Article Text

Extended report
Ankylosing spondylitis monocytes show upregulation of proteins involved in inflammation and the ubiquitin proteasome pathway
  1. C Wright1,2,
  2. M Edelmann2,
  3. K diGleria1,
  4. S Kollnberger1,
  5. H Kramer2,
  6. S McGowan3,
  7. K McHugh1,
  8. S Taylor3,
  9. B Kessler2,
  10. P Bowness1,4
  1. 1
    Human Immunology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, UK
  2. 2
    Central Proteomics Facility, Henry Wellcome Building for Molecular Physiology, Department of Clinical Medicine, Roosevelt Drive, University of Oxford, UK
  3. 3
    Computational Biology Research Group, University of Oxford, UK
  4. 4
    Nuffield Orthopaedic Centre NHS Trust, Oxford, UK
  1. Correspondence to C Wright, Human Immunology Unit, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, OX3 9DS, UK; cynthiaw{at}


Objectives: To determine if peripheral blood monocytes from patients with ankylosing spondylitis (AS) differed in protein expression compared to rheumatoid arthritis (RA) and healthy controls (HC).

Methods: Monocyte protein expression was characterised by 2D gel electrophoresis and by label-free quantitative expression profiling, using nano-ultra performance liquid chromatography coupled to electrospray ionisation mass spectrometry (ESI-MSE, where E refers to low/high collision energy switching). Data sets were analysed using the Waters expression profiling system and Ingenuity pathway analysis (IPA).

Results: Two-dimensional gel electrophoresis showed upregulation of proteasomal constituents in AS monocytes, including the β subunit of proteasome activator (PA)28. Monocyte expression profiling and IPA showed that significant changes in protein expression within the ubiquitin proteasome pathway (UPP) were restricted to AS monocytes. Statistically significant differences in protein expression involving the leucocyte extravasation, vascular endothelial growth factor, integrin and Toll-like receptor signalling pathways were seen in AS and RA monocytes compared to healthy controls. No evidence of upregulation of proteins involved in the endoplasmic reticulum stress response pathway was found in either AS or RA monocytes. Finally, the PA28 complex was shown to increase the generation of human leucocyte antigen (HLA)-B27 antigenic epitopes by the proteasome in vitro.

Conclusions: Our proteomic analyses support the hypothesis that monocytes play an important role in the pathogenesis of AS and RA, and further suggest a specific role in AS for the UPP. Quantitative proteomic expression profiling constitutes a powerful new tool for rheumatology research.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Ankylosing spondylitis (AS), the commonest of the spondyloarthritides, is an inflammatory rheumatic disease with a predilection for the axial skeleton. Clinical hallmarks of AS include sacroiliitis, uveitis, enthesitis and persistent spinal inflammation. AS remains difficult to diagnose and the pathogenic mechanisms of disease causation and perpetuation are poorly understood. AS has a highly significant association with the major histocompatibility complex allele human leucocyte antigen (HLA)-B27.1 The natural function of HLA class I proteins is to present intracellular-derived peptides at the cell surface to the T cell receptors of cytotoxic T lymphocytes in the context of an immune response. These peptides are generated by proteasomal degradation in the ubiquitin proteasome pathway (UPP). While the arthritogenic peptide theory proposes a role for peptide presentation to CD8 cytotoxic T cells,2 transgenic rodent models have suggested roles for HLA-B27 heavy chains and CD8-positive monocytes.3 An interaction between abnormal HLA-B27 forms and natural killer (NK) family immunoreceptors has been shown.4 5 The unfolded protein response has also been implicated.6 Recently, multiple genetic factors have been related to in AS pathogenesis.7

Proteomic techniques have recently shown altered protein expression in osteoarthritis and rheumatoid arthritis.8 Electrospray quadrupole – time of flight (ESI-Q-TOF) mass spectrometry identified increased levels of haptoglobin precursor isoforms in AS sera.9 However, systematic proteomic analysis of a specific cell type in AS has not yet been carried out. Monocytes are precursors of professional antigen presentation cells, including macrophages and dendritic cells. Microarray analysis of spondyloarthropathy (SpA) blood monocytes showed differentiation marker upregulation,10 and elevated expression of Toll-like receptor (TLR)4 was demonstrated on SpA monocytes.11 This led us to hypothesise that AS peripheral blood monocytes would show differences in protein expression and implicate biochemical pathways involved in disease pathogenesis.

