Ribosomopathy-like properties of murine and human cancers

Ribosomopathies comprise a heterogeneous group of hematologic and developmental disorders, often characterized by bone marrow failure, skeletal and other developmental abnormalities and cancer predisposition. They are associated with mutations and/or haplo-insufficiencies of ribosomal proteins (RPs) and inefficient ribosomal RNA (rRNA) processing. The resulting ribosomal stress induces the canonical p19ARF/Mdm2/p53 tumor suppressor pathway leading to proliferative arrest and/or apoptosis. It has been proposed that this pathway is then inactivated during subsequent neoplastic evolution. We show here that two murine models of hepatoblastoma (HB) and hepatocellular carcinoma (HCC) unexpectedly possess features that mimic the ribosomopathies. These include loss of the normal stoichiometry of RP transcripts and proteins and the accumulation of unprocessed rRNA precursors. Silencing of p19ARF, cytoplasmic sequestration of p53, binding to and inactivation of Mdm2 by free RPs, and up-regulation of the pro-survival protein Bcl-2 may further cooperate to drive tumor growth and survival. Consistent with this notion, re-instatement of constitutive p19ARF expression in the HB model completely suppressed tumorigenesis. In >2000 cases of human HCC, colorectal, breast, and prostate cancer, RP transcript deregulation was a frequent finding. In HCC and breast cancer, the severity of this dysregulation was associated with inferior survival. In HCC, the presence of RP gene mutations, some of which were identical to those previously reported in ribosomopathies, were similarly negatively correlated with long-term survival. Taken together, our results indicate that many if not all cancers possess ribosomopathy-like features that may affect their biological behaviors.

one identified as "primary tumor" and one identified as "solid tissue normal" in the column "sample type", and 2) presence of RNA-seq data for both samples. A list of 80 human ribosomal proteins was assembled from the University of Miyazaki's Ribosomal Protein Gene Database (http://ribosome.med.miyazaki-u.ac.jp). When queried, three of these gene transcripts (RPL40, RPS30, and RPS4Y) were not found in the genomic data, so the final list used for the following analysis consisted of 77 RP genes. TCGA expression data, which is stored log-transformed, was base-two exponentiated for all samples.

TCGA: survival curves
To determine if RP transcript deregulation correlated with survival, this information was combined with clinical data from TCGA regarding days to death or last follow-up. Tumor samples meeting the following criteria were excluded from the survival analysis: 1) samples without corresponding clinical information, 2) samples with no recorded "days to death" or "days to last follow up", or 3) days to death or last follow-up less than or equal to zero. There were 357 tumor samples in the HCC cohort with corresponding clinical information, 25 of which were excluded from the survival analysis on these criteria. The CRC cohort contained 278 tumor samples with clinical information and 4 were excluded. The BC cohort contained 1082 tumors with clinical information, 18 of which were excluded. Survival analysis was not performed for the PC cohort, as the requisite clinical information was available for only 6 patients. Tumor samples were then sorted according to the severity of RP transcript deregulation and placed into the upper and lower quartiles. There were 83 tumors per quartile in the HCC cohort, 69 tumors per quartile in the CRC cohort, and 266 tumors per quartile in the BC cohort. Five-year survival curves were generated comparing the top and bottom quartiles in each cohort, with significance determined by a Log-rank test P-value < 0.05. Survival differences were significant in HCC (P = 0.0435) and BC (P = 0.0046).

TCGA: mutation analysis
TCGA mutation information was accessed using cBioPortal (http://www.cbioportal.org/), from the "TCGA, Provisional" data for each cancer type. Each data set was queried for RP coding mutations in any tumor sample. A literature search was performed in order to classify these observed mutations into three general categories: mutations in RPs previously implicated in a ribosomopathy, mutations identical to those previously identified in a ribosomopathy, and all other mutations. The literature search included the LOVD Diamond-Blackfan Anemia database (http://dbagenes.unito.it/home.php) as well as PubMed searches of each individual ribosomal protein gene identifier.

Quantification of rRNA processing
Total RNAs were purified using RNeasy columns (Qiagen, Inc. Valencia, CA) and then digested with TURBO-DNA free DNAse as recommended by the supplier (Thermo-Fisher, Pittsburgh, PA). RNA concentrations were determined with a Nanodrop ND-1000 instrument (NanoDrop Technologies Inc., Wilmington, DE, USA) and RNA integrity was evaluated with an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara,CA). RNA integrity number (RIN) values for all samples in the HCC mouse model and HB tumors >9 and those for HB control livers were >7.5.
To assess rRNA processing intermediates, we quantified 18S-ITS1, ITS1-5.8S, 5.8S-ITS2 and ITS2-18S junctions as depicted in Fig  Each reaction (20 µL in a 96-well fast plate) consisted of 10 ng of RNA template, 0.16 µL RT Enzyme Mix, 10 µL the above-described RT-PCR Mix, 3.84 µL nuclease-free water and 1 µL of primer solution containing both forward and reverse primers at a concentration of 100 ng/µL each. Run conditions for RT-PCR reactions consisted of a 48°C hold for 30 minutes to catalyze reverse transcription followed by a 10-minute hold at 95°C. PCR conditions comprised 40 cycles of a 1 min 95°C melting period followed by a 1-minute 65°C annealing and extension period. Each sample was assayed in triplicate with variances seldom exceeding 5%. P-values were determined using Welch's t-test.

