MicroRNA expression profiles predictive of human renal allograft status

Abstract

Immune rejection of organ transplants is a life-threatening complication and is exemplified by alterations in the expression of protein-encoding genes. Because microRNAs (miRNAs) regulate the expression of genes implicated in adaptive immunity, we investigated whether acute rejection (AR) is associated with alterations in miRNA expression within allografts and whether expression profiles are diagnostic of AR and predict allograft function. Seven of 33 renal allograft biopsies (12 AR and 21 normal) were profiled using microfluidic cards containing 365 mature human miRNAs (training set), and a subset of differentially expressed miRNAs were quantified in the remaining 26 allograft biopsies (validation set). We found a strong association between intragraft expression of miRNAs and messenger RNAs (mRNAs), and that AR, and renal allograft function, could be predicted with a high level of precision using intragraft levels of miRNAs. Our investigation of miRNA expression in normal human peripheral blood mononuclear cells (PBMCs) showed that miRNAs (miR-142-5p, -155, and -223) overexpressed in AR biopsies are highly expressed in PBMCs, and that stimulation with the mitogen phytohaemagglutinin results in an increase in the abundance of miR-155 and a decrease in miR-223 and let-7c. Quantification of miRNAs in primary cultures of human renal epithelial cells (HRECs) showed that miR-30a-3p, -10b, and let-7c are highly expressed in HRECs, and that stimulation results in a decreased expression of miR-30a-3p. Our studies, in addition to suggesting a cellular basis for the altered intragraft expression of miRNAs, propose that miRNA expression patterns may serve as biomarkers of human renal allograft status.

Fig. 1.

Unsupervised hierarchical clustering and principal component analysis of miRNA expression profiles differentiate acute rejection biopsies from normal allograft biopsies of human renal allografts. (A) MicroRNA (miRNA) expression patterns of 7 human kidney allograft biopsies [3 showing histological features of acute rejection (AR) and 4 with normal allograft biopsy results (N)] were examined using microfluidic cards containing TaqMan probes and primer pairs for 365 human mature miRNAs. A total of 174 ± 7 miRNAs were expressed at a significant level (i.e., C_T < 35) in all samples. Gender, age, ethnicity, type of transplantation, and time from transplantation to biopsy were as follow: AR1 (Male, Black, 52 years, living donor, 161 days), AR2 (male, White, 32 years, living donor, 119 days), AR3 (Female, White, 48 years, deceased donor, 31 days), N1 (female, Black, 40 years, living donor, 203 days), N2 (Male, Indian, 50 years, living donor, 191 days), N3 (Male, Black, 31 years, living donor, 196 days), and N4 (Male, Asian, 51 years, deceased donor, 88 days). The biopsies were grouped by unsupervised hierarchical clustering on the basis of similarity in expression patterns. The degree of relatedness of the expression patterns in biopsy samples is represented by the dendrogram at the top of the panel. Branch lengths represent the degree of similarity between individual samples (Top) or miRNA (Left). Two major clusters (Top) accurately divided AR biopsies from normal allograft biopsies. Each column corresponds to the expression profile of a renal allograft biopsy, and each row corresponds to a miRNA. The color in each cell reflects the level of expression of the corresponding miRNA in the corresponding sample, relative to its mean level of expression in the entire set of biopsy samples. The increasing intensities of red mean that a specific miRNA has a higher expression in the given sample and the increasing intensities of green mean that this miRNA has a lower expression. The scale (Bottom Right) reflects miRNA abundance ratio in a given sample relative to the mean level for all samples. (B) Principal component analysis of 7 kidney allograft biopsies based on the expression of 174 small RNAs significantly expressed (i.e., C_T < 35) in all of the samples. PCA is a bilinear decomposition method designed to reduce the dimensionality of multivariable systems and used for overviewing clusters within multivariate data. It transforms a number of correlated variables into a smaller number of uncorrelated variables called principal components (PC). The first PC accounts for as much of the variability in the data as possible, and each succeeding component accounts for as much of the remaining variability as possible. PCA showed evident clustering and confirmed the separation of AR samples from normal allograft biopsies. Samples were accurately grouped by PC1, which explained 45.91% of the overall miRNA expression variability, whereas PC2 explained 21.48% of variability and did not classify the samples according to their diagnosis.

Fig. 2.

Differential expression of miRNAs in acute rejection biopsies and normal allograft biopsies at a P value <0.05. MicroRNA (miRNA) expression patterns of 7 human kidney allograft biopsies [3 showing histological features of acute rejection (AR) and 4 with normal allograft biopsy results (N)] were examined using microfluidic cards containing TaqMan probes and primer pairs for 365 human mature miRNAs. Each column corresponds to the expression profile of a renal allograft biopsy, and each row corresponds to a miRNA. ABqPCR software was used to identify miRNAs that were differentially expressed between AR biopsies and normal allograft biopsies. C_T filtering procedure was first performed. Assays with a C_T value >35 in >50% of samples in each group were called undetected. Assays that were not detected in both groups were not included in the analysis. For the remaining assays, t test was used to detect differentially expressed miRNAs. The miRNA clustering tree is shown on the Left. Branch lengths represent the degree of similarity between individual miRNAs. The higher intensities of red mean higher expression level.

