Molecular Approaches to Diagnose Diamond-Blackfan Anemia: the EuroDBA Experience

(Article begins on next page) Anyone can freely access the full text of works made available as "Open Access". Works made available under a Creative Commons license can be used according to the terms and conditions of said license. Use of all other works requires consent of the right holder (author or publisher) if not exempted from copyright protection by the applicable law. Availability: This is the author's manuscript


Introduction
Diamond Blackfan Anemia (DBA, OMIM #105650) is a rare congenital erythroblastopenia that is clinically and genetically very heterogeneous. 1 It represents part of a group of rare genetic disorders known as the inherited bone marrow failure syndromes (IBMFS), 2 and is characterized as a pure red cell aplasia that is also linked to physical malformations. 3 Because nearly all the genetic lesions driving DBA to date have been found in ribosomal protein (RP) genes, DBA is considered a "ribosomopathy". 4 This term is applied to disorders in which the pathogenic mutation results in defective ribosome biogenesis and/or the ability of ribosomes to properly translate mRNAs into protein.
The first description of DBA appears in a 1936 issue of Medicine in a chapter titled, "Anaemia of Infancy and Early Childhood" written by Hugh W. Joseph. 5 However, the recognition of DBA as a specific clinical entity is attributed to the American pediatricians Louis Diamond and Kenneth Blackfan, who published a paper describing it in 1938. 6 In its classical form, DBA affects approximately seven per one million live births and is characterized by a clinical presentation within the first year of life, macrocytic anemia with reticulocytopenia and a normocellular bone marrow with a paucity of erythroid precursors. 7 However, in recent years and by increasing the disease awareness more patients with atypical DBA manifesting later in life (or previously misdiagnosed) are referred to specialized DBA clinics. Physical malformations occur in roughly 50% of patients and include (among others) craniofacial and thumb deformities, short stature, cardiac and urogenital malformations. 8 Neurological or cognitive problems are very rare in DBA.
DBA patients generally exhibit increased levels of fetal hemoglobin and the activity of erythrocyte adenosine deaminase (eADA) is elevated in 80-85% of all patients. 9; 10 The risk of DBA patients developing cancer is higher than normal, although the risk does not appear to be as high as with other inherited bone marrow failures such as Fanconi anemia, Shwachman-Diamond syndrome, or dyskeratosis congenita. 11; 12 Genetics of DBA The DBA genotype, similar to the phenotype, is highly heterogeneous. The vast majority of allelic variations in DBA genes are mostly sporadic or de novo (55% of cases) and familial in the remaining 45%. In several instances of patients inheriting the mutation from a parent, the parent will not show any overt phenotype and are considered "silent carriers". Silent carriers may also exhibit only a macrocytosis without anemia and/or an elevated eADA. The first DBA-linked gene to be identified was RPS19 in 1999. 13 Subsequent to this initial finding, the identification of other mutations were revealed in RPS24, RPS17, RPL5, RPL11, RPS10, and RPS26. [14][15][16][17] Many other mutations in RP genes have been identified within the last ten years and today the list includes RPS7, RPS10, RPS15A, RPS17, RPS19, RPS24, RPS26, RPS27, RPS28, RPS29; RPL5, RPL9 (in review), RPL11, RPL15, RPL18, RPL26, RPL27, RPL31, RPL35, and RPL35A. [18][19][20][21][22][23][24][25] This list represents 20 of the 80 functional RP genes in humans.
Based on published observations (and unpublished observations of EuroDBA partners) it can be noted that the majority (>90%) of mutations fall in only 6 genes (RPS19, RPL5, RPS26, RPL11, RPL35A, and RPS24), while all other genes (such as RPS29, RPS17, RPS7, RPS10, RPL15, RPL9 and others) are mutated only in very few DBA patients worldwide and account for less than 10% of all mutated cases. There is little doubt that more RP or ribosome-associated genes will be identified in DBA patients in the near future.
All the RP gene mutations identified in DBA patients to date are heterozygous. Homozygosity is largely suspected to be lethal, a suspicion supported by the lethality of homozygous RP gene mutations in several animal models including zebrafish and mice 26; 27 . A wide range of mutation types is evident and at least in some cases appears to depend on the particular RP gene. Most of the missense mutations have been identified in the RPS19 gene while predominantly nonsense mutations, small deletions or insertions, and splice site mutations are found in RPL5 and RPL11. 16; 28 Partial-and whole-gene deletions have been detected (depending on the study cohort) in 10-20% of DBA patients using various copy-number methods (quantitative PCR, multiplex sequencing [MLPA], CGH and SNP arrays), mostly in RPS17, RPL35A, and RPS19 genes. [29][30][31] While DBA is considered almost exclusively linked to RP gene mutations, two non-RP genes have been reported in patients including GATA1 and TSR2. 32-34 35 The TSR2 gene is related to ribosome biogenesis since it is involved in pre-rRNA processing and binds to eS26 (RPS26) protein. GATA1 gene encodes for the major erythroid transcription factor GATA1 and is not reported to be involved in ribosome biogenesis.
In a substantial number of patients (approximately 30%) the underlying genetic defects remain unknown despite the routine screening of the known RP genes linked to DBA. However, with the increasing availability and diagnostic role of next generation sequencing methods, including multiplex gene sequencing and whole exome sequencing (WES), novel genetic defects are being slowly but steadily identified. 36

