The anti-inflammatory effect of high-density lipoprotein is blunted by delivery of altered microRNA cargo in rheumatoid arthritis patients

Abstract

Objective:

High-density lipoprotein (HDL) has well-characterized anti-atherogenic cholesterol efflux and antioxidant functions. Another function of HDL uncharacterized in RA is its ability to transport microRNAs (miRNAs) between cells and thus alter cellular function. The study’s purpose was to determine if HDL-miRNA cargo is altered and affects inflammation in RA.

Methods:

HDL-miRNAs were characterized in 30 RA and 30 control subjects by next generation sequencing and quantitative PCR. The most abundant differentially expressed miRNA was evaluated further. The function of miR-1246 was assessed by miRNA mimics, antagomiRs, siRNA knockdown and luciferase assays. Monocyte-derived macrophages were treated with miR-1246-loaded HDL, and unmodified HDL from RA and control subjects to measure delivery of miR-1246 and its effect on IL-6.

Results:

The most abundant miRNA on HDL was miR-1246; it was significantly enriched 2-fold on HDL from RA versus control subjects. HDL-mediated miR-1246 delivery to macrophages significantly increased IL6 expression 43-fold. miR-1246 delivery significantly decreased DUSP3 1.5-fold and DUSP3 siRNA knockdown increased macrophage IL6 expression. Luciferase assay indicated DUSP3 is a direct target of miR-1246. Unmodified HDL from RA delivered 1.6-fold more miR-1246 versus control subject HDL. Unmodified HDL from both RA and control subjects attenuated activated macrophage IL6 expression, but this effect was significantly blunted in RA so that IL6 expression was 3.4-fold higher after RA versus control HDL treatment.

Conclusions:

HDL-miR-1246 was increased in RA versus control subjects and delivery of miR-1246 to macrophages increased IL-6 expression by targeting DUSP3. The altered HDL-miRNA cargo in RA blunted HDL’s anti-inflammatory effect.

Graphical Abstract

INTRODUCTION

Rheumatoid arthritis (RA) is a systemic inflammatory autoimmune disease associated with accelerated atherosclerosis (1) and increased risk of myocardial infarction and early death (2). A major modifiable cardiovascular risk factor is serum cholesterol concentrations. However, despite accelerated atherosclerosis, patients with RA typically have lower low-density lipoprotein (LDL) and similar high-density lipoprotein (HDL) cholesterol concentrations compared to control subjects (1, 3, 4). Thus, the functions of these lipoproteins are likely more important than their concentrations.

HDL has several well-defined functions including the ability to remove cholesterol from tissues (cholesterol efflux) and to inhibit oxidation LDL; in some but not all reports these functions of HDL are impaired in RA (5–12). Another recently characterized function of HDL is the ability to transport miRNAs between cells (13). HDL can accept miRNAs from cells and deliver them to other cells to affect their function (13–16). For example, HDL delivers miR-223 to endothelial cells where it directly targets and downregulates intracellular adhesion molecule 1 (ICAM-1) (14), decreasing the inflammatory response.

We previously showed that plasma miRNAs and other sRNAs are altered in patients with RA and associated with RA disease activity and coronary atherosclerosis (17, 18). Because HDL is a major carrier of miRNA in plasma (19), we hypothesized that the HDL-miRNA cargo is altered in patients with RA and results in pro-inflammatory HDL.

METHODS

Study population

This study included 30 patients with RA and 30 control subjects. All subjects gave written informed consent prior to any study procedures. The study was approved by Institutional Review Boards of the Nashville VA Medical Center (IRB# 867513) and the Vanderbilt University Medical Center (IRB# 150775). Inclusion criteria were age at least 18 years and meeting classification criteria for RA (RA patients only) (20). Exclusion criteria included active cancer other than skin cancer, active infection, and the presence of an inflammatory disorder (control subjects only). Patients with RA and control subjects were frequency matched for age within 5 years. Subjects were recruited from the Nashville VA Medical Center Rheumatology and Internal Medicine clinics and from Vanderbilt University Medical Center Rheumatology and Internal Medicine clinics and from participants in prior studies.

Clinical and laboratory information

Clinical information was collected from subjects and review of medical records using standardized forms. Tender and swollen joint counts were assessed by a board-certified rheumatologist (MJO). The 28-joint disease activity score using CRP (DAS28-CRP) was used to assess RA disease activity. Plasma and serum were separated from peripheral blood and stored at −80°C until use. High-sensitivity C-reactive protein (hs-CRP) was measured in the Vanderbilt University Medical Center Hospital Laboratory. Serum lipoproteins were measured by nuclear magnetic resonance (NMR) and analyzed by the Bruker IVDr Lipoprotein Subclass Analysis, as previously described (21).

