Regulatory Roles of Rpl22 in Hematopoiesis: An Old Dog With New Tricks

Abstract

Ribosomal proteins have long been known to serve critical roles in facilitating the biogenesis of the ribosome and its ability to synthesize proteins. However, evidence is emerging that suggests ribosomal proteins are also capable of performing tissue-restricted, regulatory functions that impact normal development and pathological conditions, including cancer. The challenge in studying such regulatory functions is that elimination of many ribosomal proteins also disrupts ribosome biogenesis and/or function, preventing one from unambiguously determining whether developmental abnormalities resulting from ablation of a ribosomal protein result from loss of core ribosome functions or from loss of the regulatory function of the ribosomal protein. Rpl22, a ribosomal protein component of the large 60S subunit, provides insight into this conundrum, as Rpl22 is dispensable for both ribosome biogenesis and protein synthesis, yet its ablation causes tissue-restricted disruptions in development. Here we review evidence supporting the regulatory functions of Rpl22 and other ribosomal proteins.

INTRODUCTION

Mutations in ribosomal proteins (RP) have long been known to cause a collection of diseases termed ribosomopathies, which are characterized by disruptions in hematopoiesis, abnormalities in craniofacial and skeletal structure, and increased cancer risk. The prevailing view has been that these abnormalities result from general perturbations in the assembly or function of the ribosome; however, recent analysis of the pathophysiology of ribosomopathies, and the specific functions of particular RP, has challenged this convention and suggested that, in addition to being key components of the ribosome, RP are also capable of serving regulatory functions^1–6. Indeed, evidence is emerging that suggests that the regulatory functions of certain RP may contribute to the constellation of abnormalities observed in ribosomopathies, and may do so from within specialized ribosomes, or alternatively, in an extraribosomal capacity. By exploring these novel regulatory mechanisms and pathways, we can begin to address key questions regarding the tissue-specific defects linked to human diseases. Here we review the evidence that certain RP are capable of regulatory functions, using the RP, Rpl22, as an important example of those capabilities.

BODY

Ribosomes & Ribosomopathies

Ribosomes are intracellular machines that are responsible for protein synthesis and comprise large ribonuceloprotein (RNP) particles containing numerous RNA and protein species. Eukaryotic ribosomes contain 4 ribosomal RNAs (rRNA) and ~78 RP⁷ and their assembly is facilitated by a large array of additional RNA and protein factors⁸. Despite the vital role of ribosomes in translation in all living cells, defects in ribosome synthesis are not always lethal. Ribosomopathies are a collection of human genetic disorders resulting from mutations in RP or ribosome biogenesis cofactors, which are thought to cause disruptions in development through inducing ribosome dysfunction^9,10. These diseases have recently attracted great interest, since elucidation of the mechanisms underlying this set of diseases could have broad implications for both basic biology and therapeutic intervention¹. Ribosomopathies include Diamond-Blackfan anemia (DBA), 5q-syndrome, Shwachman-Diamond syndrome (SDS), cartilage-hair hypoplasia (CHH), X-linked dyskeratosis congenita (DC), Treacher-Collins syndrome (TCS), North American Indian childhood cirrhosis (NAIC), Bowen-Conradi syndrome, and isolated congenital asplenia (ICA)^2,11,12.

As a founding member of the collection of diseases termed ribosomopathies, DBA was initially linked to genetic mutation of the RPS19 gene, which encodes a ribosomal protein in the small subunit of ribosome¹³. Since then, 13 additional RP genes (RPS7, RPS10, RPS17, RPS24, RPS26, RPS28, RPS29, RPL5, RPL11, RPL15, RPL26, RPL31 and RPL35A) have been implicated in DBA^14–18. Indeed, more than 50% of patients who are diagnosed with DBA harbor heterozyous mutations in one of the identified RP genes^1,3. While it has never been described in DBA, monoallelic loss of RPS14 has been shown to play a critical role in the development of 5q-syndrome^19,20, a condition of altered hematopoiesis with increased cancer risk that is an independent subtype of myelodysplastic syndromes (MDS). Remarkably, in contrast to the common anemic manifestations present in DBA and 5q-syndrome, halpoinsufficiency of RPSA (a small subunit RP) was recently linked to isolated congenital asplenia by unbiased exome analysis²¹. The absence of hematopoietic defects in patients with RPSA haploinsufficiency illustrates how mutations in particular RP can have distinct and tissue-restricted effects.

