The switch from proteasome to immunoproteasome is increased in circulating cells of patients with fast progressive immunoglobulin A nephropathy and associated with defective CD46 expression

ADDITIONAL CONTENT

An author video to accompany this article is available at: https://dbpia.nl.go.kr/ndt/pages/author_videos.

INTRODUCTION

Immunoglobulin A nephropathy (IgAN) is a disease with variable course, and patients who progress to end-stage renal disease (ESRD) have an extremely variable rate of glomerular filtration rate (GFR) decline. Over the last decade, the pathogenic events leading to the development of IgAN have been elucidated and several factors have been identified [1, 2]. The production of excessive amounts of poorly galactosylated IgA1 molecules (Gd-IgA1) along with the synthesis of IgG autoantibodies, leads to immune complex formation and deposition in the mesangium, with local complement activation and subsequent glomerular damage [3, 4]. However, the mechanisms underlying different kidney disease progression rates remain unknown. Hence there is great interest in new markers for disease progression, with the hope of improving risk stratification [5].

The production of Gd-IgA1 is initiated in mucosal sites by surface receptors, such as toll-like receptors (TLRs), specifically recognizing pathogen-associated molecular patterns expressed by multiple pathogens [6–8]. The recognition of invading microbes by TLRs on dendritic cells (DCs) in the mucosae induces cytokine production and enhances antigen presentation to naive T cells, resulting in antigen-specific adaptive immune response activation. After ligand binding, TLRs trigger nuclear factor-κB (NF-κB) and interferon (IFN) regulatory factor transcription in DCs [9, 10]. IFN-γ, tumour necrosis factor (TNF)-α and TLR agonists then induce the replacement of the proteasome (PS) catalytic units β1, β2 and β5 (encoded by the genes PSMB5, PSMB6 and PSMB7), with low molecular weight proteins (LMP2 and LMP7) and a multicatalytic endopeptidase-like complex (MECL-1) (encoded by the genes PSMB8, PSMB9 and PSMB10, respectively), transforming the PS into immunoproteasome (iPS) [10–12]. These modifications result in an enhanced catalytic property and a ‘professionalization’ of PS in presenting peptides to major histocompatibility complex class I, leading to optimal lymphocyte activation. We previously reported an upregulation of iPS in patients with IgAN [13], but the role of these changes in the progression of IgAN has never been investigated.

Among the factors playing a role in inflammation and renal damage after macromolecular IgA mesangial deposition, complement activation is of great interest given the increasing number of complement-targeting drugs [14]. Several studies have tried to elucidate the possible role of defective complement control in IgAN, with conflicting results, but unanimously concluding on a possible defective regulation of an alternative complement pathway factor (CFH-H) via CFH-related protein [15–17]. We focused our interest on the complement regulator Cluster of differentiation-46 (CD46) (or membrane cofactor protein), which modulates at cellular level the C3 convertase activity of the lectin and alternative complement pathways. We investigated the transcriptional expression in peripheral white blood cells (WBCs) of CD46 in 157 patients enrolled by the Validation Study of the Oxford Classification of IgAN (VALIGA) [18] and found a significantly lower expression of CD46 messenger RNA (mRNA) in patients with progressive IgAN [19]. CD46 is not only involved in the regulation of complement activation, but is also expressed on T-cell membrane, where it promotes a switch in T helper type 1 (Th1) to type 1 regulatory T cell (Tr1) differentiation [20, 21]. Hence CD46 emerges as a key sensor of immune activation and a core modulator of adaptive immunity [22].

We explored the correlations between iPS switch and estimated glomerualr filtration rate (eGFR) decline in 157 patients with IgAN enrolled in the international VALIGA study [18], where the renal biopsies from multiple centres were centrally scored and the long follow-up available allowed an accurate calculation of GFR slope before sampling. We also evaluated the single nucleotide polymorphism (SNP) rs9357155 at the PSMB8/PSMB9 locus that encodes for LMP2/LMP7, previously found to be involved in the genetic susceptibility to IgAN [23, 24], and the expression of mRNAs encoding for TLRs and CD46.

