We sought to determine the impact of haploinsufficiency of ribosomal protein (RP) genes in C. elegans and focused on the single-copy losses of two large subunit ribosomal proteins, rpl-5 and rpl-33, along with two small subunit RPs, rps-10 and rps-23 (E. S. Cenik et al. 2019). We prioritized these four RP genes due to their involvement in human ribosomopathies, with rpl-5, rpl-33, and rps-10 relating to DBA (Farrar et al. 2008), and rps-23 relating to microcephaly and intellectual disability without the blood phenotypes (Paolini et al. 2017). Moreover, the protein products of these RP genes are incorporated into nascent ribosomes at different stages (nucleolar, nuclear and cytoplasmic) (de la Cruz, Karbstein, and Woolford 2015). We observed developmental delays across these RP haploinsufficient mutants compared to wild-type counterparts (Fig. 1A, 1B, and S1A). Protein levels were evaluated using semi-quantitative proteomics against stage-matched controls, revealing reductions of approximately 50% for RPL-33 and RPS-23, 25% for RPL-5, and 10% for RPS-10 (Fig. 1C, Table S1). The developmental delays observed in haploinsufficient strains, as compared to their time-matched controls, were found to generally correspond with the degree of protein reduction resulting from the loss of a single copy (Fig. 1C). These findings suggest a correlation between the extent of protein level reduction and the timing of developmental processes.

Similar to previous reports of reduced body size in RP mutants in other species (Marygold et al. 2007; Oliver et al. 2004), our examination revealed that, when given sufficient time to reach the same developmental stage, C. elegans RP haploinsufficient mutants were slightly smaller in body size than wild-type controls with one exception (Fig. 1D). Specifically, we observed that rps-10(0)/+ animals were slightly larger in body size compared to their stage-matched controls (Fig. 1D, p=0.02, independent Student’s t-test). The increased body size could be associated with increased cell volume, altered cytoskeletal dynamics or metabolism due to changes in signaling pathways such as TGF-β, MAPK, or cGMP (Hirose et al. 2003; Wang, Tokarz, and Savage-Dunn 2002; Watanabe et al. 2005). These results are also reminiscent of the larger wing sizes and wing discs observed in Drosophila following the single copy loss of RpL38 and RpL5 (Marygold, Coelho, and Leevers 2005). However, why single copy loss of certain RP genes leads to increased organ or body growth remains unclear.

We further conducted fecundity assays to assess impacts on reproduction. Except for rpl-5(0)/+ animals, which exhibited a significant reduction in progeny size (~25% reduction; p=0.03, independent Student’s t-test), the progeny sizes of all other RP haploinsufficient mutants were similar to those of the controls (Fig. 1E). The onset of peak fertility was delayed in all mutants except for rps-10(0)/+ animals (Fig. S1B). Additionally, RP mutants remained fertile for extended periods, thereby compensating for the overall progeny size, except for rpl-5(0)/+ mutants. Reduced fertility was observed in Drosophila minute mutants, characterized by the lack of a single copy of an RP gene (Marygold et al. 2007). The observation of similar brood sizes in the majority of RP strains in C. elegans suggests involvement of compensatory mechanisms within the germline.

Finally, we investigated lifespan in C. elegans RP mutants. Our lifespan analysis did not reveal significant differences for the majority of RP haploinsufficient mutants compared to wild-type controls, regardless of treatment with the egg-laying inhibitor Fluorodeoxyuridine (FuDR). The only exception was rps-23(0)/+ mutants that displayed a modest but significant extension of lifespan only in the absence of FuDR (Fig. 1F, S1C, p=0.007 for rps-23(0)/+ mutants, Log-rank test with Bonferroni correction). Reduced protein translation or the knockdown of RP genes are typically linked to increased lifespan (Syntichaki, Troulinaki, and Tavernarakis 2007; Steffen et al. 2012; Rogers et al. 2011; Chiocchetti et al. 2007; Curran and Ruvkun 2007; Tiku et al. 2017). However, our results suggest that haploinsufficiency for single RP genes negate the typical lifespan extension benefits associated with decreased protein synthesis due to imbalances in RP expression and the associated stress. To determine whether overall translation levels were reduced in response to RP haploinsufficiency, we quantified puromycin incorporation in stage matched animals over a limited time period. Our results suggest that haploinsufficiency for single RP genes does not result in a significant reduction in overall translation when compared to stage-matched respective controls (Fig. S1D, p-value>0.05, one-tailed Student’s t-test). Taken together, our phenotypic characterization of the haploinsufficiency of RP genes in C. elegans reveals a range of developmental and physiological consequences that broadly mimic those observed in other model organisms such as Drosophila and mice (Marygold et al. 2007; Oliver et al. 2004).