History of European DBA Registries
The rarity of diseases like DBA makes it difficult for one institute or clinician to become the centralized point of patient care. This difficulty exacerbates collecting the already sparse amount of clinical and biological data and using them to generate meaningful genotype:phenotype correlations. Thus the key to success when it comes to understanding and ultimately defeating DBA, or any other rare disease, is collaboration. Although national and international collaborations can be challenging, extraordinary progress has been made in developing, funding, and maintaining groups of clinical and biological researchers who share the same goal: To better understand and ultimately cure a specific rare disease such as DBA. Their goals were simple and straightforward: To share DBA clinical data and samples, to build registries, and to test new drugs. [37][38][39][40][41] By working together this group shared a major achievement in 1999 with the discovery RPS19 as the first known DBA-linked gene. 13 This gene, RPS19, still today remains the most commonly mutated gene found in ~25% of DBA patients and as such is routinely the first gene candidate sequenced when genotyping a patient.
The first observational DBA patient registries were initiated in the Czech Republic in 1988 and officially announced in 1992. 42 This was shortly followed by registries in the USA (DBAR, 1993), 42;43 Germany (1993), France (1995) and Italy (1995). The Italian registry is maintained as an online registry freely accessible to clinicians. 44  While the registries mentioned above contain the majority of European DBA patients due to the size of the host country's population, it is not necessary for a country to be highly populated in order to establish a meaningful registry. This is illustrated by the Israeli registry, which was founded in 2007. 46 Although Israel has a relatively low population, the respective registry contains virtually all known DBA patients in the country. This allows the registry data to be used for very precise statistical measurements of disease and phenotype frequencies that are far more difficult in larger countries. Another example is the incipient Dutch DBA registry, which was founded this year (in review). The fact that there are a limited number of clinics in close proximity that treat DBA patients in the Netherlands resulted in the establishment of a substantially sized registry in a very brief period of time.
In contrast, the initiation of patient registries in large or heavily populated countries can seem like a daunting task. This is especially true in countries that may not have access to or funding for state-of-the-art molecular diagnostics. The establishment of these patient registries represents a crucial step in creating a global DBA network. Beyond Europe, many other countries around the world have in recent years successfully established their own DBA patient registries ( Table 1). The populations of countries initiating these registries range from over a billion (China) to fewer than 3 million (Lithuania).
Thus the size and population density of any given country should not be considered a deterrent when deciding to establish a patient registry for a rare disease.

History of the European DBA Consortium
In The EuroDBA network over the next years expanded to include as associated partners other European countries that hosted DBA patient registries, such as Poland, Czech Republic, Italy, Spain, and Israel. In 2015 the funding for EuroDBA was renewed and the consortium was able to formally include many of the aforementioned countries. Moreover, the renewal allowed for the inclusion of the clinical groups in Poland, Turkey, as well as another group of biological researchers in France with expertise in pre-rRNA processing and how it is impaired by RP gene mutations.

Initial DBA Diagnostics
Patients typically present at the clinic with the basic hallmarks of anemia including pale pallor and failure to thrive. After collecting the familial history of the patient, the first test is typically a blood smear and blood cell count. DBA may be suspected if hemoglobin (Hb) is low, with absent or low reticulocyte numbers and often a macrocytosis (which is age-adjusted). Fetal Hb might also be increased, however this is an unspecific marker that is also elevated in other bone  Table 2.

Molecular DBA Diagnostics
Because the RPS19 gene is by far the most frequently mutated gene in DBA (25% of cases), most screening analysis begins with targeted Sanger sequencing of RPS19 (Table 2). This approach uses PCR amplification and sequencing of each RP gene exon and promoter region by specific forward and reverse primers in both directions. The subsequent genetic diagnostics does not fit a "one for all" approach to identify mutations, intra-exonic, full exon or whole gene deletions. Based on the availability of routine and sophisticated genomic methods, different approaches were developed in different countries ( Table 2). The first goal is to identify the most common genetic defects using routinely available methods such as Sanger sequencing or CGH array. Next generation sequencing (either targeted, or whole exome) might not yet be accessible to all laboratories, however recent developments in clinical diagnostics will likely lead to routine use of NGS instead of Sanger sequencing. Additional novel non-genetic techniques have been developed that reduce the time and cost of the molecular diagnosis of DBA.
One newly developed method takes advantage of the fact that rRNA in cells with small RP mutations typically reveals an increased 28S/18S ratio, while rRNA in cells with large RP mutation reveals a decreased 28S/18S ratio. 21  The defects in ribosome biogenesis by RP gene mutations have been proposed to activate the TP53 tumor suppressor pathway by inducing stress in the nucleolus, the cellular organelle where ribosome biogenesis originates. 55 However, one of the great puzzles of DBA is why, if RPs are expressed in every cell in the body, are erythrocytes so specifically affected when one copy of an RP gene is mutated? The specificity of the defects to erythroid cells has not been satisfactorily explained, although theories ranging from hypersensitivity of erythroblasts to elevated TP53 levels, a high protein demand in rapidly dividing erythroblasts, cell-specific translation and splicing defects and the induction of autophagy have also been proposed as mechanisms that result in the reduction of erythrocyte progenitor cells. 4; 55-61 The collaboration between the clinical and biological researchers has allowed for advancement in the pathophysiology studies of DBA that would be next to impossible for any one group to