We found that patients with IgAN presenting with rapid eGFR loss were characterized by a combination of high PS to iPS LMP7/β5 switch and low CD46 expression. This is the first report, to our knowledge, of an interrelationship between these two actors in innate and adaptive immunity control in patients with IgAN.

RESULTS

Categorization of progressive and fast progressive patients

The 157 subjects with IgAN from the VALIGA cohort had at the time of sampling a previous observation period of 6.4 years [interquartile range (IQR) 2.8–10.7] after renal biopsy, which allowed for a precise calculation of GRF slope (Table 1 reports data at renal biopsy and at sampling). Based on the median value of the eGFR slope of −0.41 mL/min/1.73 m²/year, patients were categorized into 79 ‘progressors’ with a median eGFR loss of −1.91 mL/min/1.73 m²/year and 78 ‘non-progressors’, who had a median improvement in eGFR of +0.89 (Supplementary data, Table S1). Based on the 75th centile of high rate of eGFR loss (−1.91 mL/min/1.73 m²/year), patients were categorized into 40 ‘fast progressors’, with a median eGFR decline of −3.8 (IQR −8.2 to −2.9) mL/min/1.73 m²/year, and 117 ‘non-fast progressors’, with a mild eGFR loss or some improvement during the follow-up [median −0.2 (IQR −0.8–2.2) mL/min/1.73 m²/year] (Table 2 and Figure 1).

FIGURE 1

Identification of progressive and fast progressive subjects in the cohort of IgAN patients. Histogram of frequency distribution of eGFR slopes in progressors and non-progressors and fast progressors and non-fast progressors. The median value of the eGFR slope in the 157 patients investigated (−0.41 mL/min/1.73 m²/year) allowed a categorization into progressors and non-progressors. The 75th centile of eGFR decrease (−1.91/mL/min/m²/year) allowed a categorization of the patients into 40 ‘fast progressors’, with median eGFR loss of −3.8 (IQR −8.2 to −2.9) mL/min/1.73 m²/year, and 117 ‘non-fast progressors’ with a median eGFR loss of −0.2 (IQR −0.8–2.2) mL/min/1.73 m²/year. The change of threshold from the 50th to 25th centile identifies a more severe su-population of IgAN patients with faster progression.

Open in new tabDownload slide

No difference in clinical data at renal biopsy or at sampling was found between progressors and non-progressors, with the exception of significantly higher T1–2 scores (Supplementary data, Table S1). A clinical feature of fast progressors (Table 2) was greater degree of proteinuria at renal biopsy and at sampling at the end of follow-up. The endpoint of 50% decline in eGFR or ESRD (reached in five patients with eGFR <15 mL/min at sampling) was significantly more frequent in fast progressors; P < 0.0001. Treatment with corticosteroid/immunosuppressive (Cs/IS) drugs and renin–angiotensin blockers was adopted with similar frequency in the two groups (Table 2).

Switch from PS to iPS in circulating blood cells of patients with IgAN

In the 157 IgAN patients investigated, a significant switch from PS to iPS was detected for LMP7/β5 [median 1.66 (IQR 1.31–2.14) versus 1.16 (0.95–1.41) in healthy controls, P < 0.001] and for LMP2/β1 [median 1.22 (IQR 0.94–2.06) versus 0.65 (0.35–1.00), P < 0.001]. No significant modification was found for the third switch, i.e. MECL-1/β2. LMP7/β5 and LMP2/β1 were significantly correlated with proteinuria at sampling (P = 0.03 and P = 0.004, respectively), but no correlation was found with eGFR at sampling (Supplementary data, Table S2 reports these correlations). No correlation was found with therapy at sampling and the investigated parameters. The two switches were similar in progressors versus non-progressors but significantly higher in fast progressors compared with non-fast progressors (Table 3) [LMP7/β5 1.87 (IQR 1.38–2.94) versus 1.58 (1.28–2.09), P = 0.04 (Figure 2B) and LMP2/β1 1.6 (IQR 1–5.55) versus 1.18 (0.94–1.73), P = 0.04]. However, only the LMP7/β5 switch was significantly correlated with the eGFR slope (P = 0.027) (Figure 2A).

