Abnormalities of the large ribosomal subunit protein, Rpl35a, in Diamond-Blackfan anemia

Abstract

Diamond-Blackfan anemia (DBA) is an inherited bone marrow failure syndrome characterized by anemia, congenital abnormalities, and cancer predisposition. Small ribosomal subunit genes RPS19, RPS24, and RPS17 are mutated in approximately one-third of patients. We used a candidate gene strategy combining high-resolution genomic mapping and gene expression microarray in the analysis of 2 DBA patients with chromosome 3q deletions to identify RPL35A as a potential DBA gene. Sequence analysis of a cohort of DBA probands confirmed involvement RPL35A in DBA. shRNA inhibition shows that Rpl35a is essential for maturation of 28S and 5.8S rRNAs, 60S subunit biogenesis, normal proliferation, and cell survival. Analysis of pre-rRNA processing in primary DBA lymphoblastoid cell lines demonstrated similar alterations of large ribosomal subunit rRNA in both RPL35A-mutated and some RPL35A wild-type patients, suggesting additional large ribosomal subunit gene defects are likely present in some cases of DBA. These data demonstrate that alterations of large ribosomal subunit proteins cause DBA and support the hypothesis that DBA is primarily the result of altered ribosomal function. The results also establish that haploinsufficiency of large ribosomal subunit proteins contributes to bone marrow failure and potentially cancer predisposition.

Figure 1

Array CGH of 2 DBA patients with 3q deletions. (A) CGH performed on genomic DNA of EBV lines derived from 2 patients with DBA and chromosome 3q terminal deletions demonstrates single copy terminal deletions of chromosome 3q. The region (bracket) encompassing the smaller deletion (195.46Mb-qter) is enlarged in panel B. Bars to the left of data plots represent segments of copy number loss.

Figure 2

Chromosome 3q deletion mapping in 2 DBA probands. Genomic map of chromosome 3q deletions with combined deletion mapping and RNA expression microarray identifies 3q candidate genes for DBA. Vertical bars through the chromosome 3 map with G-banding pattern indicate the deleted region, which is enlarged above. Contig coverage is shown by horizontal lines below the chromosome ideogram. Black horizontal bars indicate the position of each deletion in relation to known genes on chromosome 3q28-ter. The larger deletion begins in intron 7 of LPP and spans approximately 10 megabases, including 3 contigs (NT_005612, NT_005535, NT_029928) with 2 gaps totaling an estimated 47 kb (NCBI Build 36.1). The smaller deletion begins in the intergenic region between HES1 and CPN2 and spans over 4 megabases. The position of a previously described 3q microdeletion syndrome, which was not associated with DBA or hematologic abnormalities, is also indicated. Regions of 3q which were either not involved in both DBA deletions or include the 3q microdeletion syndrome region (shaded gray) were considered improbable for candidate genes. Seven genes (noted in black), which lie in the intervals defined by these deletions, also demonstrated haploinsufficient expression in deletion 1 EBV LCL. RPL35A, a gene encoding a structural component of the large ribosomal subunit, is located within the extreme terminal region of the chromosome 3q deletions.

Figure 3

Genotype/phenotype analysis in patient DBA019. (A) RPL35A mutational analysis in this family showed 2 first-degree relatives, a father and sister, with the heterozygous RPL35A 97 G>A mutation. Both of these patients had macrocytic anemia suggestive of subclinical DBA carriers. Hemoglobin (Hb), mean red cell volume (MCV), and eADA activity are indicated; the hemoglobin value indicated for the proband is during treatment with steroids. Erythrocyte ADA was normal in all tested members of this pedigree. Black shading represents the DBA proband; white shading, hematologically unaffected members of the paternal lineage with normal RPL35A sequence; diagonal lines, deceased members with unknown RPL35A status; cross-hatched symbols, clinically normal persons who were not tested for RPL35A mutations. (B) 2 RPL35A RNA products were amplified from patient DBA019 using RT-PCR with primers designed to amplify the full-length RPL35A message. The shorter product results from an alternative splicing event between exons 3 and 4, leading to a truncated protein. The arrows above the transcript diagram show the normal splicing event, leading to wild-type RPL35A and removal of the approximately 2691 nucleotide intron. The lines above exon 3 indicate the wild-type codon sequence within exon 3; the 97 G/A mutation is indicated in gray. The arrows below the line demonstrate the abnormal splicing event. The 97 G>A mutation results in selection of a cryptic splice donor site within exon 3 immediately upstream of the change, causing removal of 70 base pairs of 3′ exon 3 coding sequence in addition to the intron. The predicted amino acid sequence of wild-type, simple amino-acid substitution, and the splicing variant are shown below.