Tissue fractionation immuno-precipitation and immuno-fluorescent staining
All buffers were supplemented with standard protease and phosphatase inhibitors. Tissue was first washed with ice-cold PBS, minced into small pieces and then homogenized in cytoplasmic extraction buffer (200-250 mg tissue/2 ml of buffer). Cell breakage was monitored under a microscope. Homogenates were centrifuged at 500x g for 5 min at 4 0 C and the resultant supernatants were used as the cytoplasmic fraction. Pellets were further subjected to membrane extraction buffer followed by centrifugation at 3000x g for 5 min at 4 0 C to pellet nuclei. Nuclear pellets were divided into two halves. One half was used to solubilize nuclei and release chromatin-bound proteins by extracting with nuclear extraction buffer containing 5 mM CaCl 2 .
The other half was processed to isolate nucleoli as described (http://www.lamondlab.com). For this step, nuclear pellets were re-suspended in a buffer containing 10 mM Hepes, pH 7.9; 0.35 M sucrose and 0.5 mM MgCl 2 and sonicated on ice for 6 x 10 sec bursts with 10 sec cooling intervals. Sonicated lysates were layered over a buffer containing 10 mM Hepes, pH 7.9; 0.88 M sucrose and 0.5 mM MgCl 2 and centrifuged at 3000x g for 10 min at 4 0 C. Nucleolar pellets thus obtained were solubilized in a buffer containing Tris HCl, pH 8.0; 20 mM NaCl; 1 mM EDTA; 0.5% NP40 and 25 mM NaF. Fractions were analyzed by immuno-blotting for GAPDH (cytoplasmic marker), histone H3 (nuclear marker), fibrillarin (nuclear and nucleolar marker).
Expression of p53, Mdm2, and p19 ARF was assessed by immuno-blotting across each of the fractions following SDS-PAGE.
For immuno-precipitations, freshly isolated cytoplasmic liver and tumor fractions were diluted to a final protein concentration of 3 mg/ml in "IP buffer" containing Tris HCl, pH 8.0; 20 mM NaCl; 1 mM EDTA; 0.5% NP40 and 25 mM NaF supplemented with protease and phosphatase inhibitors and subjected to two rounds of pre-clearing. The first round consisted of rocking 1 ml of lysate with 20 µl Protein G PLUS-agarose beads (Santa Cruz Biotechnology) for 1 h at 4 0 C followed by centrifugation at 4000 rpm, 4 0 C for 5 min, to remove agarose beads. The second round of clearing was performed following the addition of 20 µl of isotype specific IgG1agarose conjugate (LifeSpan BioSciences inc., Seattle, WA) to the pre-cleared lysate for 1 h at 4 0 C. Upon a brief centrifugation to again remove the agarose-conjugate, the fractions were equally divided. One portion was incubated with 20 µl mouse IgG1-agarose conjugated beads while the other part was incubated with Mdm2 antibody agarose-conjugated beads (Santa Cruz Biotechnology) overnight at 4 0 C with gentle shaking. Beads were washed four times for 1 h each time at 4 0 C with IP buffer to remove any unbound protein, followed by re-suspension in SDS sample buffer and denaturation at 95 0 C for 4 min. Immunoprecipitates were further analyzed by gel electrophoresis and silver staining.
Immunofluorescent staining (Fig. 5C) was performed on liver and tumor frozen sections.
Fresh tissues were first fixed in PBS-4% paraformaldehyde for 2-4 followed by an overnight incubation in PBS-40% sucrose at 4 0 C. The fixed tissues were then embedded in Tissue Plus O.C.T. Compound (SciGen Scientific, Gardenas, CA), frozen on dry ice and stored at -80 0 C.

Figure L. Mdm2-interacting RP peptides identified by MS in liver cytoplasmic lysates following anti-Mdm2 IP (~24-35 kDa range). An analysis identical to that described in S3
Table was performed the portion of the gel denoted by the red band depicted in Fig 5E. The coverage for each RP ranged from 7-39%. Each of the 7 RPs listed here is also listed in Table   B.  Table D. Each identified peptide is indicated by yellow highlighting and is mapped to its corresponding region in the sequence of the full-length RP. The coverage for each RP ranged from 13-59%.

Figure N. Mdm2-interacting RP peptides identified in liver cytoplasmic lysates following anti-Mdm2 IP (~14-24 kDa range).
Tryptic peptides corresponding to the indicated 11 RPs in the ~14-24 kDa range identified by mass spectrometry (Fig 4E, lane 2, blue bracket) and listed in Table D are indicated by yellow highlighting and is mapped to its corresponding region in the sequence of the full-length RP. The coverage for each RP ranged from 12-38%.   Table E. Each identified peptide is indicated by yellow highlighting and is mapped to its corresponding region in the sequence of the full-length RP. The coverage for each RP ranged from 8-39%.      K75Rfs   unmatched samples) and average relative expression in normal tissue with a two-sided t-test (P <0.05). All P-values were then adjusted based on a false-discovery rate of 5%. Percent difference in relative expression for a given transcript was then calculated by dividing the difference in average relative expression between HCC and normal tissue by the average relative expression in normal tissue. Transcripts were defined as having shared directionality when differences in relative percent expression were either all increased or decreased relative to normal tissues.

Tables
Transcripts with significantly different relative percent expression in 2 or more cancers but without shared directionality were excluded.    Fig 5E). Note that 10 of 11 proteins identified as Mdm2 binding partners in normal liver were also identified in IPs from HBs, which contained seven additional RPs. See Figure M and Figure N for the exact mapping of each identified peptide to its corresponding RP.    The listed genes are more frequently co-mutated in tumors possessing a ribosomal protein mutation than would be expected by chance alone. P-values were calculated using cumulative binomial distributions and are significant after correction for false discovery.