Deep Sequencing of Serum Small RNAs Identifies Patterns of 5' tRNA Half and YRNA Fragment Expression Associated with Breast Cancer

Abstract

Small noncoding RNAs circulating in the blood may serve as signaling molecules because of their ability to carry out a variety of cellular functions. We have previously described tRNA- and YRNA-derived small RNAs circulating as components of larger complexes in the blood of humans and mice; the characteristics of these small RNAs imply specific processing, secretion, and physiological regulation. In this study, we have asked if changes in the serum abundance of these tRNA and YRNA fragments are associated with a diagnosis of cancer. We used deep sequencing and informatics analysis to catalog small RNAs in the sera of breast cancer cases and normal controls. 5' tRNA halves and YRNA fragments are abundant in both groups, but we found that a breast cancer diagnosis is associated with changes in levels of specific subtypes. This prompted us to look at existing sequence datasets of serum small RNAs from 42 breast cancer cases, taken at the time of diagnosis. We find significant changes in the levels of specific 5' tRNA halves and YRNA fragments associated with clinicopathologic characteristics of the cancer. Although these findings do not establish causality, they suggest that circulating 5' tRNA halves and YRNA fragments with known cellular functions may participate in breast cancer syndromes and have potential as circulating biomarkers. Larger studies with multiple types of cancer are needed to adequately evaluate their potential use for the development of noninvasive cancer screening.

Keywords: Y RNA; YRNA fragments; breast cancer; circulating small RNAs; serum; tRNA derivatives; tRNA halves.

Figure 1

Length distribution and annotation of reads from serum small RNAs of five healthy females and five females diagnosed with breast cancer. Length of mapped reads from five normal (A) and five cancer samples (B) is plotted against read abundance. Colors of the bars denote the individual sources of the serum small RNAs (normal and breast cancer samples). (C) Length distribution of pooled sequencing reads from all 10 samples plotted against abundance of reads according to their annotation. Length distribution is plotted against abundance of the reads annotated as miRNAs, YRNAs, tRNAs, rRNAs, or other sRNAs (snRNAs and snoRNAs). (D) Pie chart showing the percent of reads (pooled from both normal and breast cancer) mapping to the indicated specific types of small RNAs. (E) Pie chart showing the percent of reads mapping to the indicated types of small RNAs in pooled datasets obtained by sequencing of small RNAs in the sera of 42 newly diagnosed breast cancer cases.

Figure 2

Characteristics of sequencing reads that map to YRNA genes. (A) Pooled reads from all 10 samples were used to compare the number of reads that align with 5′ or 3′ ends of YRNA genes. The y-axis represents the percentage of the total YRNA reads that map to YRNA genes. (B) Percentage of reads mapping to 5′ and 3′ ends of YRNAs in the 25–29 nt peak (black) and the 30–33 nt peak (gray). Almost 100% of the reads in the 30–33 nt peak align to the 5′ end of YRNAs, while a significant majority of reads in the 25–29 nt peak align to the 3′ end of YRNAs. (C) Northern blot analysis of RNA extracted from 0.2 mL of human serum. The blot was hybridized to ³²P-end-labeled oligonucleotide probes complementary to the 5′ end (left panel) or 3′ end (right panel) of RNY4. Arrows indicate the positions of the major (30–33 nt) and minor (25–29 nt) bands. M: Decade markers.

Figure 3

Changes in serum abundance of specific 5′ tRNA halves in breast cancer cases sorted by clinicopathologic characteristics. Shown in the graph are all 5′ tRNA halves found to have significant differences in abundance when 42 newly diagnosed breast cancer cases were compared for clinicopathologic characteristics. The y-axis denotes fold change in abundance of a 5′ tRNA half when groups are compared. Eventual relapse is associated with increased serum levels of a small set of 5′ tRNA halves, while ER expression is associated with decreased serum levels of a broader group of 5′ tRNA halves. Inflammatory disease is associated with increased serum levels of a separate group of 5′ tRNA halves. Tumor stage is not shown because no significant differences were found between Stage 2 and Stage 3 cases. An adjusted P-value <0.05 was set as the level of statistical significance for the difference between two possible alternatives of each clinical characteristic (stage [2 versus 3], ER [positive versus negative], PR [positive versus negative], HER2 [positive versus negative], presence of inflammation [yes versus no], and subsequent relapse [yes versus no]).

Figure 4

Changes in serum abundance of specific YRNA fragments in breast cancer cases sorted by clinicopathologic characteristics. Shown in the graph are all YRNA fragments found to have significant differences in abundance when 42 newly diagnosed breast cancer cases were compared for clinicopathologic characteristics. The y-axis denotes fold change in the indicated YRNA fragments. Eventual relapse was associated with increased serum levels of a broad group of YRNA fragments, while ER expression was associated with decreased serum levels of an overlapping group. Inflammatory status is not shown because no significant differences were found. An adjusted P-value <0.05 was set as the level of statistical significance for the difference between two possible alternatives of each clinical characteristic (stage [2 versus 3], ER [positive versus negative], PR [positive versus negative], HER2 [positive versus negative], presence of inflammation [yes versus no], and subsequent relapse [yes versus no]).

Figure 5

Inter-individual variation in expression of 5′ tRNA halves. Expression levels of the indicated 5′ tRNA halves are represented as counts per million reads (cpm) in each individual breast cancer case. For each group, the mean (dotted line) and the standard error of the mean (error bars) are indicated. Comparisons were carried out between ER− and ER+ groups (A), and relapsed and nonrelapsed groups (B). Each tRNA isoacceptor is identified by its genomic position in the human hg19 genome. In (B), the two individuals showing increased levels of different 5′ tRNA halves are the same in each graph.