To understand the cellular mechanisms triggered by single copy losses of ribosomal protein genes in C. elegans, we performed RNA sequencing (RNA-seq) on stage-matched mutant and control animals at the L4 stage (Table S2). This analysis, which included stage-matched controls to mitigate any developmental delay effects, revealed a uniform gene expression response across all mutants. This response was characterized by overexpression of ribosomal machinery, glutathione transferase activity, and genes involved in innate immunity and stress responses, indicating a systemic adaptation to ribosomal protein loss (Fig. 2A, top, significantly enriched gene ontology [GO] categories provided in Table S3). Conversely, genes related to mitochondrial activity, fatty acid biosynthesis, cell polarity, and amino acid metabolism were significantly underexpressed (Fig. 2A, bottom, Table S3), suggesting a reprogramming of cellular metabolism in response to ribosomal protein haploinsufficiency.

Given the pronounced overexpression of glutathione transferase (gst) genes (~2.8 fold enrichment, p<0.001, GO enrichment), we assessed the expression patterns of gst genes across all RP mutants (Fig. 2B). The general overexpression signature aligns with previously established links between glutathione transferase activity and oxidative stress resistance (Ayyadevara et al. 2005; Burmeister et al. 2008), prompting us to assess the mutants’ resilience to oxidative stress.

In acute survival assays using high doses of paraquat, we observed that all RP haploinsufficient strains exhibited significantly enhanced resistance to oxidative stress compared to wild-type controls (Fig. 2C, first and second plot, p<0.01, Log-rank test with Bonferroni correction). This widespread increase in stress resistance suggests a robust, adaptive mechanism that compensates for elevated reactive oxygen species (ROS) levels.

Perturbations in ribosome biogenesis have been shown to elicit proteotoxicity (Tye et al. 2019). Moreover, ribosomal protein haploinsufficiency reduces ribosome levels (Khajuria et al. 2018), which could lead to a decrease in overall protein synthesis. To determine whether the observed elevated levels of oxidative stress in the mutants were due to proteotoxic stress or were related to a reduction in protein synthesis, we pre-treated wild-type animals with inhibitors targeting key pathways: the proteasome (bortezomib), ribosome biogenesis and the translation regulator TORC1 (rapamycin), and translation elongation (cycloheximide), before assessing survival under acute oxidative stress conditions. These treatments significantly enhanced the stress response of wild-type animals (p<0.05 for each drug, Log-rank test with Bonferroni correction), supporting the role of these pathways in mediating elevated oxidative stress resistance (Fig. S2A, left panel). Moreover, the combined use of the inhibitors did not further improve survival rates in wild-type animals (Fig. S2A, right panel). Finally, none of the treatments altered the survival outcomes of rpl-5(0)/+ mutants under oxidative stress (p ≥ 0.4, Log-rank test with Bonferroni correction), suggesting that rpl-5(0)/+ mutants inherently possess an elevated baseline oxidative stress response (Fig. S2B).

Unfolded protein and oxidative stress responses are mediated through SKN-1, which is orthologous to human NRF2 (Glover-Cutter, Lin, and Blackwell 2013; Oliveira et al. 2009; Xu et al. 2018; Hu et al. 2017; Inoue et al. 2005). SKN-1 further induces a transcriptional response that results in stress resistance when protein translation is inhibited (Wang et al. 2010). Moreover, TORC1 signaling pathway and rapamycin regulate both SKN-1 and DAF-16, orthologous to human FOXO3 (Robida-Stubbs et al. 2012). Additionally, DAF-16 is involved in repression of RP genes, serving for resistance to hypoxia resistance (Hemphill et al. 2022). Thus, we hypothesized that SKN-1 and DAF-16 might be regulators of the oxidative stress response observed in RP mutants. To dissect these regulatory pathways, we evaluated the oxidative stress survival of rpl-5(0)/+ mutants in combination with mutations in skn-1 and daf-16 genes. The skn-1(zj15) hypomorphic mutation (Tang, Dodd, and Choe 2016) in the rpl-5(0)/+ mutant background significantly diminished oxidative stress survival (Fig. 2C, third plot, p=0.02, Log-rank test), suggesting that SKN-1 is essential for the resistance of oxidative stress in rpl-5(0)/+ mutants. skn-1(zj15) mutants were not resistant to oxidative stress suggesting that SKN-1 is likely chronically activated in rpl-5(0)/+ mutants (Fig. S2C, p=0.52, Log-rank test with Bonferroni correction). We did not observe a significant survival change in rpl-5(0)/+ mutants when crossed with the daf-16(mu86) mutation (Lin et al. 1997), highlighting SKN-1’s unique contribution (Fig. 2C, fourth plot, p=0.5, Log-rank test). daf-16(mu86) mutants did not display differential resistance acute oxidative stress compared to wild-type controls (Fig. S2C, p=0.08, Log-rank test with Bonferroni correction). We also observed a mild overexpression of genes upregulated in response to skn-1 gain of function across the RNA-seq datasets for RP haploinsufficient mutants (Fig. 2D, (Nhan et al. 2019)). This finding suggests that SKN-1 activity may not be limited to rpl-5(0)/+ mutants but likely extends to other RP mutants as well.