Fig. 3.

Validation of differential expression of microRNAs in AR biopsies and normal allograft biopsies of human renal allografts. Intragraft expression levels of miR-142–5p, -155, -223, -10b, -30a-3p, and let-7c in an independent validation set of 9 acute rejection biopsies and 17 normal kidney allograft biopsies. Expression levels were quantified using modified TaqMan miRNA assays that allow absolute quantification of miRNAs (see SI Text for full details). miRNA copy numbers were normalized using the stably expressed RNU44 small nucleolar RNA, and are shown as mean (± SE) ratio of miRNA copies to RNU44 copy numbers. RNU44 copy numbers were not different between the 9 acute rejection biopsies (8.87 × 10⁶ ± 1.48 × 10⁶ copies/μg RNA) and the 17 normal allograft biopsies (8.72 × 10⁶ ± 8.42 × 10⁵ copies/μg RNA, P = 0.92). P value calculated using t test.

Fig. 4.

Positive association between miRNAs and mRNAs in human allograft biopsies. Intragraft levels of miRNAs were quantified with the use of TaqMan miRNA assays, and intragraft levels of mRNAs were quantified using real time quantitative PCR assays, and the relationship between the intragraft levels of miRNA and mRNA is shown, along with Pearson correlation (R²) and P values. A strong positive association between the levels of CD3 mRNA and the levels of miRNAs overexpressed in acute rejection biopsies was found: (A) miR-142–5p (R² = 0.72, P < 0.0001); (B) miR-155 (R² = 0.69, P < 0.0001); or (C) miR-223 (R² = 0.66, P < 0.0001). A positive association between renal tubule specific NKCC-2 mRNA and miRNAs underexpressed in acute rejection biopsies was also observed: (E) miR-30a-3p (R² = 0.53, P < 0.0001); (F) miR-10b (R² = 0.36, P < 0.0001); or (G) let-7c (R² = 0.13, P = 0.04). Results from all 33 renal allograft biopsies (red, 12 acute rejection biopsies; green, 21 normal allograft biopsies) are shown. The threshold cycle (C_T) is the fractional cycle number at which the fluorescence crossed the fixed threshold in miRNA/mRNA assays. (D) The mean (± SD) C_T values of the endogenous control for mRNAs (18S rRNA, 24.8 ± 1.3 vs. 24.7 ± 1.1, P = 0.86, t test) and (H) for miRNAs (RNU44 small nucleolar RNA, 27.1 ± 0.7 vs. 27.1 ± 0.5, P = 0.97, t test) were similar between the acute rejection samples and the normal renal allografts.

Fig. 5.

Levels of miRNAs in resting or activated normal human peripheral blood mononuclear cells. Peripheral blood mononuclear cells (PBMCs) were obtained from healthy individuals and were incubated without (open bars) or with (filled bars) 2 μg/mL PHA for 24 h (A, F, and G) (n = 7 subjects), 48 h (D) (n = 4 subjects), or 24, 48, and 72 h (B, C, E, H, and I) (n = 2 subjects), and RNA was isolated for miRNA quantification (A–E) or mRNA quantification (F–I) (see SI Text for full details). miRNA copy numbers were normalized using the RNU44 small nucleolar RNA copy numbers and mRNA copy numbers were normalized using the 18S rRNA copy numbers and are shown as mean (± SE) ratio of miRNA copies to RNU44 copy numbers or ratio of mRNA copies to 18S rRNA copies. P value calculated using paired t test.

Fig. 6.

Levels of miRNAs in resting or activated normal human renal epithelial cells. Primary cultures of normal human renal epithelial cells (HRECs) were incubated for 24 h (A, B) or 24 and 48 h (C) with cell-free supernatants of resting PBMCs (open bars) or cell-free supernatants of PBMCs activated with 2 μg/mL PHA (filled bars). Total RNA was isolated from HRECs and a subset of miRNAs found to be overexpressed (A) or underexpressed (B and C) in acute rejection biopsies were quantified with the use of modified TaqMan miRNA assays (see SI Text for full details). miRNA copy numbers were normalized using the RNU44 small nucleolar RNA copy numbers and are shown as mean (± SE) ratio of miRNA copies to RNU44 copy numbers. Results are from 2 consecutive experiments with 2 independent primary cultures of HRECs developed from 2 human kidneys. P values calculated using paired t test.