FIGURE 2

Correlation between LMP7/β5 switch in peripheral WBCs and velocity of progression of IgAN. (A) Correlation between LMP7/β5 switch and eGFR slope (P = 0.027). (B) LMP7/β5 switch in fast progressors and non-fast progressors (P = 0.04).

Open in new tabDownload slide

Correlations between iPS switch, TLR and CD46 expressions

In the entire cohort of IgAN patients, a strong correlation was found between the switch from PS to iPS and TLR mRNAs (Table 4), particularly between LMP7/β5 and TLR4 mRNA expression (P < 0.0001) (Figure 3).

FIGURE 3

Correlation between LMP7/β5 switch and TLR4 mRNA expression in peripheral WBCs. Spearman's ρ = 0.31, P = 9.02e-05 (<0.0001).

Open in new tabDownload slide

A significant difference was detected in CD46 mRNA expression in peripheral blood cells of patients with fast progressing IgAN versus non-fast progressors (P = 0.007) (Table 3 and Figure 4A). LMP7/β5 switch and CD46 mRNA expressions showed inverse trends of correlation with respect to the fast progression status (Figure 4B). The fast progressor group showed a trend for a negative correlation while in the non-fast progressor group there was a trend for a positive correlation. In a logistic regression on fast progression status in the cohort of IgAN investigated that considered clinical data at renal biopsy and categorized MEST-C (mesangial hypercellularity, endocapillary hypercellularity, segmental glomerulosclerosis, tubular atrophy–interstitial fibrosis and crescent formations) scores, we observed that the association between LMP7/β5 switch and CD46mRNA expression had a strong predictor effect (Supplementary data, Table S3). In fact, patients with both low expression of CD46 and high expression of LMP7/β5 switch in peripheral blood were at high risk for progression of IgAN {odds ratio [OR]  15.8 [95% confidence interval (CI) 2.16–158.34] (Figure 5). This distribution of risk according to the LMP7/β5-CD46 group of pertinence indicates the value of such categorization in predicting the fast progression of IgAN in our cohort sample (model area under the curve (AUC) 0.78, P = 0.004].

FIGURE 4

CD46 mRNA expression in peripheral WBCs in non-fast progressors and fast progressors and correlation with LMP7/β5 switch. (A) CD46 expression in WBCs in fast progressors and non-fast progressors (Wilcoxon–Mann–Whitney test P-value = 0.007). (B) LMP7/β5 switch and CD46 mRNA expression are oppositely correlated depending on fast progression status. The graph shows the opposite trends of correlation of CD46 and LMP7/β5 expressions in patients grouped by fast progression status. The fast progressor group shows a negative trend (Spearman ρ = −0.21), with overexpression of LMP7/β5 coupled with low CD46 expression and vice versa. The non-fast progressor group shows a positive correlation trend (Spearman ρ = 0.16). Both Spearman correlation tests result were non-significant. The expression values are log-transformed to avoid the graphical crushing effect of non-normally distributed values. The Spearman correlation test was operated on crude levels of expression to avoid a biased correlation index.