Despite the diverse clinical manifestations associated with ribosomopathies, the prevailing view is that these diseases result from activation of p53 via cellular stress induced upon ribosome dysfunction^3,22,23. Numerous reports have demonstrated that disruption in ribosome biosynthesis, now referred to as “ribosomal/nucleolar stress”, causes cell cycle arrest and apoptosis through p53 activation^2,24–27. Indeed, CD34+ bone marrow cells isolated from RPS19-mutant DBA patients generate a higher number of apoptotic cells during in vitro cell culture²⁸. Moreover, shRNA-mediated knockdown of Rps14 or Rps19 in healthy primary human CD34+ cells causes erythroid lineage-restricted induction of p53²⁹. Finally, inactivation of p53 rescues the phenotypes observed in several animal models of ribosomopathies, suggesting that aberrant p53 upregulation may be responsible for their pathogenesis^1,22,30–37. These studies raise a key question as to how disrupted ribosome biosynthesis activates p53. One mechanism involves blocking p53 ubiquitination and degradation. Numerous ribosomal proteins, including Rpl5, Rpl11, Rpl23, Rpl26, Rpl37, Rps3, Rps7, Rps14, Rps15, Rps20, Rps24 and Rps25 have been reported to physically interact with HDM2 (or Mdm2 in mouse) and suppress its E3 ubiquitin ligase activity^38–53. HDM2 is a pivotal negative regulator of p53 that polyubiquitinates p53 and causes its proteasomal degradation. Hence, the loss of HDM2 E3 ligase activity upon binding of RP results in elevated levels of p53. While numerous RP can interact with HDM2, several studies suggest that only Rpl5 and Rpl11, potentially together with 5S rRNA, are essential for the induction of p53^27,54–56. Indeed, only Rpl5 and Rpl11 accumulate in the ribosome-free fraction upon induction of nucleolar stress, while other RP are rapidly degraded by the proteasome⁵⁷. Regardless, the prevailing view is that defective ribosome biogenesis activates p53 by increasing the pool of free RP, which impair HDM2 function and increase p53 stability^1,58. p53 activation, in turn, is postulated to induce anemia because the rapid proliferation of erythroid progenitors renders them hypersensitive to p53 induction²³.

While this model may explain the anemia and bone marrow failure seen in may ribosomopathies, it fails to explain the distinct clinical manifestations observed in particular ribosomopathies. Presently, the pathophysiology of ribosomopathies remains largely unknown and is an area of active research^9,10. Among the outstanding questions is how mutations in ubiquitously expressed RP are able to cause tissue-restricted defects². Emerging evidence suggests that RP are not just static passengers on ribosomes that contribute to its structural integrity, thereby generally facilitating translation, but may also actively participate in regulating many biological processes, from within and even separate from the ribosome^2,4.

A New Paradigm: Unique Functions of Ribosomal Proteins

In principle, RP are capable of performing regulatory functions either from within specialized ribosomes of differing content or separated from the ribosome in an “extraribosomal” capacity. The “Ribosome Filter Hypothesis” was proposed in 2002 by Mauro and Edeleman and suggested that ribosomes of differing composition exert regulatory influence on the cellular proteome by controlling the extent to which particular mRNA species are translated into protein⁵⁹. While this is an attractive hypothesis that underpins more recent proposals of specialized ribosomes, it has been difficult to conclusively test; however, recent experimentation, including efforts evaluating the role of Rpl38 in controlling the translation of Hox mRNAs, is beginning to provide evidence in favor of some RP performing particular functions from within the ribosome^60,61. The alternative possibility, for which there is already substantial support, is that RP can be separate from the ribosome and regulate cellular processes in an “extraribosomal” capacity. There are three biological contexts where RP would be separated from the ribosome: 1) under conditions of disrupted ribosome biogenesis, as described above; 2) when super-stoichiometric increases in expression of one RP exceed the levels of other RP; and 3) stimulus-induced release from the ribosome. All of these circumstances enable RP, many of which are RNA-binding proteins, to engage cellular RNAs and proteins and regulate their activity.

Interactions with Proteins

Along with interactions between RP and HDM2 (discussed above), RP also interact with other members of this family of p53 regulators. Several RP, including Rps15, Rps20 and Rpl37, can bind to and negatively regulate MDMX, a homologue and partner of MDM2^4,40. Although there is no evidence of direct binding with MDMX, Rpl11 promotes MDMX degradation by binding MDM2⁶². Interestingly, Rpl11 also binds PICT1, which sequesters Rpl11 in the nucleolus and prevents its association with HDM2⁶³.

Rps3 has been shown to interact with E2F1 to induce apoptosis in neuronal cells, and this interaction is inhibited via AKT-binding and phosphorylation of Rps3⁶⁴. Rps3 also interacts with TRADD and is recruited to the death-inducing signaling complex (DISC) to induce apoptosis⁶⁵. In addition to regulating apoptosis, Rps3 has also been reported to influence the transcriptional activity of NF-κB. Rps3 associates with p65 (RelA) of the NF-κB complex and is required for enhanced DNA binding and regulation of the expression of a subset of NF-κB target genes⁶⁶. Further studies demonstrated that Rps3 translocation to the nucleus is dependent on IκB kinase β phosphorylation at a site distinct from the site phosphorylated by AKT⁶⁷. Phosphorylation of Rps3 also mediates activation of an NF-κB-mediated pro-survival pathway in non-small cell lung cancer (NSCLC) cells through disruption of the Rps3-TRAF2 complex, thereby rendering those cells radioresistant⁶⁸. In addition to phosphorylation, other post-translational modifications, such as methylation and sumoylation, also regulate the extraribosomal functions of Rps3^65,69,70.