HDL purification

HDL was separated from 8ml plasma by density-gradient ultracentrifugation (DGUC) for sequential isolation of lipoprotein-containing subfractions using potassium bromide: very low-density lipoprotein (VLDL) (1.006–1.018 g/ml), low density lipoprotein (LDL) (1.019–1.063 g/ml) and HDL (1.064–1.021 g/ml), as previously described (22).

The isolated HDL fractions were concentrated with Amicon Ultra 3K centrifugal filters (Millipore) and injected into an AKTA PURIFIER fast-protein liquid chromatography (FPLC) 10 system (Cytiva) composed of one Superose 6 (10/300 GL) column, as previously described (22). HDL FPLC fractions were assessed for total protein (ThermoFisher) by bicinchoninic acid (BCA) assay, and total cholesterol (TC) and phosphatidylcholine (PC) (Pointe Scientific) by colorimetric assays.

Each purified HDL sample was placed in a sealed 10K molecular weight cutoff (MWCO) dialysis sleeve (Thermo Scientific) and dialyzed in PBS at 4 °C, changing PBS 5 times with at least 3 hours in between changes, as previously detailed(22). The dialyzed HDL samples were then concentrated using an Amicon Ultra 3K centrifugal filters (Millipore). The HDL was filtered using a Ultrafree 0.22 μm filter (Millipore) in a cell culture hood under sterile conditions.

Cell culture

Human monocyte THP-1 cells (ATCC) were grown in RPMI 1640 medium (Gibco) supplemented with 10% fetal bovine serum (Gibco), 10 mM HEPES, 2mM L-Glutamine (Gibco), 1 × MEM Vitamin Solution (Gibco), and 0.125 mM beta-mercaptoethanol (Sigma) at 37°C with 5% CO₂. The cells were subcultured when their concentration reached 6–8×10⁵ cells/ml and were not used past passage 15. The HEK293 cells (ATCC) were maintained in DMEM medium (Corning) supplemented with 10% FBS (Gibco) and were subcultured at 80–90% confluence.

miR-1246 and DUSP3 assays

THP-1 monocytes were seeded into 24-well plates (3.5×10⁵ cells/well). The monocytes were then differentiated into macrophages using 50 ng/ml phorbol 12-myristate 13-acetate (PMA) (Sigma) for 72 hours, followed by two rinses with PBS. Macrophages activated with 0.1 μg/ml lipopolysaccharide (LPS) were transfected with 100nM miR-1246 mimics, antagomir, 10nM DUSP3 siRNA, CDC25A siRNA, or corresponding non-targeting control scramble siRNA (Dharmacon Reagents) using Lipofectamine RNAiMAX (Invitrogen) following manufacturer’s protocol. Cells were collected after 24 hours for quantitative PCR (qPCR). Supernatants were collected 48 hours and tested using Quantikine ELISA kit Human interluekin-6 (IL-6) (R&D Systems) according to manufacturer instructions.

Effect of miR-1246-loaded control HDL and native RA and control subject HDL on macrophages

For experiments using miR-1246-loaded HDL, 80 μg of purified control subject HDL was incubated with or without 200 pmol of miR-1246 mimic in HEPES buffer for final volume of 100 μl and gently rocked overnight at 37°C. THP-1 monocytes were treated with PMA for 72 hours as described above. Then cells were treated with 0.1 μg/ml LPS for another 24 hours. Next, the medium was removed, and cells were washed twice with PBS to remove remaining LPS. Then the cells were treated with control subject HDL that was either loaded or not loaded with either a miR-1246 mimic or control miRNA, cel-miR-67 mimic; or with native purified HDL from RA or from control subjects that was not loaded with miR-1246 mimic. Cells and supernatants were collected after an additional 24 hours for qPCR and ELISA, as above.

sRNA and mRNA next generation sequencing and alignments

RNA was extracted with Total RNA purification kits (Norgen) from 200 μl of plasma or 1 mg of purified HDL by protein measurement and sRNA next generation sequencing (NGS) libraries were prepared using NextFlex Small RNA-Seq Kits (PerkinElmer). The Vanderbilt Technologies for Advanced Genomics (VANTAGE) core facility performed library quality control and size selection of sRNAs by Pippin Prep (Sage Science). VANTAGE performed sRNA NGS using the Illumina NovaSeq6000. The TIGER pipeline (“Tools for Integrative Genome Analysis of Extracellular sRNAs”) (19) was used to align miRNAs and miRNAs were compared by DESeq2 with Benjamini-Hochberg adjusted p-values, as previously done (17, 23, 24).