Open in new tabDownload slide

FIGURE 5

LMP7/β5 and CD46 expression are strongly predictive for rapid progression. The graph shows the patients as grouped by LMP7/β5 switches and CD46 expression thresholds and its strong effect on rapid progression prediction. This is a graphical representation of the logistic model utilized for the prediction of rapid progression in our patients (Supplementary data, Table S3): the dots lying on the horizontal line at level 1 of the y-axis are the fast progressor patients, the ones lying on level 0 of the y-axis are the non-fast progressor patients. The S-shaped positioned dots between 0 and 1 (on the vertical axis) are the estimated risk of each patient based on the predictor variable values, transformed by the link logit function that characterizes binomial distribution. Each combination of predictor variables defines an estimated relative risk for each patient that, thanks to a logit transformation, lies between 0 (no risk) and 1 (full risk), assuming an asymptotic distribution. The vertical lines on the graph connect each patient's real status with the respective estimated risk, representing the residual of each combination. Dots are coloured based on LMP7/β5 and CD46 expression threshold group of pertinence. Thresholds of over/underexpressions are based on univariate ROC bootstrapped best thresholds for disease progression velocity. Patients with both underexpression of CD46 mRNA and overexpression of LMP7/β5 are estimated to be closer to 1 on the y-axis (higher risk, as shown by an OR of 15.82 by the model), with respect to patients with both overexpression of CD46mRNA and underexpression of LMP7/β5, that are closer to 0 on the y-axis (lower risk) by the model. Groups with intermediate estimated risk tend to have only underexpression of CD46 mRNA (OR 4.06) or overexpression of LMP7/β5 (OR 3.03). This distribution of risk according to the LMP7/β5-CD46mRNA group of pertinence testify to the great value of such categorization in predicting the rapid progression of IgAN patients in our sample.

Open in new tabDownload slide

Risk allele

The frequency of the risk allele in the SNP rs9357155-C was 0.87 in patients compared with 0.83 healthy Utah residents with European ancestry controls from HapMap-III, and the SNP was not associated with progressive IgAN (Table 3). No association was found with PS/iPS switches.

MATERIALS AND METHODS

Clinical data set and definitions

Definitions of clinical variables and outcomes followed the original VALIGA study [18]. Briefly, eGFR was estimated using the four-variable Modification of Diet in Renal Disease formula and proteinuria expressed in g/day. Mean arterial pressure (MAP) was calculated as one-third of the pulse pressure. At renal biopsy, 41 subjects were children (<18 years old): in these subjects, eGFR was estimated using the Schwartz formula (constant K = 0.55) with a maximum eGFR set at 120 mL/min/1.73 m². Proteinuria was expressed in g/day/1.73 m²; MAP was adjusted for gender and age, as in the original report. At sampling, three subjects were <18 years old. ESRD was defined as eGFR <15 mL/min/1.73 m² in all patients. Time-averaged proteinuria was determined for each year of observation. Immunosuppressive treatment was considered as intent to treat regardless of the type or duration. Renal biopsies were scored according to the Oxford classification: mesangial hypercellularity, M0/M1 (less than or equal to >50% of glomeruli with >4 mesangial cells/area); endocapillary hypercellularity, E0/E1 (present/absent); segmental glomerulosclerosis, S0/S1 (present/absent); and tubular atrophy/interstitial fibrosis, T0/T1/T2 (<25, >25 <50, >50%) [25]. Moreover, the presence of crescents (C1–C2) was considered [26].

Quantification of mRNA expressions

Blood samples were collected in PAXgene tubes (QIAGEN, Hilden, Germany) in patients with IgAN and healthy controls as previously described [19]. RNA was extracted from WBCs with the PAXgene RNA System Kit (QIAGEN). Quantitative real-time polymerase chain reaction was performed using TaqMan Universal PCR MasterMix (Life Technologies, Carlsbad, CA, USA). Probes for the detection of mRNAs encoding for PS subunits (β1, β2 and β5), iPS subunits (LMP2, LMP7 and MECL-1), TLR2, TLR4 and CD46 were used as previously reported [19, 27, 28]. Relative quantification of target genes expression was performed with the ΔΔC_t method and the relative fold changes were determined as previously detailed, hence results are expressed in corresponding arbitrary units [13].

Genotype data

The genotypes for rs2071543 were generated as part of the previously published studies [8, 24]. The quality control analyses were identical to the ones described in these studies.