Rps19, the first ribosomal gene linked to DBA¹³, has been shown to dimerize with the receptor for complement component C5a and induce a respiratory burst reaction in monocytes. Conversely, this same interaction inhibits the C5a-induced respiratory burst in neutrophils⁷¹. How this function of Rps19 influences the pathophysiology of DBA remains unknown.

A number of RP have also been implicated in regulating transformation through relatively poorly understood mechanisms. Rpl11 and Rps14 are both able to bind the c-Myc oncogene and inhibit its transcriptional activity by preventing the recruitment of its cofactor TRRAP^72,73. Rps7 suppresses ovarian tumorigenesis though regulation of the PI3K/AKT and MAPK pathways⁷⁴, although the mechanism of how Rps7 regulates those signaling networks has not been elucidated. Finally, Rpl41 regulates the phosphorylation and degradation of ATF4, which is critical for tumor cell survival in response to stress⁷⁵.

Interactions with RNA

One of the best-described extraribosomal functions of RP is that of Rpl13a, which is released from the ribosome by γ-interferon stimulation and regulates the translation of particular mRNA targets⁷⁶. In response to γ-interferon, Rpl13a is phosphorylated at Ser77, which releases it from the ribosome. Upon release, Rpl13a is able to bind to mRNAs containing “GAIT” elements (γ-interferon activated inhibitor of translation) in their 3′UTR and specifically repress their translation^76–79. Rpl13a represses translation as part of the heterotetrameric GAIT complex comprising Rpl13a, GAPDH, EPRS and NSAP1. GAIT-containing targets are highly enriched for mRNAs that encode for pro-inflammatory proteins, adding an additional layer of regulation and contributing to the resolution of inflammation^78,80,81. Recently, a virus-activated translation repression complex containing Rpl13a has also been reported⁸².

In addition to Rpl13 and the GAIT complex, several other RP have been reported to mediate transcript-specific translational control. In particular, emerging evidence suggests that RP may recognize specific IRES (internal ribosome entry site) elements⁵. For example, Rps6 and Rps25 are critical for IRES-mediated translation of viral RNAs^83–86. Rps19 and Rpl11, two DBA-associated RP, selectively regulate IRES-dependent translation of BAG1 and CSDE1, both of which are crucial for erythroid differentiation^4,5,87,88. A recent report demonstrated that Rpl38 is required for the IRES-dependent translation of HoxA during skeleton development, possibly explaining the tissue-specific developmental defects caused by Rpl38-haploinsufficiency^60,89. While the control of Hox transcript translation has been proposed to result from the function of Rpl38 from within specialized ribosomes, it is remains unclear how this would occur given that Rpl11 and Rpl38 are components of the large 60S subunit and only the small 40S subunit is associated with the mRNA during scanning for start codons.

While impaired ribosome biogenesis has been shown to induce p53 by altering p53 stability, RP have also been implicated in regulating the translation of p53 mRNA. Indeed, Rpl26 enhances p53 translation following DNA damage through interactions with the 5′ and 3′ UTR of p53 mRNA, and this is antagonized by nucleolin^90–92. Rpl22 has also been suggested to bind to p53 and repress its translation⁹³. In addition to the regulation of tumor suppressors such as p53, RP also regulate the expression of oncogenes. Indeed, Rpl11 and Rpl5 bind to the 3′-UTR of c-Myc mRNA and promotes its degradation^94,95, while Rps14 promotes c-Myc mRNA turnover through miRNA-mediated degradation⁷³. Moreover, Rpl11 has been shown to upregulate the translation of mRNAs carrying a 5′ oligopyrimidine tract (5′-TOP mRNAs), which have been implicated in growth control^27,96.

Tissue-specific and divergent functions of RP paralogs

Unlike yeast, where 75% of RP have highly-homologous functional paralogs, most RP in plants, flies, and vertebrates are encoded by a single functional gene⁵. Nevertheless, a handful of RP paralogs have been preserved in mammals, and such paralogs exhibit tissue specificity in their expression. The paralog of Rpl3, Rpl3l, is selectively expressed in muscle tissue, while that of Rpl10, Rpl10l, is restricted to the testis^12,97. The physiologic relevance of the tissue-restricted expression of RP paralogs remains largely unknown, but has been suggested to enable greater control of gene expression in higher eukaryotes⁹⁷.

Rpl22: not a typical ribosomal protein

Recent reports have demonstrated highly specialized roles for Rpl22 and its paralog Rpl22-like1 (Rpl22l1) in normal development and tumorigenesis^98–101. We will now discuss findings on how these two proteins regulate hematopoiesis and transformation. In particular, we will focus on why Rpl22 appears to be critical for normal development and function of only certain tissues, despite its ubiquitous expression. We will also address the functional relationship between Rpl22 and Rpl22l1 and the evidence that they are functioning through extraribosomal interactions with cellular target RNAs and proteins.

