The antagonism of interferon-gamma (IFN-γ) by heparin: examination of the blockade of class II MHC antigen and heat shock protein-70 expression


Dr John A Kirby, Surgical Immunobiology Unit, Department of Surgery, The Medical School, University of Newcastle, Newcastle upon Tyne NE2 4HH, UK.


IFN-γ is a pleiotropic cytokine that is primarily involved in the regulation of immune cell activation and the development of tissue inflammation. It is capable of activating a range of non-immune cells, including those of the vascular endothelium. These cells respond by increasing the expression of intracellular and cell-surface molecules such as class II MHC antigens and adhesion molecules that, together, increase the tendency for interaction with immune cells. It is known that IFN-γ can bind cell surface and extracellular heparan sulphate. Furthermore, soluble heparin can inhibit the function of this cytokine, presumably by competitive displacement from the cell surface, resulting in the failure of normal receptor signal transduction. In this study it is shown that heparin can prevent normal induction of the class II transactivator and heat shock cognate protein-70 in an IFN-γ-treated endothelial cell line. Both of these molecules are dependent on the activation of intracytoplasmic STAT-1, which is the most receptor proximal component of their respective induction pathways. This provides further evidence for the blockade by heparin of ligand activation of the specific IFN-γ receptor.


Vascular endothelium is primarily responsible for maintaining an anti-coagulant surface across which blood can flow and for regulating the passage of fluid and metabolites into body tissues. The surface of endothelium is covered by a dense layer of proteoglycan, the main functional component of which is glycosaminoglycan (GAG). The most abundant endothelial GAG is the polyanionic species, heparan sulphate (HS) [1]. The commonly used anti-coagulant, heparin, is biosynthetically related to HS, but is more extensively sulphated [2]. Heparin is known to bind with high affinity a range of growth factors including basic fibroblast growth factor and cytokines, including many of the chemokines, IL-1, IL-2, IL-3, IL-6, tumour necrosis factor-alpha (TNF-α) and IFN-γ[3,4].

Class II MHC proteins bind and present peptide epitopes to the antigen receptor of CD4+ T lymphocytes. Class II MHC molecules are present constitutively on a restricted number of ‘professional’ antigen-presenting cell types such as dendritic cells and B cells. However, many body cells, including those of the endothelium, can be induced to express high levels of class II antigens by stimulation with the proinflammatory cytokine IFN-γ.

Our knowledge of the regulation of class II MHC gene expression originated from the study of various mutant cell lines expressing defective class II MHC regulatory factors [5] and cells from patients suffering from the class II antigen deficiency characteristic of bare lymphocyte syndrome (BLS) [6]. The molecular defects in BLS fall into various groups that all act at the level of gene transcription. Two factors that are important for the transcription of class II MHC genes, namely RFX-5 [7] and RFX-AP [8], bind to the X box of MHC II promoters. Furthermore, RFXANK, a gene encoding a subunit of RFX, is essential for the interaction of both RFX-5 and RFX-AP with the MHC II promoters [5]. A fourth regulator of class II MHC gene expression is the class II transactivator (CIITA). It is likely that this protein interacts with transcription factors that are bound to the class II MHC promoter region; indeed, it is known that CIITA does not bind directly to DNA [9].

CIITA is regarded as a master regulator since the RFX transcription factors are ubiquitously expressed whilst CIITA is only expressed concurrently with the MHC II genes. The induction of CIITA is therefore an essential precursor to expression of the MHC II genes. Both constitutive and inducible class II MHC antigen expression is quantitatively regulated by CIITA. Muhlethaler-Mottet et al. [10] showed that CIITA is regulated by a variety of promoters. Two of these direct constitutive expression in dendritic cells and B cells, whilst a third governs IFN-γ-induced expression. It has recently been reported that the level of CIITA is positively correlated with IFN-γ[11]. These studies highlight the importance of CIITA as an essential and rate-limiting step in IFN-γ-mediated class II MHC expression. CIITA therefore acts as an ideal indicator for examination of the importance of interaction between IFN-γ and heparan sulphate for activation of the intracellular signalling process responsible for the induction of class II MHC antigens.