For RNA-Seq, total RNA extracted as above from macrophages treated with miR-1246 mimic or vehicle control. The VANTAGE core performed NGS library preparation using NEBNext^® rRNA Depletion Kit v2 (Human/Mouse/Rat) (New England BioLabs), quality control, and NGS using the Illumina NovaSeq6000. Reads were trimmed using Cutadapt (v2.10) and aligned to the Gencode GRCh38.p13 genome using STAR (v2.7.8a) (25). Gencode v38 gene annotations were provided to STAR to improve the accuracy of mapping. Quality control on both raw reads and adaptor-trimmed reads was performed using FastQC (v0.11.9). FeatureCounts (v2.0.2) (26) was used to count the number of mapped reads to each gene. Significantly differential expressed genes with absolute fold change >= 1.5 and FDR adjusted p value <= 0.05 were detected by DESeq2 (27).

Quantitative PCR

For plasma and HDL sRNA qPCR total RNA was extracted from 200 μl of plasma or 500 μg of purified HDL by protein measurement as described above, except that a cocktail of three sRNA mimic standards were added after the initial lysis step. RNA was polyadenylated and converted to first-strand cDNA using qScript microRNA cDNA Synthesis kit (QuantaBio). Custom sRNA forward primers were designed for miR-1246, 5’-AATGGATTTTTGGAGCAGGA-3’; cel-miR-39-3p, 5’-TCACCGGGTGTAAATCAGCTTGA-3’; cel-miR-54-3p, 5’-CCCGTAATCTTCATAATCCGAGA-3’; and cel-miR-238-3p, 5’-GTACTCCGATGCCATTCAGA-3’. PerfeCTa Universal PCR primer (QuantaBio) was used as reverse primer. A DNA mimic of known concentration was used in a standard curve in each plate to derive miRNA concentrations.

For qPCR of cellular mature miRNAs, primary and precursor miRNAs, total RNA was extracted from THP-1 cells as described above. RNA was polyadenylated and converted to first-strand cDNA using qScript microRNA cDNA Synthesis kit (QuantaBio) or Mir-X^™ miRNA First Strand Synthesis Kit (TaKaRa). Custom sRNA forward primers were designed for pre-mir-1246, 5’-GGAAATCAATCCATAGGCTAGCA-3’ and miR-1246 (as above). PerfeCTa Universal PCR primer (QuantaBio) or mRQ 3’ primer (TaKaRa) was used as reverse primer. Pri-mir-1246 has both custom forward and reverse primers: forward 5’-TAGGACTGGGCAGAGATAAG-3’ and reverse 5’-TGATTGCTAGCCTATGGATTG-3’. SNORD44 (QuantaBio) or U6 (TaKaRa) was used as internal control.

For qPCR of cellular mRNA, total RNA was extracted from cells and converted to first strand cDNA using QuantiTect Rev Transcription kit for mRNA cDNA (Qiagen) or qScript microRNA cDNA Synthesis kit (QuantaBio). mRNA primers are listed as: DUSP3, forward 5’-TGACCAGGCTTTGGCTCAAAA-3’ and reverse 5’-TCACGGTTCTGCCTCACGAT-3’; and GAPDH, forward 5’-GAAGGTGAAGGTCGGAGTC-3’ and reverse 5’-GAAGATGGTGATGGGATTTC-3’. CDC25A primer is a commercialized primer purchased from IDT. IL6 primers are commercialized primers from BioRad.

Quantitative real-time PCR was performed using PerfeCTa SYBR green (Quanta Biosciences) on Bio-Rad CFX384 instrument. Data were analyzed by the 2^−ΔCt method except as detailed above for the plasma and HDL miRNA concentrations.

Luciferase Reporter Assay

The DUSP3 3’UTR sequence was synthesized and incorporated into pEZX-MT06 plasmid (GeneCopoeia) to make DUSP3-WT plasmid. Four DUSP3 3’UTR mutants were created using the Q5^® Site-Directed Mutagenesis Kit (NEB) by altering the putative target sequence, AAUCCAUA, to UCCAUA, AACAUA, AAUCUA, and AAUCCA, respectively. The mutants were generated using specific primers with forward and reverse sequences detailed: DUSP3-delAA, TCCATAGAGTTTTCAAAATGTGAATCTTT (forward) and TAGGAGGGTTCTCCAGGTTTG (reverse); DUSP3-delUC, CATAGAGTTTTCAAAATGTGAATCTTT (forward) and TTTAGGAGGGTTCTCCAGGTTT (reverse); DUSP3-delCA, TAGAGTTTTCAAAATGTGAATCTTTGG (forward) and GATTTAGGAGGGTTCTCCAGGT (reverse); DUSP3-delUA, GAGTTTTCAAAATGTGAATCTTTGG (forward) and TGGATTTAGGAGGGTTCTCCAG (reverse).