Historical perspective: from housekeeping to RNA-binding

Rpl22 was originally identified as a ubiquitously-expressed protein that was associated with the ribosome, which led to its use as a loading control for normalizing gene expression^102,103. Subsequent studies, however, demonstrated that Rpl22 is incorporated into the 60S large subunit of the ribosome at a late stage of ribosome maturation¹⁰⁴. Rpl22 is an external, salt-extractable component on the surface of 60S, positioned away from the subunit interface (Figure 1A, blue area). Although it co-localizes with rRNA in the nucleolus and the cytoplasm^105–107, Rpl22 is not required for general ribosome assembly or global, cap-dependent translation¹⁰⁸. This is consistent with its positioning in the crystal structure of the 80S ribosome, which reveals that Rpl22 is located distal to both the mRNA entry tunnel and the nasecent protein exit tunnel of the 60S subunit (Figure 1A, arrows).

Figure 1. Rpl22 and its location in the ribosome.

A) The position of Rpl22 in the published 80S ribosome model. Red: small 40S subunit, Green: large 60S subunit, Blue: Rpl22. B) Amino acid sequence of murine Rpl22. The RNA-binding helices (Red) and nuclear localization signal (NLS; blue) are indicated. C) Cryo-EM depiction of Rpl22 interacting with 28S rRNA. D) Ribbon diagram of Rpl22 secondary structure as it interacts with 28S rRNA. The red arrow indicates the RNA-binding helices.

Rpl22 is an RNA-binding protein that not only associates with the 28S ribosomal RNA, but also with the Epstein-Barr Virus EBER-1 RNA in human B cells¹⁰⁹ and the 3′ UTR of hepatitis C virus (HCV) RNA^110,111. Rpl22 also interacts with human telomerase RNA hTR within the nucleus^112,113 and modulates the translation of its paralog, Rpl22l1, by binding to an internal hairpin structure in the Rpl22l1 mRNA¹¹⁴. Rpl22 contains classical RNA-binding helices (Figure 1B–D, red arrow), which are required for RNA-binding, as their mutation abolishes interactions with RNA¹¹⁵. SELEX experiments revealed that Rpl22 binds a consensus RNA secondary structure with three conserved nucleotides (G-CU) in the neck of a stem-loop¹¹⁰ (Figure 2A). This consensus binding motif is present in all Rpl22 RNA targets identified to date, including 28S rRNA, hTR, and the Epstein Barr Virus EBER-1 latency RNA stem-loops III and IV, which direct translocation of Rpl22 from the cytoplasm to the nucleus in EBV infected host cells^{109, 112, 113}.

Figure 2. Functional interplay between Rpl22 and its highly homologous paralog, Rpl22l1.

A) RNA consensus hairpin/motif bound by Rpl22 (adapted from Dobbelstein and Shenk, J Virology, 1995). B) Evolutionary conservation of the RNA-binding domains of Rpl22 and Rpl22l1. C) Rpl22 and Rpl22l1 play distinct, and opposing roles in HSC emergence. Rpl22 and Rpl22l1 both bind to Smad1 mRNA and control its translation, with Rpl22 playing a repressive role and Rpl22l1 acting to oppose that repression. D) HSC emergence is controlled by the opposing actions of Rpl22 and Rpl22l1 on Smad1 expression. When Rpl22 and Rpl22l1 are in balance, HSC emergence occurs normally; however, if Rpl22 dominates, it represses Smad1, blocks HSC emergence and causes anemia, whereas if Rpl22l1 dominates, HSC emergence is uncontrolled, leading to an increased predisposition to transformation.

In addition to its RNA binding ability, emerging evidence suggests that Rpl22 physically interacts with various protein partners. For example, Drosophila Rpl22 specifically interacts with and co-localizes with histone H1, and this complex is tightly associated with condensed chromatin, regulates histone modifications, and epigenetically modulates gene transcription¹⁰⁶. In addition, Drosophila Rpl22 can interact with Poly(ADP-ribose) polymerase (PARP), a nuclear enzyme involved in DNA repair and programmed cell death¹¹⁶. Casein kinase 2α was also reported to co-purify with Rpl22 in human lung cancer cells and in Drosophila^117,118. Interestingly, Rpl22 was reported to associate with Ago2, a core protein component for the miRNA-mediated translational repression machinery¹¹⁹. Together, these studies highlight the variety of proteins and RNA targets with which Rpl22 is already known to interact. While these interactions can easily be envisioned to contribute to the regulatory functions of Rpl22, their relevance to Rpl22’s role in normal development or in pathological contexts, such as in development, anti-viral immune responses or transformation, remain unknown.

