated increased production/secretion of TGF-β1 from activated microglia (Smith and Hale 1997). This secreted TGF-β1 can work in an autocrine fashion to reduce microglia activation (Spittau et al. 2013). Elevated TGF-β1 levels have also been detected in neuropathological conditions such as Alzheimer's disease, experimental autoimmune encephalomyelitis, ischemia, and traumatic brain injury (Krupinski et al. 1996; Morganti-Kossman et al. 1997; Tarkowski et al. 2002; Lanz et al. 2010). The elevated levels of TGF-β1 in these neurological conditions are inconsistent with the continued chronic microglial activation detected. Although it cannot be excluded that the TGF-β1 expressed is insufficient for deactivating microglia, it is also possible that activated microglia are less responsive to TGF-β1. Indeed, pro-inflammatory molecules have been shown to antagonize TGF-β1 signaling in numerous cell types including macrophages (Kim and Kim 2011). These effects appear to be cell and context dependent, as some of the same pro-inflammatory molecules can accentuate TGF-β1 signaling (Seki et al. 2007).

In this study, we sought to determine whether TGF-β1 signaling is antagonized in activated microglia. We treated microglia with LPS to stimulate microglia into their M1 classically activated state. We then compared expression of key genes involved in TGF-β1 signaling with that in ramified microglia. In addition, we examined the ability of TGF-β1 to transduce Smad2/3 phosphorylation in ramified versus LPS-activated microglia. Our data show that LPS drastically reduced TβR1 and TβR2 expression in primary microglia and significantly decreased levels of Smad2 protein. LPS also significantly inhibited the ability of TGF-β1 to down-regulate expression of cytokines and reduced TGF-β1-mediated cell death. These data suggest for the first time that TGF-β1 signaling is altered in activated microglia and may shed light into why microglia remain chronically activated even in the presence of high levels of TGF-β1.

Materials and methods


Recombinant human bone morphogenetic protein-2 (BMP-2) and recombinant human Activin-A were purchased from R & D Systems (Minneapolis, MN, USA) and human TGF-β1 from PeproTech (Rocky Hill, NJ, USA). The TβR1 kinase inhibitor, SB505124, was purchased from Sigma (St. Louis, MO, USA). Primers were synthesized by Integrated DNA technologies (Coralville, Iowa, USA). Dulbecco's modified Eagle's medium (15-013-CV), Hanks' balanced salt solution (21-022-CV) and antibiotic-antimycotic solution (30-004-CL) were purchased from Corning Cellgro (Manassas, VA, USA). Donor equine and fetal bovine serum were purchased from Hyclone Thermo Scientific (Waltham, MA, USA). Complete Mini, protease inhibitor cocktail tablets (04 693 124 001) were purchased from Roche (Nutley, NJ, USA).

Primary microglial culture

All animal care and procedures were approved by the Institutional Animal Care and Use Committee of the Uniformed Services University and performed in accordance with the ARRIVE guidelines. Sasco/Sprague–Dawley rats were purchased from Charles River Laboratories (Frederick, MD, USA) and housed in USU animal facility. All efforts were made to minimize both animal numbers and distress within the experiments. Mixed glial cultures were isolated from P2 Sprague–Dawley rat pups (male and female). Brains were removed from the pups, the cerebral hemispheres were dissected out, their meninges removed and the cortices placed in culture medium (Dulbecco's modified Eagle's medium, containing 10% fetal bovine serum, 1% glutamax and 1% antibiotic-antimycotic). Cortices were triturated twice with a 10 mL pipette, followed by successive triturations with an 18G needle (3X), a 22G needle (3X), and finally a 25G needle (2X). The resulting suspension was filtered using a 70 μm mesh before pelleting at 168 g for 10 min at 23°C. Cells were resuspended in culture medium and seeded into T75 flasks at a density of one cortex per flask. Cells were incubated in standard conditions (37°C; 5% CO2) and medium was replaced every 3 or 4 days. On day 14 or 15, flasks were placed on a MaxQ 2000 orbital shaker (Thermo Scientific, Rockford, IL, USA) within the incubator and shaken for 1 h at 200 rpm. The medium, containing detached microglia, was collected and centrifuged at 671 g for 5 min at 23°C. Cells were resuspended with microglial complete culture medium (Dulbecco's modified Eagle's medium, 10% normal horse serum, 1% glutamax, and 1% streptomycin/penicillin) and transferred to uncoated plates at a density of 2.5 × 105 cells/mL. Immunohistochemical analysis of the plated microglia with the microglial markers ED-1 or Iba-1 indicated that greater than 98% of the cells were microglia. Conversely, immunohistochemical analysis using the astrocyte marker, Glial fibrillary acidic protein (GFAP), indicated that less than 1% of the cells were astrocytes. Cells were grown in plates for 24–48 h before treatment with the indicated cytokine, the inhibitor SB505124 or LPS, whose stocks were diluted in complete medium. Control samples were given equivalent volumes of complete medium. In the SB505124 experiments, equivalent amounts of dimethylsulfoxide, which was used to dilute the drug, was added to complete medium in control samples. We did not detect any effects of dimethylsulfoxide on microglia.