In this study, we use proteomic techniques coupled with systems biology analysis to show that AS and RA monocytes express elevated levels of proteins involved in a variety of inflammatory pathways compared to healthy control monocytes. Monocytes from patients with AS, but not patients with RA, show statistically significant differential UPP protein expression. Lastly we show that proteosome activator (PA)28, a member of the UPP, can increase the generation of HLA-B27-restricted epitopes.


A more detailed description of the methods performed can be found in the Supplementary materials.


This study included 14 patients with AS who met the modified New York criteria12 and 10 with rheumatoid arthritis (RA) who fulfilled the American College of Rheumatology (ACR) revised criteria.13 Ethical permission was obtained from the relevant local institution (Oxfordshire REC 06/Q1606/139). All participants gave informed consent. All patients with AS were HLA-B*2705 positive by DNA typing. The mean age for patients with AS was 42 years old and the range was from 28 to 69 years old. Eight were male, seven received non-steroidal anti-inflammatory drugs (NSAIDs), four disease-modifying antirheumatic drugs (DMARDs) and none anti-tumour necrosis factor (TNF) drugs. The mean (SD) scores for patients with AS on the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), Bath Ankylosing Spondylitis Functional Index (BASFI) and Bath Ankylosing Spondylitis Metrology Index (BASMI) scales were 5.8 (2.6), 5.4 (2.8) and 5.2 (1.8), respectively. The mean age of patients with RA was 48 years old and the range was from 18 to 68 years old. Six were female, four received NSAIDs, six received DMARDs and none anti-TNF agents. The Disease Activity Score (DAS) for patients with RA was 5.2 (2.3). In all, 14 individuals who were HLA-B27− and 4 individuals who were HLA-B27+, 8 male with a mean age of 40 years old (range 28–64 years old), were included as healthy controls.

Monocyte isolation, labelling and protein extraction

Monocytes were isolated from peripheral blood mononuclear cells (PBMCs) using MACS CD14 Microbeads (Miltenyi Biotec, Bisley, UK). Purity was >95% (see fig 1A). A total of 5×106 purified monocytes were labelled with 13.25 MBq of [35S] Promix (GE Healthcare, Chalfont St Giles, UK) overnight. Lysate proteins were precipitated via methanol-chloroform precipitation.14

Figure 1

Two-dimensional gels show upregulation of proteasome constituents in ankylosing spondylitis (AS) monocytes. (A) Representative fluorescence-activated cell sorting (FACS) characterisation of CD14-bead sorted peripheral blood monocytes before proteomic analysis: left, 95.9% of cells expressed CD14; middle, 0.54% of cells expressed CD3; right, 0.62% of cells expressed CD19. (B) Two-dimensional gel electrophoresis comparison of [35S]-labelled monocyte proteins of a patient with AS (left) and a healthy control (right). Magnified box shows five of seven spots consistently upregulated in the patients with AS. Gels shown are representative of four patients with AS and eight healthy controls.