HEK293 cells were seeded in 24 well plates and allowed to attach overnight. The following day, the cells were co-transfected with 20 ng/μl plasmid DNA (DUSP3-wide type (WT) and 4 corresponding mutants: DUSP3-delAA, DUSP3-delUC, DUSP3-delCA, and DUSP3-delUA) and 2 μM miR-1246, co-transfected with 20 ng/μl DUSP3-WT and 2 μM negative control miRNA or transfected with 20 ng/μl DUSP3-WT only by DharmaFECT Duo (Horizon Discovery) following manufacturer’s protocol. Cells were collected 24 hours after treatment.

HEK293 cells were lysed using lysis buffer from Luc-Pair^™ Luciferase Assay Kit 2.0 (GeneCopoeia). The lysate placed on a gentle shaker for 10–15 min at room temperature (RT). The lysate was then transferred to an opaque 96-well plate for measurement of firefly luminescence and Renilla luminescence. Luciferase activity was calculated as the ratio of normalized firefly luminescence/Renilla luminescence for each sample well.

General Statistics

The sample size for this study was based on fold difference in HDL-miRNA cargo between RA and control subjects. A total of 30 RA and 30 control subjects gives over 80% power based on Wald test and adjusted for 5% false discovery rate to detect all HDL-miRNAs with ≥ 2-fold difference in RA versus control subjects. This is based on preliminary data that about approximately 90–100 human miRNAs are detected on HDL and an expectation that about 10 HDL-miRNAs would be altered with at least 30 read counts per miRNA. This was calculated using RNAseqPS(28). Data analysis and figure preparation were conducted using IBM SPSS Statistics (Version 27) and GraphPad Prism version 9.4.1. Data were presented as mean ± standard deviation (SD), number (percent). Continuous data were compared using unpaired t-test or Mann-Whitney U test depending on the distribution. Categorical data were compared using Chi Square test. The independent association between RA and elevated HDL-miRNA concentration was examined with linear regression adjusting for the presence of ischemic heart disease and diabetes mellitus type 2 and the log-transformed HDL-miRNA concentration to normalize residuals. Spearman correlation was used to assess the relation between delivered miRNA and responding mRNA.

RESULTS

Subject characteristics

The study included 30 patients with RA and 30 control subjects. The groups were of similar age, race, and sex. Mean age was similar between RA and control subjects (56 years versus 55 years). Subjects were predominately White and non-Hispanic or Latino/a. Percentage of females was similar between groups (47% for RA and 57% for control). Among patients with RA, the average DAS28 CRP score was 3.2 units, which is low-to-moderate disease activity (Table). Serum HDL and other lipid panel measurements were not significantly altered in RA versus control subjects (Table). Significantly more patients with RA had ischemic heart disease and diabetes mellitus type 2 compared to control subjects (Table).

Table.

Subject demographics

	Control (n=30)	RA (n=30)	P
Age, years	59 [43, 65]	59 [50, 67]	0.55
Race			0.83
White, #	27 (90)	25 (83)
Black or African American, #	2 (7)	3 (10)
Native American or ≥ 1 race, #	1 (3)	2 (6)
Ethnicity, # Hispanic or Latino/a	0 (0)	1 (3)	-
Sex, # female	17 (57)	14 (47)	0.61
Total Cholesterol, mg/dl	180 [159, 203]	168 [138, 199]	0.12
Triglycerides, mg/dl	88 [56, 123]	106 [72, 137]	0.14
HDL-Cholesterol, mg/dl	59 [50, 75]	52 [46, 67]	0.11
LDL-Cholesterol, mg/dl	98 [84, 115]	87 [66, 123]	0.21
Ischemic heart disease, #	0 (0)	7 (23)	0.01
Diabetes mellitus type 2, #	0 (0)	8 (27)	0.005
DAS28 CRP, units	-	2.93 [2.27, 4.18]	-
Swollen joints, #	-	1 [0, 3]	-
Tender joints, #	-	3 [0, 7]	-
hsCRP, mg/l	1.35 [0.68, 3.18]	3.55 [1.80, 8.55]	0.002
Methotrexate use, #	0 (0)	20 (67)	-
Leflunomide use, #	0 (0)	4 (13)	-
Sulfasalazine use, #	0 (0)	3 (10)	-
Hydroxychloroquine use, #	0 (0)	7 (23)	-
Anti-TNFα use, #	0 (0)	15 (50)	-
Abatacept use, #	0 (0)	2 (7)	-
Rituximab use, #	0 (0)	2 (7)	-
Any Biologic use, #	0 (0)	19 (63)	-
JAK inhibitor use, #	0 (0)	1 (3)	-
Corticosteroid use, # current	0 (0)	20 (67)	-
NSAID use, # current	11 (37)	13 (43)	0.60