Figure 4

shRNA directed against RPL35A mRNA causes decreased proliferation and apoptosis. (A) The efficacy of expression knockdown was assessed by real-time quantitative RT-PCR 4 days after transduction of UT-7/Epo or TF-1 cells with the lentiviral siRNA construct. Bar graph represents expression of RPL35A transcripts after normalization to GAPDH in each of 4 shRNA constructs compared with cells infected with a control shRNA-lentiviral construct targeting firefly luciferase. Bars represent the aggregated mean expression level from 5 independent experiments, 3 in TF-1 and 2 in UT-7/Epo. Error bars indicate SD. (B-E) UT-7/epo cell proliferation was assessed by quantitation of fluorescent dye DNA binding. GFP-positive UT-7/epo cells were sorted 3 days after lentiviral infection, subsequently plated (day 0) in 96-well plates in 5 replicates at 10³ cells/well, and assayed at the indicated times after sorting. Plating variation was corrected by adjusting the LOG2-transformed intensity at each day by the LOG2-transformed intensity on the initial day of plating. Curves represent the average of 4 platings from 2 independent experiments. Error bars indicate SDs. Differences between shLuc control-infected cells and RPL35A sh-2 were not significant. *sh-1, P < .05, compared with shLuc control cells. **sh-3 and 4, P < .01, compared with shLuc control cells. (F) Apoptosis was quantified by flow cytometric analysis after annexin V staining. GFP-positive UT-7/epo cells were sorted 3 days after infection and returned to culture for 24 hours. Cells were subsequently assessed on day 1 (top panel) and day 5 (bottom panel) after sorting. The histogram plots show cell counts versus annexin V intensity in shLuc (gray shading) and sh-3 (solid line) with apoptotic cells from each group enumerated above the gating threshold (M1). After sorting, no significant increase in annexin V–positive cells was discernable between control and sh-3–infected cells (K-S, P > .1). After adjustment for the proportion of annexin V–positive cells on day 1 (treatment group day 5 − treatment group day 1), RPL35A knockdown by day 5 resulted in a nearly 2.5-fold increase in annexin V–positive cells compared with controls (4.7% vs 11.7%; K-S, P < .001). Of note, a significant shift in the overall intensity of annexin V staining of the entire population of cells was also observed, suggesting that the overall effect on apoptosis was greater. Similar results were observed in sh-1– and sh-4–infected cells (not shown). Sh-Luc indicates Luciferase-control transduced cells; sh-1, -2, -3, and -4, respective RPL35A shRNAs.

Figure 5

Rpl35A is required for large ribosomal subunit assembly and pre-rRNA processing. (A) To identify potential sources of altered proliferation, polysome analysis was performed on UT-7/Epo (left column) and HEK293A (right column) cells infected with control (top row) or RPL35A (bottom row) sh-1. Cells were sorted 3 days after lentiviral infection and returned to media for 24 hours before treatment with cycloheximide and fractionation of lysates on a sucrose gradient. The 40S, 60S, and 80S peaks are indicated by; the polysome fraction lies below the horizontal line. In comparison to control shRNA-infected cells, cells infected with RPL35A shRNA demonstrated a decreased 40S:60/80S ratio, indicating a relative reduction of free 60S subunits. Similar results were seen in UT-7/Epo and HEK293A cells transduced with RPL35A sh-3 (not shown). Experiments were performed once with 2 different RPL35A shRNA for HEK293A cells and twice with 2 RPL35A shRNAs in UT-7/Epo cells. (B) Metabolic labeling of nascent RNA with ³²P was used to identify abnormalities of pre-rRNA processing. Lentivirus-infected cells sorted 6 days after infection were plated in phosphate-free media in 6-well plates for 2 hours before the addition of ³²P orthophosphate for 1 hour, washed, and then incubated in complete media for 4 hours. RNA was fractionated on 1.3% agarose/formaldehyde gels, dried, and autoradiographed. denotes the indicated mature and pre-rRNA species. RPL35A knockdown resulted in marked decrease of 32S, 28S, and 12S labeling without affecting mature 18S labeling. An increased exposure of the gel demonstrates reduced 7S and 5.8S rRNA. An ethidium bromide stain of the gel is also shown. (C) Intensity profile of lanes 1 and 2 demonstrates the reduction of 28S, preservation of 18S, and the appearance of a 41S band (shoulder adjacent to the 45S peak) not seen in control cells. Eth indicates ethidium bromide; sh-Luc, Luciferase-control transduced cells; sh-1, -2, -3, and -4, respective RPL35A shRNAs.

Figure 6

Rpl35A is required for pre-rRNA processing in ITS1 and ITs2. (A) A schematic of human pre-rRNA processing. Mature ribosomal RNA species are indicated by shaded boxes: ■, 18S; , 5.8S; □, 28S. External and internal transcribed spacers are indicated as lines between the mature species and labeled above the primary pre-rRNA transcript. Cleavage sites, as originally proposed by Hadjiolova et al, are shown by numbered arrows above the 45S and 45S′ transcripts. The sequence of cleavage of the 45S′ pre-RNA at sites 1 and 2 results in 2 alternative processing pathways. Two additional human cleavage sites (2b and 4a) shown as numbered arrows below the transcript are inferred from these studies. The presence of a 7S precursor to 5.8S rRNA implies an additional cleavage (4a) within ITS2. An additional cleavage site corresponding to the yeast A3 site (2b) within ITS1 is also proposed. The positions of oligonucleotide probes used for Northern analysis are shown in gray below the primary transcript. (B,C) Northern analysis of rRNA from RPL35A knock-down in UT7-Epo demonstrates steady-state increases in 45S:32S and 32S:12S ratios, indicating a disruption of 32S pre-rRNA maturation with resultant decreases in mature 28S (B) and 5.8S (C) rRNA. The rRNA species are indicated to the right of each panel; the probe used is indicated in gray to the left of each panel. Eth indicates ethidium bromide; Sh-Luc, Luciferase-control transduced cells; sh-1, -2, -3, or -4, respective RPL35A shRNAs.

Figure 7

DBA EBV cell lines display altered large subunit pre-rRNA processing. RNA isolated from DBA EBV cell lines was probed with a 5.8S rRNA probe: lanes 1 to 4, healthy subject LCL; lane 5, deletion 1; lanes 6 to 8, DBA LCL with S19 mutations; lanes 9 to 12, non-S19-mutated DBA LCL. An altered 32S:12S ratio was observed in deletion 1 (*lane 5) and DBA019 EBV (+lane 11) cell lines. The sample in lane 12 also has increased 32S:12S with normal RPL35A transcript levels by qPCR and normal sequence for RPL35A and RPS19, suggesting additional gene defects may impair this pre-rRNA processing pathway in DBA.