Loss of Rpl22 in model organisms

In humans, haploinsufficiency of RP is the pathophysiological basis for DBA, leading to clinical phenotypes such as organ malformations, defective hematopoiesis and increased cancer risk¹¹. To gain insight into the normal physiologic role of Rpl22 in development, Rpl22 has been ablated in a number of model systems. Due to the ubiquitous expression of Rpl22 in all organisms, it would be predicted that the loss of Rpl22 would be the lethal. Indeed, deletion of Rpl22 in Drosophila and knockdown of Rpl22 in C. elegans are both embryonic lethal^11,120. However, ablation of Rpl22a or Rpl22b in yeast does not affect their viability, and such yeast exhibit only mild defects in bud site selection and slowed growth^121,122. The discrepant results in these model systems prompted an assessment in mammals. The Rpl22 knockout mouse, which was generated by gene-trap mutagenesis, was viable and fertile with no obvious morphologic defects⁹⁸, which is quite unusual as RP-deficiencies are generally embryonic lethal or at least display gross morphologic abnormalities^123,124. Interestingly, the most obvious initial phenotype in Rpl22-deficient mice was a highly selective arrest in development of a subset of T lymphocytes^98,125. This raises two critical questions: 1) Why is Rpl22-deficiency not lethal; and 2) Why do only selected tissues exhibit Rpl22-dependency, when Rpl22 is ubiquitiously expressed?

The most likely explanations are either that: 1) Rpl22 is dispensable for ribosome biogenesis and function and the tissue-restricted developmental abnormalities result from the loss of its tissue-restricted regulatory functions; or 2) Rpl22l1, the highly homologous paralog of Rpl22, might compensate for Rpl22 loss in some tissues. In Drosophila, Rpl22l1 is specifically enriched in testes¹²⁶ but not in the ovary or other tissues¹²⁷. Rpl22l1 can incorporate into the ribosome and associate with active translational machinery¹²⁷. The RNA-binding helices in Rpl22l1 are highly conserved and identical to those in Rpl22, suggesting the potential for Rpl2 and Rpl22l1 to bind a similar set of RNA targets (Figure 2B). Moreover, in Rpl22-deficient mice, Rpl22l1 expression is increased and it is more extensively incorporated into the ribosome, suggesting that the potential for compensation exists¹¹⁴. Conversely, while knockdown of Rpl22l1 in MEF impairs growth, the knockdown of Rpl22 promotes MEF growth and proliferation, raising the possibility that these paralogs perform distinct functions¹¹⁴. Accordingly, the potential for compensation for Rpl22 loss by Rpl22l1 exists, but at least in some tissues, the functions of these paralogs may be distinct. The surprising possibility that Rpl22 and Rpl22l1 may have distinct and even antagonistic functions is addressed below.

Antagonistic Regulatory Functions of Rpl22 and Rpl22l1 in Hematopoiesis

The question of whether Rpl22 and Rpl22l1 are functionally redundant was evaluated using the zebrafish model, which has a number of attributes that make it particularly useful for such comparisons⁹⁹. Zebrafish are extremely amenable to assessing the regulatory roles of RP in development because: 1) blood cell development in zebrafish is extremely well conserved with that in mammals; 2) development of all blood cell lineages can be easily monitored in the transparent embyros using transgenic fluorescent reporters that mark each blood cell lineage^128–130; and 3) gene expression can be easily manipulated by knocking genes down using morpholinos or by overexpressing them through injection of mRNA or heat inducible expression vectors.

Both Rpl22 and Rpl22l1 are ubiquitously expressed in all tissues in zebrafish. Their functional relationships were assessed by knockdown using morpholino oligonucleotides, which revealed that both are essential for development of T cell progenitors in the thymus⁹⁹; however, Rpl22 knockdown arrested T cell development in a p53-dependent manner, whereas the arrest of T cell development caused by Rpl22l1 knockdown was p53-independent. Rpl22 knockdown arrested the development of T cell progenitors after their arrival in the thymus, but knockdown of Rpl22l1 arrested T cell development indirectly, by arresting the emergence of hematopoietic stem cells (HSC) in the aorta-gonad-mesonephros (AGM). Because of the arrest of HSC emergence caused by Rpl22l1 knockdown, it was not possible to determine in zebrafish if Rpl22l1 function was also required in downstream progeny, including T cell progenitors in the thymus. Nevertheless, it was definitively demonstrated that Rpl22 and Rpl22l1 are not functionally interchangeable in the hematopoiesis, since enforced expression of Rpl2211 was unable to alleviate the arrest of T cell progenitors caused by knockdown or Rpl22, nor was enforced expression of Rpl22 able to restore HSC emergence in zebrafish embryos in which Rpl22l1 had been knocked down. Together, these data make clear that Rpl22 and Rpl22l1 are performing distinct functions in hematopoiesis.