Primary microglia were fixed with ice cold 4% paraformaldehyde. Cells were permeabilized with phosphate-buffered saline (PBS)/0.1% Triton X-100 and blocked with 2% fish skin gelatin (FSG)/PBS. Cells were incubated with primary antibody (in 1% FSG) for 2 h at 23°C or overnight at 4°C. Cells were blocked again (2% FSG) briefly (15 min) before incubation with the appropriate secondary antisera for 1 h (1 : 500). Cells were washed and stored in PBS (4°C) until images were obtained. The following antisera were used: anti-GFAP, rabbit polyclonal (1 : 1000, DAKO Z0334; DAKO, Carpinteria, CA, USA) for astrocytes; anti-ED-1, mouse monoclonal (1 : 500, Millipore MAB1435; Millipore, Billerica, MA, USA) for microglia; and anti-Iba-1, rabbit polyclonal (1 : 1000, 019-19741; Wako, Richmond, VA, USA) for microglia. Fluorescent labeling with secondary antibodies was performed using Alexa Fluor 568 goat anti-mouse and Alexa Fluor 488 goat anti-rabbit (1:500, Invitrogen A11031 and A-11034, respectively; Invitrogen, Carlsbad, CA, USA). Images (10 and 20X) were acquired using a Nikon Diaphot 300 inverted microscope (Nikon, Linthicum Heights, MD, USA) attached to a Retiga 1300i CCD camera (Qimaging, Surrey, BC, Canada) using Qcapture software.

Quantification of cell number

Microglia were fixed (4% paraformaldehyde) and nuclei stained with 4′,6-diamidino-2-phenylindole (DAPI). Images were obtained at 5× magnification, which acquired nearly half a well. All nuclei were counted in each image utilizing Image J software (National Institutes of Health, Bethesda, MD, USA). Cells were counted in a minimum of four wells per treatment.

LDH assay

Primary microglia (50 000 cells/well) were plated in 96-well plates and cultured in full culture medium. One day later, cells were treated with or without TGF-β(1 ng/mL) for 24 h. Lactate dehydrogenase (LDH) released into the medium was measured with the Pierce LDH Cytotoxicity Assay Kit (Thermo Scientific) according to the manufacturer's instructions. Briefly, 50 μL of microglial conditioned medium was added to 50 μL of substrate in reaction medium and incubated for 30 min at 23°C. The reaction was terminated with Stop solution and the absorbance read at 490 nm.

RNA isolation and Quantitative PCR

RNA was isolated from microglial cultures using Trizol reagent (Life Technologies, Carlsbad, CA, USA) according to manufacturer's instructions. cDNA was synthesized using Superscript III (Life Technologies) and QPCR performed using SYBR Green (Qiagen) in a CFX96 real time system (Bio-Rad, Hercules, CA, USA) with the primers listed in Table 1. Target RNA expression levels were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Relative changes in mRNA expression levels between control and treated samples were calculated by the delta threshold cycle (∆∆Ct) method (Susarla et al. 2011) using Bio-Rad CFX Manager 2.0. A minimum of four independent samples was used per group and the effects were observed in two to five independent experiments. Heat maps were generated to show effects of treatment in individual experiments and bar graphs were generated to show significance.

Table 1. Primers used for qPCR

Western blots

Total protein was isolated from treated microglia by 20 s sonication on ice in radioimmunoprecipitation lysis buffer containing phosphatase and protease inhibitors. Loading buffer containing sodium dodecyl sulfate and β-mercapethanol were added to the lysate and samples were boiled for 8 min before storing at −80°C. Equal amounts of protein (7.5–15 μg) were loaded onto sodium dodecyl sulfate-polyacrylamide gels. Gels were transferred to nitrocellulose membranes and membranes were probed with various primary antisera before probing with the appropriate horseradish peroxidase-linked secondary antisera. Blots were developed with Supersignal enhanced chemiluminescence reagent (Thermo Scientific). Chemiluminescence signals were detected using a Fuji LAS-3000 image acquisition system (Fuji, Stamford, CT, USA) equipped with a cooled CCD camera. Blots were reprobed with GAPDH as a loading control. Immunoreactivity for each protein was quantified using Fuji Image analysis software (Multi Gauge). The following antisera were used for western blots: inducible nitric oxide synthase (iNOS) (1 : 1000, 610333; BD Biosciences, San Jose, CA, USA); GAPDH (1 : 5000, 9545; Sigma); pSmad2 (1 : 1000, Ab3849; Millipore); Smad2 (1 : 1000, 5339S; Cell Signaling, Danvers, MA, USA); pSmad3 (1 : 1000, 9520S; Cell Signaling); Smad2/3 (1 : 1000, 07408; Upstate, Billerica, MA, USA); pSmad2/3 (1 : 1000, 8828; Cell Signaling); Smad3 (1 : 1000, 06-920; Upstate); pSmad1/5/8 (1:1000, 9511S; Cell Signaling) and Smad1/5/8/9 (1:1000, Ab72504; Abcam, Cambridge, MA, USA). The following secondary antisera were used: horseradish peroxidase-linked anti-rabbit IgG (1 : 2500, 7074; Cell Signaling).