Statistical methods

The rate of renal function decline (eGFR slope) was determined by fitting a straight line through available eGFR data from renal biopsy to sampling, as previously described [18]. Disease progression and velocity were defined according to the median and higher 75th centile value of eGFR slope in the study cohort, as described above. The functional form of all continuous variables was assessed. Data were tested for normal distribution using the Shapiro–Wilk test. Descriptive variables were presented as median (IQR, owing to non-normal distributions) or frequency (count) and compared across relevant groups using the Mann–Whitney U or Wilcoxon signed-rank test for continuous variables and Fisher exact test for categorical variables. The Oxford MEST-C scores were categorized as reported. Therapy was the intent to treat regardless of the type or duration, as in Coppo et al. [19]. Exposure to Cs/Is at sampling was also considered. Spearman rank correlation coefficient was used to investigate correlations, due to the non-linearity of assessed data. Group tests were two-sided, with P-values <0.05 considered statistically significant. Univariate and multivariate regression analyses were performed to calculate associations between iPS expression and other clinical variables, including disease progression. A reciever operating characteristics (ROC) curve served to identify the optimal threshold point of CD46 mRNA and LMP7/β5 expression based on disease progression velocity status. Multivariate logistic regression analyses were chosen to investigate the accuracy of disease progression prediction models. The best predictive model including clinical data at renal biopsy (sex, age, eGFR, proteinuria and MAP) and MEST-C scores categorized into 0 (no positive score) versus 1 (at least one score > 0) was implemented with a categorical variable defined by LMP7/β5 and CD46 mRNA combined expression thresholds, identified through the ROC of univariate models on disease progression velocity. Patients were categorized into four risk groups, from lower (CD46 overexpressed and LMP7/β5 underexpressed) to higher predicted risk (CD46 underexpressed and LMP7β5 overexpressed). Overall model fit was assessed using the Akaike's information criterion, AUC and log-likelihood test. Analyses were performed and figures generated using R 3.4.4 (R Foundation, Vienna, Austria).

DISCUSSION

We previously reported an upregulation of the iPS in peripheral blood mononuclear cells of patients with IgAN in a single-centre exploratory study [13]. The increased expression of iPS was particularly significant for the LMP7/β5 switch that correlated with persistent proteinuria.

This study investigated a larger group of 157 patients with IgAN from multiple European centres contributing to the VALIGA cohort with centrally reviewed renal pathology scores and complete long-term follow-up after renal biopsy [19]. The switch from PS to iPS was assessed as a ratio of subunit mRNA expression, as in the previous study of our group [19], as well as in recent literature [29]. The individual PS and iPS proteins are present in cells in various combinations and do not change after cell maturation [30]. In this selected cohort, we confirmed the upregulation of iPS, with increased switch of LMP7/β5 and LMP2/β1, comparied with healthy controls. No evidence of genetic conditioning was detectable for the risk allele in the SNP rs9357155-C of the PSMB8/9 gene encoding for LMP7/β5 and LMP2/β1.

The novelty of the present report is the finding of a significant association between the rate of eGFR decline and the increased expression of the LMP7/β5 in peripheral cells in patients with IgAN. Patients with fast progression (arbitrarily defined by an eGFR slope >75th centile in the whole group) had significantly higher PS-iPS switch, independent from the level of renal function at sampling or at renal biopsy. LMP7/β5 and LMP2/β1 switch correlated with the velocity of yearly loss of eGFR and proteinuria at sampling. The remarkable finding is that the same subcohort of fast progressing patients had a concomitant significant reduction in CD46 mRNA expression in peripheral cells. The combination of increased PS–iPS switch and reduced CD46 expression was strongly predictive of a fast progressor status.

Recent data have suggested a dysregulation in gut-associated lymphoid tissue (GALT) in patients with IgAN. In these patients, genetic determinants, gut dysbiosis and reactions to diet components may play a combined role [31]. iPS is highly represented in the GALT, acting in defence against pathogens [12]. In inflammatory bowel disease, an increase in LMP7/β5 switch was reported at a local intestinal mucosa level [32, 33]. LMP7 and LMP2 are encoded by the PSMB8 and PSMB9 genes and are involved in NF-κB activation, maintenance of the intestinal epithelial barrier and local control of inflammatory responses to infection [24]. In a previous genome-wide association study, rs9357155 at the PSMB8/9 locus was found to be associated with increased risk of IgAN [23]. However, this SNP was not associated with increased iPS switch in our study.