IFN-γ is able to stimulate its receptor on endothelial cells by achieving a high concentration on the cell surface. There is growing evidence that an important function of endothelial cell surface HS is to bind cytokines, so that their local concentration is maintained. In the absence of such a mechanism, cytokines would lose biological activity rapidly by dilution into the blood or tissue fluids; this is an important function at the luminal surface where the blood flow is rapid. Previous studies by our group and others have shown that soluble heparin-like GAGs can suppress the induction of class II MHC antigen expression, ordinarily produced by the treatment of endothelial cells with IFN-γ[12]. It has been proposed that this effect is mediated by inhibition of cytokine binding to cell surface HS molecules as a result of direct competition for IFN-γ binding by soluble heparin and cell surface HS.

In this study we examine how the expression of class II MHC protein by endothelial cells changes in response to stimulation by IFN-γ in presence and absence of heparin. In order to verify the direct blockade by heparin of IFN-γ receptor stimulation, we have examined the effect of heparin on expression of the CIITA.


Cell culture

The HMEC-1 cell line [13] was propagated in MCDB-131 medium (Sigma, Poole, UK) supplemented with epidermal growth factor (EGF; 10 ng/ml), hydrocortisone (1 μg/ml), 10% fetal calf serum (FCS) and gentamycin (50 μg/ml). Cells were cultured in 25-cm2 flasks (Falcon Plastics; Becton Dickinson, Cowley, UK) at 37°C in an atmosphere containing 5% CO2 in air. Confluent cell monolayers were detached by treatment with trypsin-EDTA (Sigma) and were split in a ratio of 1:3; the cells were routinely subcultured every 2–3 days.

Cytokine treatment

For experimental use, HMEC-1 cells were transferred from flasks and grown in 24- or 12-well cluster plates (Falcon), using 3 ml of culture medium per well. Once confluent, the medium was supplemented with 100–1000 U/ml of recombinant IFN-γ (R&D Systems, Abingdon, UK) for varying periods. In some experiments the medium was also supplemented with heparin (Sigma) at concentrations varying between 0 and 1000 μg/ml. Cells were maintained in complete MCDB-131 during the stimulatory period.

Cultured cells from representative plates were recovered by brief treatment with trypsin EDTA for flow cytometric antigen quantification or RNA extraction for reverse transcriptase-polymerase chain reaction (RT-PCR).

Class II MHC antigen expression assay

The expression of class II MHC antigens was determined after IFN-γ stimulation by semiquantitative immunofluorescence staining of detached HMEC-1 cells, using methodology previously described by this group [12]. Briefly, cells recovered from each well were labelled with an optimized concentration of FITC-conjugated MoAb specific for class II MHC antigens (CR3/43; Dako, High Wycombe, UK). After incubation for 25 min at 4°C, the cells were washed by centrifugation and analysed by flow cytometry (FACScan; Becton Dickinson); a minimum of 10 000 events was accumulated. Unstained cells were used to assess autofluorescence and an isotype-matched, FITC-conjugated irrelevant antibody (Dako) was used for negative control staining. In each case the median fluorescence values were determined by application of Lysis II software (Becton Dickinson).

Each assay was performed in triplicate on several occasions and the mean results were calculated.

RT-PCR and cloning of the amplified product

Total RNA was isolated from HMEC-1 cells using RNAzol B (Biogenesis Ltd, Poole, UK) according to manufacturer’s instructions. RNA was quantified by absorbance at 260 nm and was stored at −20°C.

First strand complementary DNA synthesis was performed using 1 μg of total RNA in the presence of MuLV reverse transcriptase (RT) and oligo (dT)16 primer. Reaction mixtures for PCR contained the complementary DNA template 0·5 μmol each of sense and anti-sense primers and 2·5 U of Taq DNA polymerase (Perkin-Elmer-Cetus, Warrington, UK). A DNA fragment encoding residues 484–1105 of human CIITA was amplified. Controls of RNA integrity consisted of PCR amplifications of the 3′-untranslated region of glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Amplifications were performed in a DNA thermal cycler (Perkin-Elmer-Cetus 480) for 35 cycles under the following conditions; 94°C for 1 min, 55°C for 1 min and 72°C for 2 min. The sense and anti-sense primers for CIITA and GAPDH were (5′ACCAGATGAAGTGATCGGTG3′; 5′GTTGGAGACCTCTCCAGCTG3′) and (5′ AGTCAACGGATTTGGTCGTA3′; 5′ AAATGAGCCCCAGCCTTCT3′), respectively.