Functional validation of DBA mutations
The functional validation of DBA-linked RP gene mutations may be achieved by analyzing the maturation of ribosomal RNA precursors by northern blot. 16; 17; 20; 21; 23; 29; 52; 62; 63 Mutations in DBA-linked RP genes invariably lead to haploinsufficiency of the corresponding protein. Since most ribosomal proteins are progressively incorporated into pre-ribosomal particles concomitantly to pre-ribosomal RNA maturation, lack of a given RP impairs processing of preribosomal precursors (pre-rRNAs) in a specific manner. 64 Modifications of the pre-rRNA pattern can thus be visualized by northern blot and used as a "molecular signature" for the defect of this RP. This characteristic pre-rRNA pattern can be determined by northern blot analysis of RNAs extracted from a patient's cells in order to validate the functional impact of a mutation. In case of a mutation in a new RP gene suspected to be pathogenic, the patient pre-rRNA profile is compared to that obtained after knocking down expression of the corresponding ribosomal protein with siRNAs in a cell line (see Figure 1A). Control samples from unrelated individuals, and/or unaffected parents or siblings are used for comparison. Because ribosome processing is affected ubiquitously in DBA patient cells, a variety of cell types can be used to prepare total RNAs including peripheral blood lymphocytes, LCLs, or fibroblasts. This technique is also useful to examine whether ribosome biogenesis is affected in patients for whom sequencing failed to reveal any mutation/deletion among the known DBA genes.
A complementary approach consists of analyzing ribosomes from cytoplasmic fractions on sucrose gradients. By providing the relative abundance of small and large ribosomal subunits and the distribution of polysomes (translating ribosomes), this technique reveals to which extent a RP defect not only impairs either pathway, but also impacts translation. Figure 1 provides an example of coupling these techniques for a DBA patient from the EuroDBA registry, for whom no RP gene defect was found by sequencing. Figure 1A reveals a clear ribosome biogenesis dysfunction in patient LCLs, with an accumulation of both 30S and 32S pre-rRNAs (precursors to rRNA constitutive of the small and the large ribosomal subunits, respectively). Quantifications of product to precursor ratios relative to the controls further ascertained these findings ( Figure   1A), which strongly support the diagnosis of DBA despite the lack of a candidate gene. Figure   1B illustrates how polysome profiling revealed a substantial loss of 60S subunits in the patient LCLs compared to the healthy control cells, suggesting a defective RP from the large ribosomal subunit. Figure 1C illustrates  54 This is best observed in RNAs extracted from peripheral blood lymphocytes (PBMCs) subjected to activation by phytohemagglutinin. Although this technique is not sensitive enough to see the vast majority of rRNA precursors, it allows detection of an increase of 32S pre-rRNAs when the large subunit production is impaired. Assessment of the 18S/18S ratio is now routinely used by the Italian members of the consortium prior to Sanger sequencing in order to determine which of the small or the large subunit pathway is affected, and to prioritize the RP genes to be sequenced. In future years, sensitive analytical approaches adapted to pre-ribosomal precursor analyses need to be adapted to clinical environments, in order to routinely validate ribosome biogenesis defects and help diagnosis of DBA.

Improving treatment and aiming for a cure
The registering of DBA patients, systematic genotyping, and the continued efforts in the laboratory have already been invaluable for establishing important genotype:phenotype relationships such as those discussed above. The molecular signatures of the different RP gene mutations are already beginning to be used to improve diagnostics. The continuing inclusion of more clinical and genetic data in patient registries means at this rate it won't be long before the results may be translated into meaningful patient management protocols. The hope is that in the future a patient's genetic information will be able to single-handedly predict, for example, a successful steroid treatment, the susceptibility to iron overload upon chronic blood transfusions, the likelihood of undergoing treatment independence one day, or the likelihood of developing cancer.
The other more obvious hope for the future of any rare disease is a cure.  [67][68][69] In the context of DBA, this technique has already been used to introduce and drive exogenous wild type RPS19 expression in inducible pluripotent stem cells (iPSCs) derived from a patient carrying a truncating mutation in one allele of RPS19. 70 This approach was able to successfully revert the ribosome biogenesis defects of the mutant cells. All this said, it should be kept in mind that CRSIPR/Cas9 technology is still very incipient. However when one considers its potential for curing monogenic inherited disorders in humans, there is no question that this technology will advance quickly in the near future.

Discussion
The advent of genome sequencing has resulted in significant advances in rare disease research this past decade. While researchers are now more likely than ever able to identify diseasecausing genes, the lack of understanding the pathphysiological mechanisms underlying these mutations remains a cumbersome bottleneck in terms of finding a therapeutic cure. In fact,     Table 2