Data presented as median [interquartile range] or number (%). Analyses by Mann-Whitney U test or Chi Squared test. DAS28 CRP = disease activity score based on 28 joints and CRP. hsCRP= high sensitivity C-reactive protein. TNF= tumor necrosis factor. JAK = janus kinase. NSAID = non-steroidal anti-inflammatory drug.

miR-1246 is enriched in RA HDL

HDL-miRNA cargo was screened using next generation sequencing (NGS). Ninety-one miRNAs were observed on HDL from patients with RA and control subjects (Supplemental Table 1). The most abundant HDL-miRNA was miR-1246 (baseMean [average of normalized counts of all RA and control subjects] = 5781). In contrast, miR-1246 was not abundant compared to other miRNAs in plasma, ranking 195 most abundant with baseMean = 212 (though direct comparisons in baseMean between HDL and plasma miRNAs are imprecise given normalization strategies) in plasma sequenced from the same subjects (Supplemental Table 2). This suggests that HDL is the main carrier of miR-1246 in plasma.

Results were validated by qPCR; miR-1246 was approximately 2-fold enriched in RA versus control HDL (mean= 4937 zmole/μg HDL versus 2417 zmole/μg HDL, P=0.02) (Figure 1a), but not significantly altered in RA versus control plasma (mean= 193 zmole/μl plasma versus 289 zmole/μl plasma, P=0.15) (Figure 1b). Prior studies demonstrated altered HDL-miRNA cargo in the setting of heart disease and diabetes mellitus type 2 which are increased in patients with RA; however, HDL-miR-1246 remained significantly elevated in patients with RA after adjustment for ischemic heart disease and diabetes mellitus type 2 (P=0.04).

Figure 1.

miR-1246 is enriched in plasma-derived HDL but not plasma of patients with RA compared to control subjects. (a) miR-1246 concentration in purified HDL is increased 2-fold, P= 0.02 in RA versus control subjects. (b) miR-1246 concentration in plasma is similar in RA versus control subjects P=0.16. miR-1246 concentration was assessed by qPCR. Data are presented as mean ± SD. *=P<0.05, P>0.05 is designated as not significant (ns).

miR-1246 increases proinflammatory cytokine expression

The function of miR-1246 was examined in macrophages because of their role as major producers of proinflammatory cytokines in RA (29) and within the atherosclerotic plaque (30). There is very low baseline miR-1246 expression in human THP-1 monocyte-derived macrophages, but transfection of a miR-1246 mimic increased detection of miR-1246 (Figure 2a) and increased LPS-induced IL6 mRNA expression 8.6-fold (P=0.0003), (Figure 2b) and secreted IL-6 1.9-fold (P=0.016) (Figure 2c). Transfection of miR-1246 antagomir, which should bind to the endogenous cellular miR-1246, had no significant effect on IL6 mRNA expression (Figure 2b).

Figure 2.

miR-1246 increases expression of IL6 mRNA and secreted IL-6 while decreasing expression of CDC25A mRNA and DUSP3 mRNA in macrophages. LPS-activated THP-1 macrophages were transfected with miR-1246 mimic, miR-1246 antagomir or control miRNA cel-miR-67. (a) Transfection efficiency is shown. miR-1246 increased expression of (b) IL6 mRNA by qPCR and (c) secreted IL-6 by ELISA, and decreased expression of (d,e) DUSP3 mRNA and (d,f) CDC25A mRNA by NGS and qPCR. There was no significant effect of cel-miR-67 on outcomes. Data are presented as mean ± SD. *, P <0.05; **, P <0.01; ***, P <0.001; ns, not significant. All qPCR experiments were performed three times, and representative experiments are shown.