Two-dimensional gel electrophoresis

[35S]-Labelled proteins were separated according to isoelectric point and molecular weight and analysed as described.15 Unlabelled monocyte protein spots were excised from duplicate silver-stained gels,16 digested and identified by mass spectrometry.17

Monocyte fractionation, quantitative proteomic and Ingenuity pathway analysis (IPA)

Monocytes were fractionated as described,18 and proteins methanol/chloroform precipitated and trypsin digested.15 Monocyte protein digests were subjected to ultra performance liquid chromatography tandem mass spectrometry (nano-UPLC-MSE, where E refers to low/high collision energy switching) analysis using a Waters Nano-Acquity UPLC system coupled to a Waters QTOFpremier tandem mass spectrometer (Waters, Milford, Massachusetts, USA). Data was acquired in MSE mode and processed with ProteinLynx Global Server (PLGS V.2.2.5, Waters, Milford, Massachusetts, USA). Peaklists were searched against the SwissProt/NCBInr databases using PLGS and Mascot (Matrixscience, London, UK), Quantitative analysis of MS data was performed using the Waters Expression Profiling System (WEPS), only changes greater than ±1.3 with a p value of <0.05 were included in the analysis. IPA (Ingenuity Systems, identified significantly regulated pathways using the right-tailed Fischer exact test. The Benjamini–Hochberg method was used to adjust p values for multiple comparisons.

Peptide synthesis, in vitro digestion, analysis and quantification

Extended peptides (LELRSRYWAIRTRSGSN and NQQITANRELQQELAAA) containing the HLA-B27 epitopes SRYWAIRTR,19 and NRELIQQEL20 were synthesised and purified to >98% by high performance liquid chromatography (HPLC) chromatography. In vitro proteasome digests were performed as described.21 Proteasomes were pelleted and supernatants containing digested peptides were analysed using a nano-UPLC-MSE mass spectrometer.


2D gel electrophoresis shows elevated levels of proteasomal proteins in AS monocytes

We first compared the peripheral blood monocyte proteome of four patients with AS with that of eight healthy controls by 2D gel electrophoresis. Figure 1B shows representative 2D gels from a patient with AS (left) and a healthy control (right). Seven proteins were found to be upregulated more than twofold in all patients with AS compared to controls. These proteins are lettered a to g (fig 1B). Corresponding spots were excised from duplicate silver-stained gels, digested with trypsin and analysed by liquid chromatography electrospray ionisation tandem mass spectrometry (LC-ESI-MS/MS). The identities of these proteins, together with those of three expression controls (1–3), are shown in table 1.

Table 1

Ankylosing spondylitis (AS) monocyte proteins identified by 2D gel electrophoresis and liquid chromatography tandem mass spectrometry (LC-MS/MS)

Spot a was identified as the metabolic enzyme superoxide dismutase, however with a pI shift indicating a post-translational modification. Spot c was identified as the constitutive proteasome subunit β type 4. The proteasome activator complex subunit β (PA28β) was confidently identified in spot f (Mascot score 350). The aspartic protease cathepsin D was also detected in this spot, however with a lower mascot score of 103. Furthermore, immunoblotting on additional five patients and three control samples showed equal cathepsin D expression (data not shown). Since quantitative expression profiling on further patients with AS (see below) independently confirmed increased PA28 expression, we concluded that the increased intensity in spot f was caused by PA28 upregulation. Spot g was identified as the constitutive proteasome subunit β type 3.