Statistical analysis

All data are expressed as mean ± SEM. Data were analyzed using Prism software (Graphpad, La Jolla, CA, USA) by one-way or two-way anova or by Student's t-test. Specifically, in experiments where the effects of LPS ± individual TGF-β cytokines were examined the following statistical analysis were performed: comparisons to examine microglia activity in samples treated individually with one of the TGF-β cytokines (TGF-β1, Activin-A, or BMP-2) versus their corresponding co-treated conditions with LPS were performed using Student's t-test; comparisons to determine whether TGF-β cytokines alter microglia activity compared to control were performed using one-way anova followed by a Bonferroni correction; comparisons to determine whether co-treatment (LPS plus TGF-β cytokine) alters microglia activity as compared to LPS alone were performed using one-way anova followed by a Bonferroni correction. In addition, one-way anova followed by a Bonferroni correction was used to test for significance when examining whether LPS alters microglia activity versus control. This analysis was also used when testing the effects of TGF-β1 or the ALK5 inhibitor, SB505124, on gene expression in microglia. To examine daily effects of TGF-β1 on microglia cell survival, two-way anova followed by a Sidak correction was performed. For western analyses, comparisons were made between each condition using one-way anova followed by a Tukey correction. Finally, Student's t-test was performed in all other comparisons where the effect of a single treatment involving one time point was analyzed. A p-value of less than 0.05 was considered statistically significant.


We first verified that our primary microglial cultures responded to TGF-β1 and BMP-2, another member of the family of TGF-β cytokines. Whereas TGF-β1 treatment (1 ng/mL for 30 min) resulted in increased phosphorylation of Smad2 (Fig. 1a), we did not detect an appreciable increase in pSmad3 (Fig. 1b). BMP-2 treatment (100 ng/mL for 1 h) resulted in an increase in the phosphorylation of Smad1/5/8 (Fig. 1c), indicative of canonical BMP signaling. Immunohistochemical analysis demonstrated that untreated or TGF-β1 treated cultures were almost exclusively ED-1 positive indicating very high purity for microglia (>98%) (Fig. 1d). We did, however, observe an increase in GFAP immunoreactivity in the cultures after 24 h of BMP-2 treatment (Fig. 1d) indicating that BMP-2 caused proliferation of contaminating astrocytes in the cultures. Morphological assessment of microglia demonstrated that TGF-β1 treatment led to a significant rounding of ED-1 positive cells by 24 h whereas BMP-2 treatment did not significantly affect microglial morphology (Fig. 1e).

Figure 1.

Response of primary microglia to transforming growth factor-beta 1 (TGF-β1) and bone morphogenetic protein-2 (BMP-2). Western blot of lysates from primary microglia treated with TGF-β1 (1 ng/mL; 30 min) (a, b) or BMP-2 (100 ng/mL; 60 min) (c). Blots were probed with total and phosphorylated Smad2 (a), Smad3 (b) or Smad1/5/8 (c). Westerns were replicated with a minimum of three independent cell cultures. Immunohistochemical staining (10X) of microglia treated for 5 days with either BMP-2 (100 ng/mL), TGF-β (1 ng/mL) or vehicle (d). Astrocytes (GFAP-red; arrowheads) were almost undetectable in control or TGF-β1-treated cultures, where almost all cells were microglia (ED-1, green). However, GFAP+ cells were more common following BMP-2 treatment. ED-1 staining (20X) demonstrates the rounded morphology of TGF-β1-treated microglia as compared to untreated or BMP-2-treated microglia (e). Yellow arrows denote morphology of untreated microglia, whereas white arrows denote the more rounded morphology.