miR-1246 directly targets DUSP3 leading to increased IL6 expression

The effect of miR-1246 on macrophage gene expression was assessed by NGS to identify potential targets of miR-1246 which may increase proinflammatory cytokine expression. Because miRNAs typically downregulate their target mRNAs, downregulated genes were examined. A total of 116 mRNAs were decreased after miR-1246 mimic transfection versus vehicle control. Among these 45 had cumulative weighted context score of −0.2 or less (Figure 2d, Supplemental Table 3), which is considered high to moderate confidence for miRNA targeting. Among the 45 mRNAs, 11 (CCNG2, CDC25A, CREBL2, DENND2D, DSCC1, DUSP3, FAM53C, GOLT1B, PCBP2, SLC12A2, and VIM) had literature support as miR-1246 targets and/or had >1.5-fold reduction after miR-1246 transfection. Among these 11 transcripts, qPCR expression of CDC25A mRNA and DUSP3 mRNA were most significantly reduced by miR-1246 in validation experiments (Figure 2e, f).

DUSP3 siRNA knockdown (Figure 3a) increased LPS-induced IL6 mRNA expression 2.9-fold (P=0.005) (Figure 3b) and secreted IL-6 1.9-fold (P=0.01) (Figure 3c) in macrophages, while CDC25A has no significant effect on IL6 mRNA expression (Supplemental Figure 1). To test whether DUSP3 expression is regulated post-transcriptionally by miR-1246, the entire 3’ UTR from DUSP3 was fused to reporter constructs that contain the firefly and renilla luciferase genes as the DUSP3-wild type (WT) plasmid. Four different mutant constructs were made with two nucleotide deletion within target site (Figure 3d). Co-expression of miR-1246 with DUSP3-WT significantly reduced luciferase activity (Figure 3e). In contrast, four different DUSP3 mutations within the putative target sequence for miR-1246 completely abrogated the negative effect of miR-1246 on expression of DUSP3 (Figure 3e). These data indicate that DUSP3 is a direct target of miR-1246.

Figure 3.

miR-1246 directly targets DUSP3 to increase IL6 expression. The DUSP3 siRNA decreased DUSP3 (a) and increased IL6 expression (b) by qPCR, as well as increased secreted IL-6 (c) by ELISA. (d) A schematic shows the sequence alignment of WT and four different mutants within the putative miR-1246 target sites on the 3’UTR of DUSP3 used for the luciferase assay. (e) miR-1246’s ability to directly target DUSP3 is demonstrated by luciferase assay. A reduction in firefly/renilla activity indicates binding of miR-1246 to the WT 3’UTR of DUSP3. No firefly/renilla activity change is detected in mutated 3’UTR of DUSP3. All qPCR experiments and luciferase assay were repeated three times, and representative results are shown. Data are presented as mean (SD). **, P <0.01; ***, P <0.001; ****, P <0.0001.

HDL delivers miR-1246 leading to increased IL6 expression

Control subject HDL was loaded with miR-1246 to investigate HDL’s ability to deliver miR-1246 to cells. Macrophages were activated with LPS then washed prior to treatment with HDL because HDL is capable of binding LPS (31), and could therefore confound the results. The HDL-miR-1246 complex versus unmodified control HDL delivered miR-1246 (Figure 4a), decreased DUSP3 mRNA expression 2-fold (P= 0.006) (Figure 4b), and increased IL6 mRNA expression 5-fold (P=0.03) (Figure 4c) and secreted IL-6 2.4-fold (P=0.004) in the activated macrophages (Figure 4d). There was no significant effect of HDL delivery of cel-miR-67 on outcomes.

Figure 4.

HDL delivers miR-1246 to increases IL6 expression in macrophages. LPS-activated THP-1 macrophages were incubated with control HDL loaded with or without miR-1246 (LPS->HDL-miR-1246 and LPS->HDL, respectively) or with control miRNA cel-miR-67 for 24 hours. Control HDL loaded with miR-1246 delivered significantly more miR-1246 (a), decreased DUSP3 mRNA (b), and increased IL6 mRNA (c) significantly more than unloaded HDL based on qPCR, as well as increased secreted IL-6 (d) significantly more based on ELISA. There was no significant effect of cel-miR-67 on outcomes. Each experiment was repeated three times, and representative results are shown. Data are presented as mean (SD). **, P <0.01; ***, P <0.001.