The amplified products were electrophoresed on a 1·5% agarose gel, the band was excised and the product DNA purified and eluted on a Quiquick column (Qiagen, Crawley, UK) . The DNA was cloned into the TA cloning vector pCR2.1 (Invitrogen, Groningen, The Netherlands) and the fidelity and orientation of the DNA was confirmed by sequencing. Additional bands of slightly larger size were also amplified by the CIITA primers; for purposes of identification these bands were purified, cloned and sequenced in the same manner.

Statistical analysis

All results are expressed as mean ± s.e.m. of the corresponding replicates, the significance of changes in class II MHC antigen expression was assessed by application of Student’s t-test. Data were stored and analysed using Prism 3 Software, GraphPad.


Initial experiments demonstrated that IFN-γ induces the HMEC-1 cell line to express high levels of class II MHC antigens. Figure 1a shows the effect of stimulation of these cells with a range of concentrations of IFN-γ for 72 h. Class II MHC antigens were significantly up-regulated on the cell surface by stimulation with IFN-γ at concentrations as low as 100 U/ml. The expression of class II antigens increased following stimulation with higher concentrations of IFN-γ; this effect was not saturated by an IFN-γ concentration as high as 6000 U/ml. An IFN-γ concentration of 1000 U/ml was chosen for subsequent experimentation.

Figure 1.

Flow cytometric data showing the median immunofluorescence associated with expression of class II MHC antigens by cultured HMEC-1 cells. (a) Expression of class II antigens after stimulation with a range of IFN-γ concentrations for 72 h. (b) Expression of class II antigens after stimulation with IFN-γ at 1000 U/ml for a range of incubation periods. In each case the data points represent the mean ± s.e.m. of three determinations.

The results in Fig. 1b show how the expression of class II MHC antigens varied with the time of stimulation of HMEC-1 cells with 1000 U/ml IFN-γ. Up-regulation of antigen expression was significant by 36 h (P < 0·05) and reached a level after 60 h that did not increase further by 84 h (P > 0·05). An optimum time of 72 h was chosen for subsequent experiments.

The results in Fig. 2 show the effect on class II MHC antigen expression by HMEC-1 cells of stimulation with IFN-γ in the presence of a range of concentrations of heparin. It was found that heparin antagonized the activity of IFN-γ, with a concentration of only 50 μg/ml reducing the class II protein induction at 72 h to 50% of the control level. The addition of heparin at 1000 μg/ml was not toxic to the cells but completely abrogated the induction of MHC II antigens, with antigen expression by these cells being no different from that by unstimulated cells (P < 0·05).

Figure 2.

Results from a representative experiment demonstrating the effect of various concentrations of heparin on the median immunofluorescence associated with class II MHC antigens induced by treatment with 1000 U/ml of IFN-γ for 72 h. The bars represent the mean of duplicate determinations; similar data were obtained in three separate experiments.

A series of experiments was performed to determine the effects of removal of IFN-γ or of the addition of heparin at varying times after the initiation of cytokine stimulation of HMEC-1 cells; the results are shown in Fig. 3. When IFN-γ was removed from cultured HMEC cells after 12 h, the expression of MHC class II proteins after 84 h was found to be no different from that of cells that had been incubated continuously with IFN-γ (P > 0·05). However, it was found that the expression of class II antigens at 60 h and 84 h was markedly reduced (P < 0·05) when heparin (500 μg/ml) was added to cultures that had already been stimulated with IFN-γ for 12 h. Addition of the GAG chondroitin sulphate, which is known to have no effect on the activity of IFN-γ[12], did not reduce the potential for class II antigen expression (data not shown).

Figure 3.

Flow cytometric data showing the median immunofluorescence associated with expression of class II MHC antigens by cultured HMEC-1 cells at varying times after stimulation with IFN-γ at 1000 U/ml. ▪, Cultures maintained with IFN-γ throughout the experimental period; hatched bars, the effect of removing IFN-γ from the medium after 12 h; □, the effect of adding heparin at 500 μg/ml after 12 h. The bars represent the mean of duplicate determinations; similar data were obtained in three separate experiments.

The regulation of class II MHC genes is primarily mediated at the transcriptional level by CIITA. In order to investigate further the mechanism by which heparin can regulate IFN-γ-induced MHC II protein expression, a series of experiments was performed to measure changes in the expression of mRNA encoding CIITA in IFN-γ-stimulated HMEC-1 cells. Figure 4 shows the results of RT-PCR identification of mRNA encoding the CIITA; mRNA for the housekeeping protein GAPDH was also amplified for positive assay control. It was found that unstimulated HMEC-1 cells did not express CIITA mRNA. However, after treatment of the cells with IFN-γ, the product of RT-PCR using primers specific for CIITA showed three bands on agarose gel electrophoresis; increasing the stringency of the reaction did not eliminate the extra bands. These bands were all excised, cloned into pCR2.1 and sequenced.