To further examine the effects of TGF-β1 on microglia, we treated cultures overnight with TGF-β1 and/or the ALK5 (TβR1) inhibitor SB505124 (2 μM). TGF-β1 induced the transcription of the genes encoding the TGF-β receptors, TβR1 and TβR2, and the inhibitor Smads (Smad6 and Smad7), which are involved in feedback inhibition of TGF-β signaling (Fig. 2a) (Park 2005). As expected, TGF-β1 reduced the expression of the pro-inflammatory cytokines, tumor necrosis factor alpha (TNF-α) and Il-1β (Fig. 2a). In contrast, when cells were treated with SB505124, we observed a significant decrease in expression of TβR1, Smad6 and Smad7 and a significant increase in the expression of pro-inflammatory cytokine genes compared to expression in untreated cells (Fig. 2a). These data suggest that microglial culture medium contains active TGF-β1 that alters the basal level of gene expression in untreated cells.

Figure 2.

Effects of transforming growth factor-beta 1 (TGF-β1) treatment on primary microglia. QPCR analysis of RNA isolated from microglia treated with TGF-β (1 ng/mL) for 18 h with or without a pre-treatment (6 h) with the ALK5 inhibitor SB505124 (2 μM) (a, b). Data are presented as a heat map illustrating fold change over untreated controls (a). Graphical representation of QPCR analysis of TβR1 (b) (mean ± SEM, n = 4, ***p < 0.001). Detailed analysis of all other genes on the heat map are shown in Figure S1. Photomicrograph (5X, DAPI staining) and quantification of microglial cell number after 5 days treatment with either TGF-β1 (1 ng/mL) and/or SB505124 (2 μM) (c, d). Significantly fewer microglia remain after TGF-β1 treatment as compared to untreated samples while significantly more microglia were observed after SB505124 treatment (**p < 0.01). Quantification of LDH in medium from microglia cultured with and without TGF-β1 (1 ng/mL) for 24 h (e). TGF-β1 significantly increased the amount of LDH released by microglia. Time course of microglial cell number with and without TGF-β1 (1 ng/mL) treatment (f). Microglia cell numbers decreased with time in culture. The rate of cell loss was significantly increased by exogenous TGF-β1 (****p < 0.0001).

TGF-β1 reduces microglial proliferation (Spittau et al. 2013). We observed that treatment of microglia with SB505124 resulted in significantly more microglia than untreated cells, providing additional evidence of basal TGF-β1 signaling in untreated cells (Fig. 2c, d). In SB505124 treated cultures, all cells were still ED-1+, showing that inhibition of TGF-β1 signaling induced microglial and not astrocytic proliferation. In contrast, TGF-β1 treatment resulted in a significant decrease in the number of microglia compared to untreated conditions (Fig. 2c, d, f), and an increased release of LDH (Fig. 2e) suggesting that at least part of the decrease in microglial number was because of TGF-β1-induced cell death. Together, these data indicate that low levels of TGF-β1 (e.g., endogenously released) exert anti-proliferative effects on microglia, whereas higher levels of TGF-β1 can induce microglial cell death. Although 2 μM SB505124 exerted potent effects on microglia when given alone, this dose did not completely antagonize the effects of exogenous administration of TGF-β1 (Fig. 2a–d).

We next determined the effects of LPS treatment on TGF-β1 signaling in microglia. LPS treatment (1, 10, and 100 ng/mL) resulted in a significant reduction in TβR1, TβR2, and Smad6 transcript levels at 8 and 24 h (Fig. 3a–c). Western analysis also demonstrated reduced TβR1 protein levels in LPS-treated microglia (Fig. 3d, e). As expected, LPS resulted in a significant increase in the levels of transcripts encoding TNF-α and IL-1β (Fig. 3a) and iNOS protein (Fig. 4h). Together, these data indicate that LPS treatment of microglia reduces the expression of genes/proteins known to be significantly involved in TGF-β1 signaling.

Figure 3.

Effects of lipopolysaccharide (LPS) on genes involved in transforming growth factor-beta 1 (TGF-β1) signaling. QPCR analysis of RNA isolated from microglia treated with LPS for (a) 8 or (b) 24 h. Data are presented as a heat map illustrating fold change over untreated controls. LPS reduced the expression of several genes involved in TGF-β1 signaling, whereas genes encoding pro-inflammatory molecules were induced. Graphical representation of QPCR analysis of expression of TβR1 after LPS treatment for 8 h (c) (mean ± SEM, n = 4, ***p < 0.001). Detailed analysis of all other genes on the heat map are shown in Figures S2 (8 h) and S3 (24 h). Representative western blot demonstrating that TβR1 is reduced after 24 h LPS (100 ng/mL) treatment (d). Quantification of western blots of TβR1 expression (e) (mean ± SEM, n = 3, *p < 0.05). TβR1 expression was normalized to GAPDH expression.

Figure 4.