Differential protein expression in the UPP in AS monocytes

We then studied the monocyte proteome from seven further patients with AS using label-free quantitative nano-UPLC-MSE mass spectrometry and IPA. Two experiments were conducted; the first using pooled monocytes from three patients with AS, three patients with RA and three healthy controls, and the second using pooled monocytes from four patients with AS, four patients with RA and four healthy controls. Pooling was necessary in order to obtain sufficient protein concentration to derive comparative quantitative data, and holds the advantage of revealing consistent biological trends while averaging out individual variation. Nano-UPLC-MSE mass spectrometry peak list files were processed by WEPS to obtain comparative quantitative data on protein expression levels between groups. Quantitative data from both experiments were combined and the AS and RA data, yielding 811 and 729 proteins, respectively, were each compared to that from the healthy controls using IPA. Pathway significance was calculated using the right-tailed Fischer exact test and corrected for multiple comparison using the Benjamini–Hochberg method. The most significant pathways for the AS and RA pools, based on the percentage of differentially expressed proteins in a pathway compared to the healthy control pool (absolute numbers of proteins and p values shown in brackets), were the leucocyte extravasation (AS: 28, p = 0.001, RA: 21, p = 0.026), vascular endothelial growth factor (VEGF) (AS: 15, p = 0.011, RA: 12, p = 0.028), Janus kinase (JAK)/signal transducer and activator of protein (STAT) (AS: 11, p = 0.033, RA: 12, p = 0.014) and TLR pathways (AS: 11, p = 0.036, RA: 10, p = 0.031) (see fig 2A). Table 2 shows a subset of these differentially expressed proteins. No statistically significant evidence of involvement of the endoplasmic reticulum stress response was found in either the AS or RA pools.

Figure 2

Ingenuity pathway analysis (IPA; identifies differential protein expression in inflammatory pathways in ankylosing spondylitis (AS) and rheumatoid arthritis (RA) monocytes, and in the ubiquitin proteasome pathway (UPP) in AS monocytes alone. The x axis shows probability of involvement of given pathway, corrected for multiple comparisons, using the Benjamini–Hochberg method. (A) Quantitative comparison: the right-tailed Fischer exact test was used to calculate the p value from the combined data of two expression analysis experiments determining the probability that each signalling pathway assigned to that data set was due to chance alone. The leucocyte extravasation pathway had a p value of 1.32×10−3 for the AS monocytes and 2.59×10−2 for the RA monocytes. The vascular endothelial growth factor (VEGF) pathway p value was 1.06×10−2 for the AS monocytes and 2.77×10−2 for the RA monocytes. The Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway p value was 3.26×10−2 (AS) and 1.41×10−2 (RA), and the Toll-like receptor (TLR) pathway 3.62×10−2 and 3.14×10−2, respectively. The UPP was identified as statistically significant (p value of 1.82×10−3) for the AS group but not for the RA group. (B) Independent (qualitative) comparison of proteins identified uniquely in AS, RA or healthy control monocytes. No inflammatory pathways were identified in the healthy control group. The UPP was identified as statistically significant in only the AS group (p value = 9.54×10−4). The UPP p value in the RA group was 1.27×10−1. p = 0.05 corresponds to −log(1.3). Combined results from two expression analysis experiments are shown, the first using monocytes from three patients with AS, three patients with RA and three healthy controls, the second using monocytes from four patients with AS, four patients with RA and four healthy controls.

Table 2

Proteins within the leukocyte extravasation, vascular endothelial growth factor (VEGF), integrin, Toll-like receptor (TLR) and ubiquitin proteasome pathway (UPP) upregulated in peripheral blood monocytes of patients with ankylosing spondylitis (AS) and patients with rheumatoid arthritis (RA) compared to healthy controls

The only pathway in which a marked difference between patients with AS and patients with RA was observed was the UPP. A total of 25 proteins in this pathway were differentially expressed in the AS pool (p = 1.82×10−3), including 5 ubiquitin-specific peptidases. By contrast, only 12 of 202 UPP proteins were differentially expressed in the RA pool. Both subunits of the proteasome activator complex were upregulated within AS and RA pools. In addition, the rpn-6 subunit of the 19S cap of the proteasome, which is replaced by the PA28 complex in the immunoproteasome, was twofold downregulated in the AS pool (see table 2).

We next looked for proteins expressed at increased levels in AS monocytes whose expression is known to be upregulated as a consequence of TNFα; 14 such proteins were found and are listed in table 3.

Table 3

Identification of proteins upregulated in ankylosing spondylitis (AS) monocytes that are known to be tumour necrosis factor (TNF)α inducible

In view of a recent gene array study on AS macrophages,37 we also looked for evidence of interferon (IFN)γ-inducible proteins. In addition to the PA28 subunits and TLRs listed in table 2, we identified several IFNγ inducible proteins upregulated in the AS monocytes, five of which are listed in table 3.