The arrest of HSC emergence caused by Rpl22l1 knockdown is independent of effects on p53 and instead results from impaired BMP4/Smad1 signaling, which is required for induction of the essential transcriptional regulator for definitive hematopoiesis, Runx1⁹⁹. Rpl22l1 knockdown interferes with BMP4/Smad1 signaling by post-transcriptionally reducing the expression of Smad1. This reduction in Smad1 expression is associated with a shift of smad1 mRNA out of the polysome fraction⁹⁹. Moreover, smad1 mRNA contains consensus Rpl22/Rpl22l1 binding sites (Figure 2A) and can be bound by Rpl22l1, as well as Rpl22. These data suggest that Rpl22l1 controls the translation of smad1 mRNA by direct binding. Moreover, the repression of Smad1 protein expression caused by knockdown of Rpl22l1 is reversed by simultaneous knockdown of Rpl22, indicating that Smad1 expression is reduced upon Rpl22l1 knockdown because the Rpl22 that remains represses the translation of smad1 mRNA. Thus, Smad1 expression is antagonistically controlled by Rpl22 and Rpl22l1, with Rpl22l1 acting to facilitate Smad1 expression and Rpl22 acting to interfere with Rpl22l1 and block Smad1 expression (Figure 2C). By extension, HSC emergence is controlled by the antagonistic balance of Rpl22 and Rpl22l1. When Rpl22 and Rpl22l1 expression is balanced, HSC emergence occurs normally in a tightly regulated manner (Figure 2D); however, when Rpl22 and Rpl22l1 are not balanced, pathology results, with excess Rpl22 blocking HSC emergence and causing anemia, while Rpl22l1 dominance leads to excessive generation of HSC that are predisposed to transformation (Harris et al., In preparation). More generally, these findings suggest that Rpl22 and Rpl22l1 are likely to bind an extensively overlapping set of RNA targets, but they have distinct and in many instances antagonistic effects on those targets (Figure 2C).

Rpl22 in Lymphocyte Development and Transformation

While the networks of cell surface receptors and transcription factors that regulate the development and function of B and T cells have been extensively studied^131–133, relatively little is known regarding the influence of post-transcriptional regulation on lymphocyte development and function by RNA-binding proteins such as Rpl22. Despite the ubiquitous expression of Rpl22, its germline ablation does not result in embryonic lethality or gross abnormalities in mice, but does result in profound reductions in selected peripheral lymphocyte populations⁹⁸. Indeed, Rpl22-deficient mice have striking decreases in splenic CD4 and CD8 T cells, and somewhat more modest reductions in splenic B2 cells and B1 B cells present in the peritoneal cavity^98,134. Surprisingly, Rpl22-deficient peripheral B cells appear fully functional, as they are able to proliferate in response to mitogenic stimulation, upregulate activation markers, and undergo class switch recombination. Initial studies reported that Rpl22-deficient T cells were unable to proliferate in response to TCR stimulation⁹⁸; however, this is likely to be due to the extensive homeostatic proliferation that occurs in Rpl22-deficient mice, which are lymphopenic because of limited thymic output¹³⁴. While these studies do not rule out a role for Rpl22 in peripheral immune responses, they suggest that the predominant role of Rpl22 is likely to be in supporting normal lymphoid development.

B cell development

The development of B cells occurs in the bone marrow of adult mammals¹³². The first committed B cell progenitor in the bone marrow is the pro-B cell, in which rearrangements at the immunoglobulin heavy chain (IGH) locus occur. Successful rearrangement of the IGH locus results in production of μ heavy chain protein, which pairs with the surrogate light chains, λ5 and VpreB, to form the pre-BCR. Expression of the pre-BCR initiates the signals required for traversal of the pre-BCR checkpoint, which ensures that only progenitors with productive IGH rearrangements are selected to differentiate to the large pre-B cell stage. Following several rounds of cellular division to expand the pool of B cell progenitors with successful IGH rearrangements, the large pre-B cells differentiate to the small pre-B cell stage where immunoglobulin light chain (IGL) rearrangements occur, first at the κ locus and then at the λ locus. Successful rearrangement of the IGL locus results in expression of IGL chain protein, which pairs with μ IGH protein to form the BCR, which initiates differentiation to the immature B cell stage. Immature B cells are vetted through a tolerance checkpoint to ensure that only non-autoreactive B cells are selected into the peripheral B cell pool.

While initial studies reported potential defects in B cell development in Rpl22-deficient mice^98,100, it was only recently that the stages of B cell development impacted by the loss of Rpl22 were reported¹³⁴. Rpl22-deficiency causes a mild reduction in pro-B cells that becomes more pronounced at the pre-B and immature B cell stages, demonstrating that Rpl22 is required for early B cell development. Pro-B cells are highly dependent on the cytokine interleukin 7 (IL-7) for their survival and proliferation¹³⁵. Despite expressing normal cell surface levels of the IL-7 receptor α chain, Rpl22-deficient pro-B cells were unable to respond to IL-7, exhibiting both increased apoptosis and reduced proliferation. Responsiveness of Rpl22-deficient pro-B cells to IL-7 signaling in vitro is restored by the knockdown of p53, and ablation of p53 completely restores the development of B cells in Rpl22-deficient mice. The mechanistic basis by which Rpl22 loss induces p53 in developing B cells remains unknown.