Effect of lipopolysaccharide (LPS) treatment on transforming growth factor-beta 1 (TGF-β1)-mediated phosphorylation of Smad2 in primary microglia. Western blots of pSmad2, total Smad2, pSmad3, total Smad3 and iNOS expression after 24 h treatment with LPS (10 ng/mL) followed by 30 min treatment with TGF-β1 (1 ng/mL) (a). Quantification of western blots (mean ± SEM, n = 6; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001), normalized to the protein indicated (b–h).

The ability of TGF-β1 to induce phosphorylation of Smad2 and Smad3 was also examined in LPS-activated microglia. Microglia were pre-treated with LPS for 24 h before addition of TGF-β1 for 30 min. Treatment with TGF-β1 resulted in increased levels of pSmad2 (Fig. 4a, b). In contrast, pSmad3 was not induced by TGF-β1 (Fig. 4a, c). Treatment of microglia with LPS for 24 h led to a decrease in the expression of Smad2 (Fig. 4d) which resulted in a decrease of TGF-β1 induced pSmad2 (Fig. 4a, b), but overall an increase in the pSmad2/Smad2 ratio (Fig. 4f) in TGF-β1-treated microglia compared to control. Neither Smad3 or pSmad3 expression nor the ratio of their expression were altered by pre-treatment with LPS (Fig. 4 c, e, g). Finally, induction of iNOS confirmed LPS activation of the microglia (Fig. 4a and h).

As TGF-β1 treatment decreased microglial survival, we wanted to determine whether LPS could antagonize TGF-β1 effects thereby enhancing microglial survival. A five day TGF-β1 treatment resulted in an approximate 70% reduction in the number of microglia compared to untreated, control cells (Fig. 5a, b). In contrast, LPS (1, 10, and 100 ng/mL) resulted in an increase in the number of microglia compared to control. Co-treating microglia with TGF-β1 and low doses of LPS resulted in significantly more microglia by day 5 as compared to the TGF-β1-only group but less than that observed for the LPS-only group (Fig. 5a, b). However, when microglia were treated with TGF-β1 together with a high dose of LPS (100 ng/mL), microglia cell numbers were equivalent to those treated only with LPS (Fig. 5a,b). This effect occurred irrespective of whether LPS was given before or after TGF-β1. Together, these experiments indicate that the ability of TGF-β1 to alter microglia survival was reduced with increasing doses of LPS.

Figure 5.

Effect of transforming growth factor-beta (TGF-β) cytokines on microglial survival. Quantification of microglial cell number (DAPI staining) after a 5 day treatment with lipopolysaccharide (LPS) (1, 10 or 100 ng/mL) with or without TGF-β1 (1 ng/mL) treatment (a, b). Microglia were treated with TGF-β1 for 90 min before the addition of LPS (a). Microglia were treated with LPS for 90 min before the addition of TGF-β1 (b). Five days later, cells were fixed with 4% paraformaldehyde, stained with DAPI and labeled with Iba-1. Quantification of DAPI staining reveals that TGF-β1 reduces microglia survival, while LPS induces proliferation. Values indicated within the black bars are the percentage survival of TGF-β1-treated microglia (number of cells remaining in the TGF-β1/LPS group divided by the number of cells remaining in the corresponding LPS group). Note that percentage survival increased with increasing doses of LPS. Photomicrographs (10X, Iba1) of microglial cultures after treatment for 5 days with TGF-β1 (1 ng/mL), Activin-A (10 ng/mL) or bone morphogenetic protein-2 (BMP-2) (100 ng/mL) with or without LPS (100 ng/mL) (c, d). Cytokines were added 90 min before the LPS. Quantification of cells in (c) by counts of DAPI-stained nuclei (d) (mean ± SEM, n = 4. *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001). Both TGF-β1 and Activin-A resulted in significantly less microglia as compared to control while BMP-2 had no effect. LPS significantly inhibited the ability of TGF-β1 and Activin-A to induce microglial cell death while inducing death in BMP-2-treated microglia at a rate similar to that observed in LPS only group.

We also analyzed whether other members of the TGF-β superfamily could alter microglial survival and whether LPS could counter such actions. We examined Activin-A, which like TGF-β1 is capable of signaling via pSmad2/3 and has been shown to exert anti-inflammatory actions in a murine microglia cell line (Sugama et al. 2007). As observed with TGF-β1, Activin-A treatment for 5 days led to a significant loss of Iba-1 positive cells (Fig. 5c, d). LPS treatment substantially reduced the ability of Activin-A to induce microglial cell death (Fig. 5c, d) increasing the number of microglial cells in the culture. This occurred even though high dose LPS (100 ng/mL) in this batch of microglia resulted in decreased survival rather than proliferation (Fig. 5d). BMP-2, which signals primarily through pSmad1/5/8 rather than pSmad2/3 has been suggested to exert pro-inflammatory actions (Dmitriev et al. 2010, 2011) and therefore could potentially induce proliferation of microglia. However, BMP-2 did not lead to an alteration in microglial cell number (Fig. 5c and d), neither did LPS co-treatment increase microglial survival in the presence of BMP-2 (Fig. 5d). Thus, different members of the TGF-β superfamily of cytokines had very different effects on microglial survival: TGF-β and Activin-A treatment led to microglial cell death, whereas BMP-2 treatment did not alter microglial survival. Similarly, LPS antagonized the effects of TGF-β and Activin-A on microglial survival, but had little interaction with BMP-2.