We then repeated IPA with an independent non-overlapping protein data set, analysing proteins that were uniquely identified in either the AS group, the RA group or the healthy control group. This analysis (fig 2B) supported the analysis of the quantitative data set. Thus the only pathway uniquely implicated in the AS group was again the UPP (p value = 9.54×10−4). Of note, five proteasomal subunits, including the immunoproteasomal subunit low molecular mass polypeptide 7 (LMP-7) (β5i), were uniquely present in the AS group, compared to one proteasomal subunit in the healthy control and none in the RA group. The leucocyte extravasation, VEGF and integrin pathways were identified in the AS and RA groups. Interestingly the nuclear factor κB (NFκB) pathway was uniquely implicated in the monocytes from patients with RA in quantitative and qualitative analyses. No inflammatory pathways were implicated from the proteins uniquely present in the healthy control group.

PA28 can increase the generation of HLA-B27 restricted peptide ligands

As we found evidence of upregulation in proteasomal proteins in AS monocytes including both components of the PA28 complex, we therefore tested the hypothesis that PA28 may modulate the generation of HLA-B27 epitopes. We conducted an in vitro proteasomal digestion assay of extended HLA-B27-restricted epitopes in the presence or absence of the PA28 complex and quantified the production of the epitopes generated by mass spectrometry. The concentration of the HLA-B27-restricted influenza nucleoprotein epitope SRYWAIRTR38 increased fourfold in the presence of PA28 (fig 3A). Likewise, the HLA-B27-restricted Chlamydia NRELIQQEL epitope20 was generated with a threefold increase in the presence of PA28 (fig 3B).

Figure 3

Proteosome activator (PA)28 increases the generation of two human leukocyte antigen (HLA)-B27-restricted epitopes. (A) Ion extraction chromatogram of triple-charged 1208 Da HLA-B27 restricted influenza nucleocapsid protein (NP) SRYWAIRTR epitope shows ion intensities of 522 at t = 20 min, 820 at t = 40 min and 1380 at t = 60 min in the absence of PA28. In the presence of PA28 ion intensities are 2840 at t = 20 min, 3920 at t = 40 min and 4850 at t = 60 min. (B) Generation of the double-charged 1142 Da HLA-B27 restricted Chlamydia epitope NRELIQQEL increases shows ion intensities of 1770 at t = 20 min, 3660 at t = 40 min and 4440 at t = 60 min in the absence of PA28. In the presence of PA28 ion intensities are 4050 at t = 20 min, 7460 at t = 40 min and 10 900 at t = 60 min. Relative abundance of 100% was set to the maximum ion intensities, respectively 4850 in (A) and 10 900 in (B).


We have used traditional and novel quantitative label-free proteomics approaches to show changes in protein expression in AS and RA monocytes when compared to healthy controls. Upregulation of proteins involved in inflammatory responses, specifically those in the leucocyte extravasation, VEGF, JAK/STAT and TLR signalling pathways, were found in AS and RA monocytes. Our findings of common upregulation of these inflammatory pathways emphasise the importance of using inflammatory rheumatic disease controls. However, changes in protein expression in a statistically significant number of proteins belonging to the UPP were restricted to AS monocytes. This included upregulation of the proteasome activator complex, which we have shown can increase the generation of two HLA-B27-restricted antigenic peptide epitopes in vitro.