Figure 4.

Agarose gel fractionation of reverse transcriptase-polymerase chain reaction (RT-PCR) products from resting and cytokine-stimulated HMEC-1 cells. IFN-γ was applied for 72 h where indicated at 1000 U/ml; heparin was added where indicated at 500 μg/ml. Lanes 1a,b, PCR products from unstimulated cells; lanes 2a,b, cells stimulated with IFN-γ; lanes 3a,b, cells stimulated with IFN-γ in the presence of heparin. The marker lanes (M) contain a 100-bp ladder. Lanes marked ‘a’ in each case demonstrate the amplification of a house-keeping gene, glyceraldehyde 3-phosphate dehydrogenase (GAPDH). Lanes marked ‘b’ show RT-PCR products amplified with class II transactivator (CIITA)-specific primers.

Sequence matching revealed the middle band ( Fig. 5) to be completely homologous to a region of 525 bp of cDNA encoding the heat shock cognate protein 70 (HSC70; sequence-accession Y00371 [14]). This sequence was amplified as a consequence of coincidental primer homology. The top band (approximately 600 bp) was not identified, but the lower band corresponded to CIITA cDNA. Detailed sequencing of multiple clones of the CIITA product consistently revealed a deletion of 148 bp between residues 596 and 744.

Figure 5.

Agarose gel fractionation of reverse transcriptase-polymerase chain reaction (RT-PCR) products from HMEC-1 cells stimulated for varying periods with IFN-γ (1000 U/ml) either with or without the presence of heparin (500 μg/ml). At each time point, cells stimulated with IFN-γ yielded three product bands; in each case these bands were not produced, or very weakly produced, by cells stimulated in the presence of heparin. At each time point the lanes on the left represent cells stimulated with IFN-γ only, whereas the lanes on the right show the PCR products amplified when cells were stimulated with IFN-γ in the presence of heparin.

It was found that both CIITA and HSC70 were induced with similar kinetics by treatment of HMEC-1 cells with IFN-γ. Messenger RNA encoding both species was detected after stimulation for 6 h, reached a peak after 8–12 h and remained detectable for 72 h (results not shown). Apart from the common primer sequences, no significant sequence homology exists between HSC70 and CIITA.

Further experiments were performed to determine the effect on CIITA transcription of the addition of heparin during the stimulation of HMEC-1 cells with IFN-γ. Cells were treated with either IFN-γ or IFN-γ plus heparin (1000 μg/ml) over a period of 72 h and RT-PCR was performed. It was found that addition of heparin blocked the normal induction by IFN-γ of transcripts for both CIITA and HSC70.


This study was performed to investigate further the molecular mechanism by which heparin-like GAG molecules are able to prevent the induction and up-regulation of class II MHC antigens during stimulation of endothelial cells with IFN-γ. An immortalized human microvascular endothelial cell line HMEC-1 was used in this series of experiments [13]. These cells are known to share many phenotypic and functional properties with primary endothelial cell lines, including adhesion molecule expression, cytokine responsiveness and the capacity for primary activation of alloantigen-specific T lymphocytes [15].

Of particular significance for this study is the experimental reproducibility offered by use of a continuous cell line. It is known that primary endothelial cell lines from different donors show responses to cytokine stimulation that are highly variable between donors [16]. However, initial experiments did demonstrate that the kinetics for class II MHC antigen induction by HMEC-1 cells was comparable with that typically observed for IFN-γ-stimulated primary human endothelial cells, showing maximal expression after stimulation for 60 h [17].

IFN-γ is thought to bind to the surface of endothelial cells by interaction between basic amino acids located within the C-terminal region of the molecule and HS containing proteoglycan on the cell surface. This interaction is believed to be necessary for optimal activation of endothelial cells by this cytokine. It has been reported previously by this group [12] and others [18] that heparin can antagonize the capacity of IFN-γ to induce the expression of class II MHC molecules. Soluble heparin is thought to act by competitive sequestration of the cytokine in solution, thereby preventing the interaction with HS on the cell surface. Significantly, it is demonstrated in this study that addition of heparin after initial stimulation of cells with IFN-γ is also sufficient to reduce class II MHC antigen up-regulation. This is suggestive of competitive displacement from the cell surface of a reservoir of active IFN-γ that would otherwise be capable of maintaining chronic stimulation of the cells, producing optimal class II antigen expression. This finding is consistent with the observation that intravenous heparin can displace radiolabelled cytokines from binding sites on the luminal surface of the vascular endothelium [19].