Lastly, we examined the effect of co-treatment of LPS with the TGF-β cytokines on gene expression in microglia (Fig. 6a). Although Activin-A and TGF-β1 exerted differential effects on genes related to TGF-β signaling, both reduced the expression of pro-inflammatory markers at the transcript level. BMP-2 treatment failed to alter the expression of anti- or pro-inflammatory genes in primary microglia. LPS treatment alone, or together with either Activin-A, TGF-β1 or BMP-2 led to a strong reduction in the expression of many genes important to TGF-β signaling (Fig. 6a). LPS treatment together with TGF-β1 or Activin-A induced IL-1β transcripts, whereas TGF-β1 or Activin-A alone reduced its expression, demonstrating that LPS can overcome the anti-inflammatory actions of these TGF-β cytokines in these cells (Fig. 6a, b). However, Activin-A or TGF-β1 co-treatment with LPS also led to a large increase in transcription of arginase, a marker for M2 microglia, indicating that Activin-A or TGF-β may influence the effect of LPS signaling on some genes to reduce the inflammatory phenotype (Fig 6a, c). LPS had differential effects on TGF-β mediated transcription of the inhibitory Smads, Smad6 and Smad7. LPS, which down-regulated the expression of both when given alone, reduced the TGF-β mediated induction of Smad6, but had no effect on TGF-β mediated induction of Smad7 (Fig. 6d and e, respectively). Overall, LPS has a suppressive effect on TGF-β or Activin-A signaling and counteracts many of these cytokines' anti-inflammatory actions, at least partially through reducing the expression of genes important to TGF-β signaling.

Figure 6.

Effect of lipopolysaccharide (LPS) and transforming growth factor-beta (TGF-β) cytokines on expression of genes in primary microglia. QPCR analysis of RNA isolated from microglia treated with TGF-β1 (1 ng/mL), Activin-A (10 ng/mL) or bone morphogenetic protein-2 (BMP-2) (100 ng/mL) for 90 min before the addition of LPS (100 ng/mL) (a). Microglia were harvested 24 h after the addition of cytokines. Graphical representation of qPCR data for individual genes (b–e) (mean ± SEM, n = 4. *p < 0.5, **p < 0.01, ***p < 0.001, ***p < 0.001; ****p < 0.0001). Detailed expression of other genes analyzed on the heat map are shown in Figure S4.

Figure 7.

Schematic illustrating antagonistic actions of lipopolysaccharide (LPS) on transforming growth factor-beta 1 (TGF-β1)-signaling in microglia. Binding of TGF-β1 to its receptor leads to phosphorylation of Smad2 and Smad3, complex formation with Smad4, and subsequent translocation to the nucleus where Smads regulate gene expression. TGF-β1 also signals through non-Smad pathways. In microglia, TGF-β1 treatment leads to a down-regulation of pro-inflammatory cytokines and an up-regulation of several genes involved in TGF-β1 signaling including the iSmads, Smad6 and Smad7, as well as TGF-β receptors. TGF-β1 can also inhibit microglial proliferation (at low doses) and induce death (at higher doses). LPS induces pro-inflammatory cytokines and down-regulates several genes that are involved in TGF-β1 signaling. LPS also impairs the ability of TGF-β1 to regulate gene expression and to induce microglial cell death. This in turn, results in prolonged survival of ‘activated’ microglia.


TGF-β1 signaling is known to counteract the effects of LPS/toll like receptor 4 (TLR4) signaling in many cell types including microglia (Kim et al. 2004; Le et al. 2004). Interactions between these two signaling pathways are cell and context dependent (Seki et al. 2007). However, although the effects of LPS activation of microglia have been extensively characterized, little is known of the effects of LPS on TGF-β1 signaling in microglia. As TGF-β1 is a critical anti-inflammatory cytokine in the CNS, antagonism of its microglial signaling could enhance the ability of LPS to maintain microglial activation for prolonged periods. In this article, we report that LPS treatment of microglia resulted in decreased protein levels for the TGF-β receptor, TβR1, decreased protein levels of the transcription factor Smad2, and reduced transcription of genes prominently involved in TGF-β signaling. LPS treatment also reduced TGF-β1 stimulated gene expression, inhibited TGF-β1-mediated suppression of pro-inflammatory cytokines and reduced TGF-β1-mediated death of primary microglia. Thus, in microglia, LPS and TGF-β1 actions are mutually antagonistic.