RA HDL-miR-1246 increases IL6 expression

Highly purified HDL from 6 RA and 6 control subjects were used to examine native HDL’s delivery of miR-1246 and effect on IL6. Subject demographics are available in Supplemental Table 4. Activated macrophages treated with unmodified, purified HDL from RA versus control subjects increased cellular miR-1246 1.6-fold (P=0.047) (Figure 5a), while there was no significant change in pri-mir-1246 (Figure 5b) and pre-mir-1246 expression (Figure 5c). This indicates that the increase in cellular miR-1246 was due to HDL delivery of the naturally occurring miR-1246 it was carrying, and not increased transcription. Unmodified purified HDL from both RA and controls subjects decreased activated macrophage IL-6 expression, but compared to control HDL cells treated with the RA HDL had 3.4-fold (P=0.03) increased IL-6 expression (Figure 5d) and decreased DUSP3 expression 1.3-fold mRNA (P=0.0007) (Figure 5e). Delivery of miR-1246 was significantly associated with IL6 mRNA expression (Rho=0.76, P=0.006; Figure 5f). These findings together indicate that RA HDL’s anti-inflammatory effect is blunted compared to control HDL due to miR-1246 delivery to macrophages.

Figure 5.

RA HDL increased IL6 expression. LPS-activated THP-1 macrophages were incubated with purified control or RA HDL (370 μg protein) for 24 hours. RA HDL delivered significantly more miR-1246 than control HDL (a), pri-mir-1246 and pre-mir-1246 were not increased with control or RA HDL treatment, showing the increase in miR-1246 after treatment was due to delivery rather than de novo transcription (b and c), the typical anti-inflammatory effect of HDL on IL6 expression was significantly dampened by RA HDL treatment compared to control treatment (d), accompanied by DUSP3 expression reduction (e). Expression was assessed by qPCR. These data are presented as mean (SD). *, P <0.05; ***, P <0.001; ns, not significant. Cellular miR-1246 after HDL treatment and IL6 were significantly correlated based on Spearman correlation (f).

HDL-miR-1246 relationship with RA disease features

HDL-miR-1246 was approximately 2-fold higher in those with active disease (DAS28-CRP ≥ 2.6, mean HDL-miR-1246 = 5421 zmole/ug HDL) compared to those in remission (DAS28-CRP<2.6, mean HDL-miR-1246 = 2606 zmole/ug HDL, P=0.03) (Supplemental Figure 2). There was a modest, but non-significant linear correlation with DAS28-CRP (Rho=0.28, P=0.14) and hs-CRP (Rho=0.18, P=0.37). Patients with RA taking glucocorticoids had slightly lower HDL-miR-1246 concentration (users = 3981 zmole/ug HDL) compared with non-users (non-users = 4742 zmole/ug HDL, P=0.04). There was no difference in HDL-miR-1246 concentration among RA patients taking NSAIDs (P=0.75), methotrexate (P=0.053), leflunomide (P=0.19), sulfasalazine (P=0.28), hydroxychloroquine (P=0.28), or any biologic agent (P=0.96) compared to non-users of the agent.

DISCUSSION

To our knowledge, this is the first study to examine HDL-miRNA cargo and its effect on inflammation in patients with RA. There are several major findings of this study. We identified that HDL is a main carrier of miR-1246 in plasma, that miR-1246 is highly abundant on HDL compared to other miRNAs, and that miR-1246 is enriched in RA versus control HDL. We found that compared to controls, RA HDL delivers more miR-1246 to activated macrophages leading to amplified IL-6 expression through targeting DUSP3 and that this is a mechanism behind HDL’s blunted anti-inflammatory effect in RA compared to control subjects.

HDL-miRNA cargo has not been studied widely, but is altered in several other disease states such as familial hyperlipidemia(13, 32), coronary artery disease (15), and diabetes mellitus type 2 (33). HDL can accept miRNAs and other small RNAs from many cell types such as macrophages, neutrophils, neurons, and pancreatic beta cells (34), but the export mechanism for specific miRNAs is not clear. Like this study, others have shown that delivery of HDL-miRNA cargo can regulate recipient cell gene expression. For example, we previously showed that HDL can deliver miR-223 to human coronary artery endothelial cells leading to decreased inflammatory response by inhibiting expression of intracellular adhesion molecule 1 (ICAM-1) (14). Others showed that HDL from patients with coronary artery disease versus control subjects was enriched in miR-24-3p, and that HDL mediated delivery of miR-24-3p to human umbilical vein endothelial cells downregulates vinculin leading to increased production of proinflammatory reactive oxygen species (15).

It is unclear which cells or conditions contributed to the increased miR-1246 on RA HDL. The majority of the literature relevant to miR-1246 relates to cancer. One such study demonstrated that in the setting of hypoxia, which is found in the tumor environment, miR-1246 transcription and export from glioma cells increased and hypoxia inducible factor – 1 alpha (HIF-1α) played a major role in regulating these processes (35). The RA synovium is a hypoxic environment (36) and consequently there is increased expression of HIFs in synovial tissue from RA versus osteoarthritis patients(36–38). Specifically, HIF-1α is highly expressed in the intimal layer of the RA synovium (39), which could contribute to increased miR-1246 export. Because HDL is enriched in synovial fluid from RA compared to osteoarthritis patients (40), it is possible that RA HDL could take up miR-1246 within the joint, the main target organ in RA. Futures studies are needed to further evaluate this.