Further information concerning the functional role of cell surface heparan sulphate has been provided by examination of the response to stimulation by IFN-γ of endothelial cells that express under-sulphated HS [12]. In this study it was found that cells treated with chlorate, a metabolic inhibitor of GAG sulphation, showed a reduced expression of class II MHC antigens after stimulation with IFN-γ. This suggests that the polyanionic charge produced by GAG sulphation might be important for sequestration of basic residues on cytokine molecules. This model is supported by the observation that transient cleavage of HS from the surface of endothelial cells by treatment with heparinase II results in a significant reduction in the binding of radio-iodinated IFN-γ[12].

Despite this evidence, it has been largely inferred that sequestration of IFN-γ onto cell surface HS is a prerequisite for specific receptor engagement resulting in intracytoplasmic signal transduction and class II MHC antigen expression. The current study provides evidence for this model by demonstrating for the first time that soluble heparin can block the intracellular induction of CIITA produced normally by stimulation of endothelial cells with IFN-γ. This failure of transactivator production accounts for the absence of class II MHC induction by cells stimulated with IFN-γ in the presence of heparin. The 148-bp deletion observed in the CIITA sequence amplified from HMEC-1 cells has been reported previously for B lymphocytes and involves a region of the transcript that would otherwise be rich in serine, threonine and proline residues [20].

It is also reported for the first time that heparin can prevent cytokine-induced up-regulation of a member of the 70-kD heat shock protein family, HSC70. The main function of the heat shock proteins is to bind to, or chaperone, unfolded proteins, thereby protecting them from damage. The 70-kD family shows a high degree of interspecies homology and is the most abundant family of the heat shock molecules [21]. In addition to their maintenance of normal protein structure, heat shock proteins have been implicated in the development of certain immune responses, including the development of autoimmune diseases [22]. Indeed, it has been suggested that the 70-kD family may play a role in antigen processing and presentation [23].

There are conflicting reports about the effect of IFN-γ on the expression of members of the 70-kD heat shock protein family. For example, Schett et al. observed that IFN-γ had no effect on the expression of these proteins by synovial fibroblast-like cells [24]. However, Dsouza observed that IFN-γ could induce the expression of these proteins by cultured oligodendrocytes [25]. Unlike other members of the 70-kD heat shock protein family, HSC70 is normally considered to be expressed constitutively [14]. However, it has been shown that the expression of this protein in certain melanoma cell lines can be up-regulated by heat shock, implying variability in the control of this protein in different cell types [26].

In the current study it was shown that mRNA encoding HSC70 was not present in resting HMEC-1 cells, but was induced rapidly following stimulation with IFN-γ. As with the activation of CIITA, it was also demonstrated that heparin could antagonize the potential of IFN-γ to induce HSC70.

The intracellular signalling cascade resulting in the transcription of CIITA is known to include members of the Janus kinase/signal transducers and activators of transcription (JAK/STAT) pathway. Indeed, several studies have implicated the activation of STAT-1 as playing a role in the control of CIITA expression [27–29]. This was recently confirmed by Muhlettaler-Mottet et al.[30], who demonstrated that treatment of cells with IFN-γ resulted in the activation of STAT-1, which could then bind to a region on the CIITA promoter. In addition, it has also been shown that the up-regulation of heat shock proteins by IFN-γ is mediated by the activation of STAT-1 [31]. Previous work by our group has shown that treatment with heparin can inhibit the normal activation of STAT-1 that follows stimulation of endothelial cells with IFN-γ[32].

The series of experiments described in this study provide further evidence to support the conclusion that heparin binds to IFN-γ in solution, preventing normal sequestration onto cell surface GAG molecules. In turn, this prevents specific receptor ligation that would otherwise result in STAT-1 activation and the transcription of both CIITA and HSC70. This model is entirely consistent with the known immunomodulatory properties of heparin. It is possible that further clarification of this mechanism will enable the development of novel and safe immunosuppressive agents for clinical use.


The authors are grateful to the Northern Counties Kidney Research Fund and Newcastle University Research Committee for supporting this research.