This is the first study of a specific cell type in rheumatic disease that couples new powerful quantitative proteomic techniques with systems biology analysis. The limitations of traditional proteomic methods can now be overcome using sophisticated analytical tools to distil and interpret large data sets. Firstly, we have used nano-UPLC-MSE to identify far greater numbers of proteins than two-dimensional electrophoresis. Secondly, the datasets generated were subjected to IPA to identify signalling pathways significantly involved in AS and RA monocytes. Our results strongly suggest a role for proinflammatory monocytes in both these rheumatic diseases. Thus, AS monocytes showed significant changes in expression level of 63 proteins within the leucocyte extravasation, VEGF, JAK/STAT and TLR pathways, with a further 25 in the UPP. We also found upregulation of proteins whose expression is increased by TNFα. Leucocyte extravasation is a critical precursor to macrophage differentiation and subsequent effector functions such as antigen presentation at the site of inflammation. Polymorphisms in the VEGF pathway, which was identified by IPA to be significant in both rheumatic conditions, have been implicated in psoriatic arthritis39 and in the severity of AS.40 In agreement with the previously reported upregulation of TLRs in PBMCs of patients with SpA and RA,11 we observed an upregulation in TLR1, TLR2 and TLR9 in the AS group, and of TLR2 in the RA group (see table 2). In agreement with the study by Gu and colleagues,10 we also observed a 2.3-fold upregulation of myeloid-related protein 8 (MRP8) in AS monocytes (see table 3), a TNF-inducible protein expressed by macrophages in chronic inflammation.

In contrast to recent findings in a gene array study on AS macrophages,37 we did not find evidence of IFNγ dysregulation. This may be explained by differences in treatment (50% of patients in the above-mentioned study had received anti-TNF therapy) or study protocol (which included ex vivo culture). We did, however, observe a signature of proteins whose expression is increased by TNFα (table 3). No patients in our study had received TNFα blockers at the time of sampling, although most patients with AS and patients with RA were eligible for anti-TNF therapy based on their disease activity scores. Taken together, our data show clear evidence of an active monocyte inflammatory response in AS and RA, supporting and extending previous reports. The fact that similar data was obtained using several independent proteomic techniques serves to validate our approach. Nevertheless in future our results should be corroborated with gene expression, immunoblotting and fluorescence-activated cell sorting (FACS) analysis.

The UPP controls cellular homeostasis and generates antigenic peptides for presentation to T cells. This was the only pathway uniquely identified in the monocytes of patients with AS from quantitative analysis of differentially expressed proteins and from proteins uniquely present in each group. Five proteins upregulated in the UPP were identified as ubiquitin-specific peptidases. These enzymes remove ubiquitin from proteins targeted for degradation, playing a key regulatory role in numerous cellular processes. However, with the exception of ubiquitin-specific peptidase 7 (USP7), which is involved in transcriptional regulation, DNA replication and apoptosis,41 these enzymes remain uncharacterised and their substrates unknown. Also within the UPP, PA28 and proteasomal subunits were shown to be upregulated. As no previous studies on PA28 dependence have included HLA-B27 restricted epitopes, we conducted in vitro proteasomal digestion of peptide precursors containing HLA-B27 epitopes in the presence of PA28. Our preliminary data show for the first time that PA28 can increase the generation of HLA-B27 epitopes in vitro. We do not suggest that increased levels of PA28 affect the presentation of HLA-B27-restricted epitopes preferentially, and further detailed in vivo analysis of epitope generation is required to establish any biological implications with respect to AS.

Our proteomic analyses support the hypothesis that monocytes play an important role in the pathogenesis of AS (and RA), and further suggests a specific role in AS for the UPP. Quantitative proteomic expression profiling constitutes a valuable new tool for rheumatology research that can delineate disease-specific protein pathways.


Supplementary materials


  • Funding This study was supported by the Medical Research Council, UK, the Natural Science and Engineering Research Council, Canada, the Arthritis Research Campaign, UK, the National Ankylosing Spondylitis Society, UK and the Oxford Biomedical Research Centre.

  • Competing interests None declared.

  • Ethics approval Ethics approval was obtained (Oxfordshire REC 06/Q1606/139).

  • ▸ Additional data (Supplementary methods) are published online only at