T cell development

T cells develop in the thymus through characteristic stages defined by the expression of cell surface molecules¹³³. The earliest T cell progenitors lack surface expression for the co-receptors CD4 and CD8 and are termed double negative (DN) progenitors. The DN compartment is further segregated by CD44 and CD25 expression into DN1 (CD44+ CD25−), DN2 (CD44+ CD25+), DN3 (CD44− CD25+) and DN4 (CD44− CD25−) stages. The early thymus-seeding progenitors are contained with the DN1 compartment. Upon entry into the thymus, interactions of these progenitors with Notch ligands initiate commitment to the T cell lineage. At this stage, rearrangement of the Tcrb, Tcrg, and Tcrd loci begins. Successful rearrangement of the Tcrg and Tcrd loci results in surface expression of the γδ T cell receptor (TCR) and commitment to the γδ T cell lineage. Successful rearrangement of the Tcrb locus results in production of TCRβ protein, which pairs with the pre-T cell receptor α (pTα) to form the pre-TCR. Expression of the pre-TCR initiates the signaling events required for traversal of the β-selection checkpoint at the DN3 stage, which ensures that only those progenitors with productive Tcrb rearrangements are selected to differentiate. β-selected T cell progenitors undergo robust proliferation and downregulate CD25 as they differentiate to the DN4 stage, following which they upregulate CD4 and CD8 as they differentiate to the CD4+ CD8+ double positive (DP) stage. In DP thymocytes, the Tcra locus is rearranged. Successful rearrangement of the Tcra locus initiates two selection checkpoints, positive and negative selection, which ensures generation of T cells that are not autoreactive, yet recognize peptides in the context of self MHC molecules.

The striking decreases in peripheral CD4 and CD8 αβ T cells, but not γδ T cells, in Rpl22-deficient mice suggested that Rpl22 might play a crucial role specifically during the development of αβ T cells⁹⁸. Consistent with this notion, γδ T cell development is minimally affected by Rpl22-deficiency; however, Rpl22-deficiency profoundly impairs the development of αβ lineage T cells. Indeed, αβ T cell development in Rpl22-deficient mice is arrested at the β-selection checkpoint, leading to severe reductions in DP, as well as CD4 and CD8 single positive, thymocytes. The arrest at the β-selection checkpoint in Rpl22-deficient mice does not result from impaired rearrangement of the Tcrb locus or pre-TCR signaling, but rather resulted from p53-induced apoptosis that began to manifest in DN3 thymocytes. Thus, Rpl22-deficiency impaired the development of both B and αβ T lymphocytes in a p53-dependent manner, as p53-deficiency completely reverses the developmental arrest of both cell lineages.

Enhanced p53 protein expression in the absence of Rpl22

The induction of p53 is frequently associated with mutations that inactivate RP^{53,98,134,136}; however, a major distinction is that the induction of p53 in response to loss of other RP is widespread, explaining the embryonic lethality of most RP mutations, whereas p53-induction in Rpl22-deficient mice is highly selective, even among closely related cell lineages. While p53 induction was observed in B and αβ T lineage progenitors, it was not induced in closely related γδ T progenitors. p53 causes the developmental arrest of B and αβ T lineage progenitors through the induction of proapoptotic targets, including Puma, Noxa, and Bax, as well as the cell cycle regulator p21^134,136. p53 also induces the target microRNA miR-34a, which is able to replicate the developmental arrest of p53 induction upon ectopic expression in Rpl22-expressing αβ T lineage progenitors. Of the potential downstream targets of p53, Puma was the only molecular effector whose elimination by itself was able to partially rescue the developmental defects observed in Rpl22-deficient mice¹³⁶. The combined loss of Puma and Bim (not a p53 target) resulted in a near complete restoration of T cell development in the absence of Rpl22, further demonstrating that p53 induction in Rpl22-deficient progenitors controls early T cell development primarily through effects on cell survival. Collectively, these studies demonstrated that Rpl22 regulates survival during early B and T cell development by repression of p53 (Figure 3).

Figure 3. Rpl22 and Rpl22l1 have multiple roles during hematopoiesis.

Rpl22 and Rpl22l1 have opposing effects on the translation of Smad1 mRNA, and the balance of their antagonistic activities control HSC emergence by controlling the expression of Smad1. Rpl22 also controls key developmental checkpoints during B and T cell development by controlling p53 expression in a lineage-restricted manner.