LPS mounted a robust pro-inflammatory response in primary microglia even in the presence of TGF-β1. Whereas the LPS-mediated induction of TNF-α was inhibited by TGF-β1 treatment, LPS-mediated induction of iNOS was not altered and LPS-mediated up-regulation of IL-1β was enhanced by TGF-β treatment, the latter an indication that the two signaling pathways can synergize. Such a synergistic effect was also observed when examining the expression of arginase, a purported marker of M2 microglia. As previously reported in macrophages (Salimuddin et al. 1999), LPS strongly up-regulated microglial expression of arginase (Fig. 6c) illustrating the difficulty in neatly categorizing microglia into these predefined states (Mosser and Edwards 2008; Gordon and Martinez 2010). TGF-β treatment alone induced arginase expression a little, but TGF-β substantially enhanced induction mediated by LPS (Fig. 6c). The mechanism through which TGF-β and LPS signaling pathways synergistically interact to induce arginase or IL-1β is not known. However, this synergistic interaction seems restricted and may denote unique transcriptional control elements in these genes. More common was the antagonism between LPS and TGF-β1 pathways as illustrated by the ability of LPS to prolong survival of TGF-β1-treated microglia.

TGF-β's anti-inflammatory actions result in part from inhibiting microglial activation and proliferation (Suzumura et al. 1993). TGF-β1 has also been shown to cause microglial cell death although the literature is somewhat confusing (Xiao et al. 1997; Kim et al. 2004). The confusion partially results from the different effects of TGF-β1 on primary microglia and the microglial cell line BV2. TGF-β1 induces apoptosis in primary microglia (Xiao et al. 1997) but prevents cell death induced by ligands such as LPS (Kim et al. 2004) in BV2 cells. Our experiments indicated that 1 ng/mL TGF-β1 induced cell death of primary microglia (Fig. 5a), whereas doses of TGF-β1 up to 10 ng/mL did not affect the survival of BV2 cells (data not shown). Interestingly, antagonizing basal TGF-β1 signaling with the ALK5 inhibitor, SB505124, resulted in significant proliferation of primary microglia (Fig. 2d), suggesting that the low levels of TGF-β1 in the medium, potentially secreted by microglia (Spittau et al. 2013) inhibits microglial proliferation. In agreement, neonatal mice that are deficient for TGF-β1 show a pronounced microgliosis, suggesting that in wild type mice TGF-β1 suppresses the activation and proliferation of microglia (Brionne et al. 2003). In addition, exogenous intrathecal application of TGF-β1 in a neuropathic pain model led to a reduction in spinal microgliosis and microglial activation in the spinal cord (Echeverry et al. 2009). Thus, TGF-β1 in cell culture and in vivo has profound inhibitory effects on microglial proliferation, and may also reduce microglial survival.

Previous studies have demonstrated that LPS/TLR4 signaling antagonizes TGF-β1-signaling in macrophages (Kim and Kim 2011). One consequence of LPS/TLR4-mediated inhibition of TGF-β1 signaling in macrophages is hyperactivation and a sustained production of pro-inflammatory cytokines, which may be beneficial in the short term. Interestingly, Alzheimer's transgenic mice that also have macrophage-specific defects in TGF-β signaling have improved pathology over those with normal TGF-β signaling, and increased clearance of toxic Aβ molecules, demonstrating the potential benefit of interfering with TGF-β actions in macrophages (Wyss-Coray et al. 2001). It is unclear, however, whether inhibiting TGF-β1 signaling in microglia would produce beneficial effects in the CNS. First, unlike macrophages, microglia have considerably longer life spans and can thus spend a significant amount of time in the activated state (Kofler and Wiley 2011; Norden and Godbout 2013). Indeed, chronic activation of microglia underlies neuropathology in many conditions as microglia release molecules that are neurotoxic (Block et al. 2007). Thus, LPS/TLR4-mediated inhibition of TGF-β1-signaling in microglia may initially be beneficial in vivo, when microglial activation is needed, for example, to attack invading pathogens. However, a prolonged down-regulation of microglial TGF-β1 signaling may contribute to prolonged activation. Indeed, a single intraparenchymal injection of LPS can trigger microglial activation for up to a year (Herrera et al. 2000). It is not known whether LPS-mediated suppression of TGF-β1 signaling contributes to such chronic microglial activation.