Regardless of the mechanism for increased miR-1246 on HDL, HDL’s increased delivery of miR-1246 to macrophages increasing IL-6 expression has important implications for RA. miR-1246 has proinflammatory effects in several key cells in RA. For example, extracellular vesicle-mediated miR-1246 delivery increases inflammation in condylar chondrocytes by targeting GSK3β and Axin2 (41). Exosomes derived from human colon cancer tumorspheres are abundant in miR-1246 and deliver miR-1246 to fibroblasts leading to increased IL-6 secretion (42). On a larger scale HDL-miRNAs may have an important role in RA. Critical cells for RA pathogenesis such as macrophages, B cells, T cells, RA synovial fibroblasts and others express scavenger receptor class B type 1 (SR-B1) (43–46), which may be required for HDL-miRNA delivery (13). Thus, more studies are needed to determine the effect of HDL-miRNA delivery on other key cells in RA, and the potential for HDL-miRNA targeting for RA treatment.

This work has important implications for the accelerated atherosclerosis in RA. It demonstrates that HDL, which is thought to be cardioprotective, is abnormal in RA. This builds on prior studies of HDL function in RA which focused mostly on HDL’s cholesterol efflux and antioxidant capacities (5). HDL cholesterol efflux capacity is typically tested as HDL’s or HDL enriched serum’s ability to remove cholesterol from lipid-laden macrophages (5). It remains controversial whether cholesterol efflux capacity of HDL is altered in RA. For example, we and others previously showed that cholesterol efflux capacity is similar in RA and control subjects, but among RA patients with high disease activity it is impaired compared to low disease activity or remission (6, 7). However, others have shown decreased cholesterol efflux capacity in RA (8, 9), and inverse association with plaque burden and cardiovascular events (47). Differences in study findings may be due to the different populations tested and variations in methodology (5).

RA HDL’s antioxidant capacity has been tested as HDL or HDL-enriched serum’s ability to prevent oxidation of LDL but has sometimes also been called its anti-inflammatory capacity (5). This is not to be confused with what we have called the anti-inflammatory effect of HDL in the current study in which we tested the effect of HDL on macrophage proinflammatory cytokine production, or as others have measured as the ability of HDL to suppress TNF-alpha induced vascular cell adhesion molecule-1 (VCAM-1) (48).

Studies of RA HDL’s antioxidant capacity are conflicting with some showing impaired HDL antioxidant capacity in RA (10–12) and another showing similar antioxidant capacity in RA versus control subjects(49). A prior study showed that RA HDL with impaired antioxidant capacity compared to HDL with normal antioxidant capacity has altered protein cargo, such as higher levels myeloperoxidase, which generates reactive oxygen species (11). Just as HDL’s protein cargo is altered in RA patients with impaired antioxidant capacity, we found HDL’s miRNA cargo is altered in patients with RA leading to impaired anti-inflammatory capacity.

A limitation of this study is that we have a relatively small sample size for both profiling the HDL-miRNAs and studying native HDL delivery of miR-1246. This has been a limitation for other prior HDL-miRNA studies with sample sizes ranging from 5 to 10 subjects per group in some studies (13, 33, 50). Though there have been two larger studies with 32 to 95 subjects per group (15, 32). As opposed to some studies examining HDL function using HDL enriched serum (i.e., apolipoprotein B depleted serum), which would still contain extracellular vesicles and other miRNA carriers, we needed to study highly purified HDL. Preparing the purified HDL requires a large volume of plasma due to losses during processing, limiting the number of samples available to examine, but it provides a purer evaluation of HDL function. In the experiments testing native HDL’s effect on macrophage IL-6 expression, we have not accounted for the effect of other HDL-miRNAs as well as other HDL properties. However, testing the control HDL loaded with and without miR-1246, demonstrates the effect of the delivered miR-1246 in a more controlled experiment. We also only examined the impact of HDL-miR-1246 on macrophage cytokine expression, other functions will be of interest to examine in future studies.

This study demonstrates a novel, targetable function of HDL that is abnormal in RA and could contribute to excess inflammation and atherosclerosis observed in the disease. Future studies will focus on consequences of RA’s altered HDL-miRNA cargo on disease activity and atherosclerosis in vivo.