A critical remaining question is why the inactivation of other RP causes widespread induction of p53 and embryonic lethality, while Rpl22-deficiency selectively induces p53 expression in only selected lineages. One important factor is that Rpl22 is not essential for ribosome biogenesis or global protein synthesis, suggesting that the developmental defects caused by Rpl22-deficiency resulted from the loss of some lineage-restricted regulatory function of Rpl22^98,100. Rpl22 loss does not increase p53 expression by enhancing its stability, as occurs upon loss of other RP. Instead, Rpl22 regulates p53 expression by altering its translation⁹⁸. Indeed, a recent study reported that Rpl22 can directly bind to p53 mRNA and control its translation⁹³, suggesting direct repression. However, this finding is insufficient to explain why p53 translation is increased upon Rpl22 loss in only certain cell types, such as in αβ, but not γδ, T cell progenitors. p53 is activated in response to a variety of distinct stimuli, including DNA damage and the unfolded protein response^137,138. The DNA damage response is activated and leads to activation of p53 during early lymphocyte progenitors in response to V(D)J recombination of the antigen receptor loci¹³⁹. Moreover, DNA-damage or nucleotoxic stress unrelated to V(D)J recombination might also be expected to occur during the robust proliferation undergone during development of susceptible αβ and B cell lineages, but not by γδ T cell progenitors¹⁴⁰. The unfolded protein response is also activated during early B and T cell development, possibly as a consequence of the robust proliferation required following successful antigen receptor rearrangements¹⁴¹. Thus, the lineages that are susceptible to developmental arrest by Rpl22-deficiency may be those whose development entails abrupt and stress-inducing transitions, such as those that occur at the pre-BCR and β-selection checkpoints. The mechanistic basis for lineage restriction of p53 induction remains a critical, unanswered question.

Transformation

RP mutations are often associated with an increased risk for hematologic malignancies¹¹; however, the underlying mechanistic basis by which RP mutations facilitate transformation remain elusive. Interestingly, haploinsufficiency of Rpl24 and Rpl38 delay the development of B and T cell lymphomas in mouse models^142,143. In contrast, Rpl22 haploinsufficiency results in the rapid development of thymic lymphoma in a mouse model of T acute lymphoblastic leukemia (T-ALL) driven by enforced expression of oncogenic Akt2 in T cell progenitors¹⁰¹. Rpl22 appears to be functioning as a haploinsufficient tumor suppressor, as the remaining Rpl22 allele in Rpl22^+/− lymphomas is both intact and expressed. The link between Rpl22 inactivation and enhanced transformation and/or accelerated tumor progression is also observed in human cancer. Indeed, RPL22 is inactivated in approximately 10% of primary T-ALL and specific inactivating point mutations in RPL22 have been observed in relapsed T-ALL samples and are associated with reduced survival¹⁰¹. In addition, Rpl22 mRNA levels are reduced in a number of T cell malignancies, including adult T-cell lymphoma/leukemia¹⁴⁴. These studies suggest that, unlike Rpl24 and Rpl38, Rpl22 acts as a tumor suppressor.

The mechanism by which Rpl22 haploinsufficiency accelerates the development and progression of T cell malignancies is distinct from the mechanism by which it controls traversal of the β-selection checkpoint as it is unrelated to effects on p53. Indeed, p53 was not inactivated in Rpl22^+/− thymic lymphomas, suggesting that Rpl22 suppresses transformation by an alternative mechanism¹⁰¹. Instead, Rpl22 inactivation enhances transformation potential by inducing NF-κB, which then transactivates the stemness factor Lin28B¹⁴⁵. Lin28B, a negative regulator of processing of let-7 miRNA, is overexpressed in 15% of human primary tumors and is often associated with the advanced stages of disease¹⁴⁶. Importantly, among the downstream targets of let-7 are the critical oncogenes, Myc and Ras, which are upregulated upon Rpl22 inactivation. Consequently, Rpl22 haploinsufficiency appears to promote transformation by activating the NF-κB/Lin28B axis, which induces Myc and Ras in a let-7 dependent manner¹⁰¹. The mechanistic basis by which Rpl22 inactivation induces NF-κB activity to initiate this oncogenic cascade remains unclear.

CONCLUSIONS

The ability of Rpl22 and its paralog Rpl22l1 to play critical roles in regulating hematopoiesis, while not adversely affecting ribosome biogenesis or global, CAP-dependent translation, illustrates the regulatory, moonlighting functions that can be attributed to at least certain RP. These observations underscore an emerging paradigm shift in the way that RP are viewed. Indeed, because most RP are RNA-binding proteins, one can view the ribosome as a large collection of RNA-binding proteins with the potential to exist separate from the ribosome and in that state to bind cellular proteins and RNAs and regulate their activities, thereby critically influencing a variety of biological processes. Studying these regulatory functions remains challenging, as simple loss-of-function approaches for most RP will compromise both the core function of the RP in facilitating ribosome biogenesis and function, as well as its regulatory function, thereby preventing clear attribution of the cause of any resultant developmental abnormality. Fortunately, because Rpl22 and Rpl22l1 are dispensable for ribosome biogenesis and function, their analysis is not impeded by this complication, and the developmental defects caused by their loss can reasonably be attributed to their moonlighting functions. Nevertheless, a number of critical questions regarding their roles in development remain to be addressed. How are such highly homologous proteins as Rpl22 and Rpl22l1 (73% identical) able to perform distinct and even antagonistic functions? What are the cellular target RNAs through which they mediate their effects? How are these widely expressed proteins capable of exerting such lineage-restricted effects on p53 induction? How is Rpl22 and Rpl22l1 incorporation into the ribosome controlled and does Rpl22 and Rpl22l1 found outside the ribosome the result of stimulus-induced release or blocked incorporation? Addressing these questions represents a small part of an emerging effort to understand the ribosome and its protein components as a critical regulatory platform.