In this study, LPS treatment resulted in the down-regulation of several genes important for TGF-β signaling. We observed decreased transcript levels for TGF-β receptors, TβR1 and TβR2, which are known to contribute to TGF-β signaling in microglia, as well as a down-regulation of ALK1, whose role is unknown in microglia. We also showed that Smad2 and Smad3 transcripts were down-regulated by LPS treatment. Interestingly, western analysis confirmed that Smad2 protein levels were decreased by LPS whereas Smad3 remained unchanged. We cannot rule out the possibility that the selective decrease in Smad2 protein is mediated by post-translational events such as ubiquitination (Heldin and Moustakas 2012). We also observed that exogenous TGF-β1 resulted in increased phosphorylation of Smad2 but not Smad3. It is currently unknown whether Smad2 phosphorylation is critical for events such as TGF-β1-mediated cell death in microglia or whether LPS-mediated reduction in Smad2 expression is critical for the ability of LPS to interfere with TGF-β1-signaling.

TGF-β1-mediated induction of the inhibitor Smads (I-Smads), Smad6 and Smad7 provides an intrinsic negative feedback mechanism. Smad6 predominantly inhibits BMP-2-signaling whereas Smad7 inhibits Activin-A- and TGF-β-mediated signaling (Itoh et al. 1998; Ishisaki et al. 1999). Other signaling pathways induce I-Smads to down-regulate TGF-β1 signaling. In some cell types, TNF-α and IL-1β induce Smad7 through NF-κB signaling (Bitzer et al. 2000). Exposure to laminar shear stress (Topper et al. 1997), UV radiation (Quan et al. 2005), the phorbol ester, 12-O-tetradecanoylphorbol-13-acetate (Tsunobuchi et al. 2004) and IFN-γ (Ulloa et al. 1999) can also suppress TGF-β1 signaling by up-regulating Smad7. However, in microglia, we found that LPS reduced, rather than induced the expression of Smad6 and Smad7. TGF-β1, as expected, induced both I-Smads, but co-treatment with TGF-β and LPS had a differential effect on the I-Smads. LPS did not reduce the TGF-β1-mediated induction of Smad7 although it completely blocked the TGF-β-mediated induction of Smad6 (Fig. 6d, e). This differential regulation of Smad6 and Smad7 obviously indicates different mechanisms of transcriptional control or RNA stability, but may also point to different functions for these two proteins.

Smad6 can repress NF-κB signaling in macrophages (Kim and Kim 2011) through sequestration of Pellino-1 (Choi et al. 2006), a protein upstream of NF-κB activation. Thus, TGF-β1 may reduce pro-inflammatory cytokine production through induction of Smad6 expression which could bind to Pellino and reduce NF-κB activation. LPS suppression of Smad6 expression would release the brake on Pellino and allow activation of NF-κB and transcriptional induction of inflammatory genes. It remains to be seen whether these mechanisms operate in microglia.

We also examined the effects on microglia of Activin-A and BMP-2, two additional members of the TGF-β superfamily. In general, Activin-A, similar to TGF-β1, exerted anti-inflammatory effects reducing expression of TNF-α and IL-1β as previously shown (Sugama et al. 2007). However, unlike TGF-β1, Activin-A did not induce transcription of Smad6, Smad7, TβR1 and TβR2 suggesting that a different repertoire of genes are activated by TGF-β1 and Activin-A and that these cytokines exert anti-inflammatory effects via different mechanisms in microglia possibly involving non-Smad signaling. Activin-A, similar to TGF-β1, induced microglial cell death that was suppressed by co-treatment with LPS, whereas BMP-2 had no significant effect on microglial cell survival. BMP-2, which can be pro-inflammatory (Dmitriev et al. 2010, 2011), did not exert major pro-inflammatory actions on microglia. BMP-2 treatment for one hour induced phosphorylation of pSmad1/5/8, whereas 24 h treatment induced expression of Smad6 and noggin (Fig. 6a). We cannot exclude the possibility that these BMP-2 effects occurred in astrocytes even though these cells accounted for less than 2% of the total cell population at the start of the BMP-2 treatment.

In conclusion, we have shown that LPS significantly antagonizes the anti-inflammatory actions of TGF-β1 on primary microglia. Indeed, it appears logical that LPS/TLR4 signaling could down-regulate TGF-β1 signaling in order to prevent death of microglia and to ensure activation of these cells, while LPS/TLR4 signaling is still active. It will therefore be interesting to determine whether inhibition of TGF-β1 signaling is a common mechanism of pro-inflammatory molecules, especially those that signal through TLRs. Such studies should provide further insight into mechanisms underlying chronic activation of microglia, and may also provide new therapeutic directions for reducing chronically activated microglia in various neuropathological conditions.


This study was supported by a grant from the Defense Medical Research Development Program (AJS). We are grateful to members of the Symes laboratory for their helpful comments and suggestions. The opinions and assertions contained herein are the private opinions of the authors and are not to be construed as reflecting the views of the Uniformed Services University of the Health Sciences or the US Department of Defense. The authors have no conflict of interest to declare.