We observed a highly significant association between LMP7/β5 switch and TLR4 mRNA expression in circulating cells. TLR4 is highly expressed at the intestinal mucosa surface, as a receptor for Gram-positive bacterial lipopolysaccharide [34, 35]. We previously detected a significant increased expression of TLR4 mRNA in peripheral mononuclear cells of patients with IgAN [27]. TLR4 activation induces a release of NF-κB and IFN-β [34] aimed at defending against pathogens while maintaining tolerance to commensal bacteria. The concomitant increased expression of LMP7/β5 and TLR4 mRNAs suggests the hypothesis that an altered GALT response to intestinal components may favour an increased iPS switch in immune-competent cells that later reach the systemic circulation.

iPS is involved also in autoimmunity, and an up-regulation of iPS subunits has been detected in human autoimmune disorders [32, 36, 37]. In IgAN, an autoimmune reaction against Gd-IgA1 leads to the production of autoantibodies correlated with the progression risk [38]. Cells of the adaptive immune system are regulated by iPS, mainly by LMP7/β5 switch, inducing a reduction in regulatory T (Treg) cell differentiation and feedback. This event may lead to uncontrolled immune response and persistent T-cell activation [39]. A reduced circulation of Treg cells in patients with IgAN was reported [40] and we previously reported a lower expression of the transcriptional factor forkhead box protein P3 in peripheral mononuclear cells of children with primary IgAN and IgA vasculitis with nephritis, with increased iPS switch [28]. Treg cells play a powerful anti-inflammatory role, maintaining tolerance through the direct contact or the release of anti-inflammatory cytokines and balancing the pro-inflammatory effects of Th17. Patients with progressive IgAN have an increased renal presence of Th17-positive cells [40, 41]. These data collectively suggest that the increased expression of iPS may favour autoimmune reactions, defective regulatory T-cell control and persistent inflammatory conditions favouring disease progression in IgAN.

The coexistance of iPS high switch and CD46 mRNA low expression in patients with progressive IgAN is unlikely to be a chance finding, since the effect estimates for these associations are large. CD46 modulates at the cellular level C3 convertase activity of the lectin and alternative complement pathways; its genetic defect can lead to uncontrolled complement activation at the endothelial level and atypical haemolytic uraemic syndrome [22]. Apart from this function, CD46 has been recently identified as a costimulatory molecule for T-cell activation; CD46-costimulated human T cells induce a Tr1 phenotype with high production of the anti-inflammatory interleukin (IL)-10 [41]. As reported above, Tregs are reduced in patients with IgAN. The CD46-mediated differentiation pathway is defective in several chronic inflammatory diseases, underlying the importance of CD46 in controlling T-cell function. CD46-mediated Tr1 differentiation is altered in patients with multiple sclerosis [42], presenting with impaired IL-10 secretion upon CD46 costimulation [41, 43] in patients with lupus [44]. Hence CD46 is a multifunctional molecule, which can be down-regulated by genetic conditioning and by acquired factors, playing a role in immune activation and as a modulator of adaptive immunity. We observed that the decreased expression of CD46 mRNA in the peripheral cells of patients with IgAN significantly improved prediction of the progression rate when added to clinical and pathology features [19].

Here we report that the association of reduced CD46 mRNA expression and enhanced iPS switch in peripheral cells increases the risk of IgAN progression.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.

ACKNOWLEDGEMENTS

We thank Drs. Krzysztof Kiryluk and Ali Gharavi from Columbia University for providing genotype data for the PSMB8/9 locus and for helpful feedback on the study design.

FUNDING

The study was supported by the Immunopathology Working Group of the ERA-EDTA. The VALIGA study was granted by the first research call of the ERA-EDTA in 2009. Preliminary and partial data were presented as an oral communication at the ERA-EDTA 2016 Congress.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. We declare that the results presented in this article have not been published previously in whole or in part. The study had approval from local ethics committee and was performed in accordance with the Declaration of Helsinki.

REFERENCES