In S. cerevisiae, aberrations in ribosome biogenesis lead to upregulation of targets of Hsf1, a key regulator of proteotoxic stress response (Li, Labbadia, and Morimoto 2017; Tye et al. 2019; Albert et al. 2019). To investigate whether a similar pattern occurred in our RP haploinsufficient mutants, we examined the expression levels of genes that are known targets of HSF-1 overexpression (Sural et al. 2019). Our analysis indicates a trend toward overexpression of HSF-1 regulated genes in two of the mutants, rpl-33(0)/+ and rps-23(0)/+ (Fig. S2D). However, the overexpression of HSF-1 targets did not translate into observable differences in acute heat resistance over a 2-hour period (Fig. S2E, p>0.05). Notably, in C. elegans, the responses to oxidative stress and heat stress act in opposition and attenuate each other’s effects (Crombie et al. 2016). Taken together, these observations suggest that while there may be alterations in the expression of HSF-1 targets under RP haploinsufficiency, this does not necessarily indicate activation of HSF-1 or enhanced stress tolerance. It may point to a complex interplay in the cellular stress response which requires further investigation.

To investigate the impact of single-copy ribosomal protein (RP) gene loss on translation, we used ribosome profiling (Ribo-seq) alongside RNA-seq to analyze stage-matched mutant and control C. elegans animals at the L4 developmental stage. This approach allowed us to identify translational efficiency (TE) alterations across four RP mutants (Table S4). Notably, genes such as ccdc-47, ddb-1, F32D1.5, pab-1, and rad-50 exhibited significantly decreased TE across all RP mutants (p_adj<0.05), with pab-1 and rad-50 also showing RNA overexpression, hinting at a potential compensatory mechanism in response to reduced TE (Fig. S3A).

Further analysis revealed distinct expression trends. Cell matrix adhesion and defense response genes were overexpressed at both RNA and TE levels (Fig. 3A, “RNA and TE over”, Table S4, significantly enriched GO category list provided in Table S5). In contrast, genes involved in histone H3K36 methylation and sister chromatid segregation were underexpressed at both levels, suggesting a systematic downregulation in these functional categories (Fig. 3A, “RNA and TE under”, Table S4, S5). Given the involvement of H3K36 methylation in processes such as RNA polymerase II-mediated elongation (Carrozza et al. 2005) and the regulation of alternative splicing (Luco et al. 2010), a reduction in the expression of components of this pathway could further contribute to alterations in transcription and the diversity of transcript isoforms being produced.

Interestingly, ribosomal component genes were overexpressed at the RNA level but had reduced TE, notable in both large and small subunit genes (Fig. 3A, “RNA under TE over” and “RNA over TE under”, respectively, Table S4, S5). This differential regulation highlights a complex response that balances transcription and translation in response to RP haploinsufficiency to maintain overall ribosome numbers.

In S. cerevisiae, mutations in small subunit genes do not impact the expression of large subunit genes, resulting in an accumulation of unpaired large subunits (Cheng et al. 2019). In contrast, our observations in C. elegans indicate a coordinated response across both subunits at the RNA and TE levels: reductions in ribosomal proteins from either subunit led to a generalized overexpression at the RNA levels (p ≤ 0.01 except rps-10(0)/+, ROAST–rotation gene set testing, (Wu et al. 2010)) but a decrease in TE levels (p ≤ 0.05 for rpl-33(0)+ and rpl-5(0)/+, ROAST) (Fig. 3B, S3B). To assess potential accumulation of unpaired subunits, we analyzed 28S/18S ratios by measuring the area underneath 28S and 18S peaks in total RNA samples from RP haploinsufficient and wildtype animals using Bioanalyzer. We did not observe significant changes in this ratio (Fig. S3C, p > 0.05; RP haploinsufficient mutants were compared to wild-type controls using Student’s t-test with Bonferroni correction). Given that such a response was not observed in yeast, we asked whether a similar response occurs in response to DBA-specific RP gene reductions in human hematopoietic cells (Khajuria et al. 2018). Re-analyzing the RNA-seq and Ribo-seq datasets from this study revealed that the results in C. elegans mirror those from human RPL5 and RPS19 knockdowns in hematopoietic cells, where TE of all subunits are significantly decreased (Fig. S3D, p<0.05, ROAST, Table S6). Hence, the human cellular response is more similar to that in C. elegans, in comparison to yeast, highlighting a potentially conserved regulatory mechanism regulating translation for maintaining ribosome production under conditions of single-copy loss of RP genes.

In RP haploinsufficient mutants, we observed upregulation of glutathione transferase activity and potential activation of SKN-1, a critical factor in maintaining cellular redox balance and facilitating mitochondrial retrograde signaling (Palikaras, Lionaki, and Tavernarakis 2015; Dinkova-Kostova and Abramov 2015). These observations suggest a potential impairment in mitochondrial function. In particular, genes of the electron transport chain (ETC) require coordinated translation by both cytoplasmic and mitochondrial ribosomes, creating a vulnerability in proteostasis (Soto et al. 2022).

We next compared changes in RNA expression and translation efficiency (TE) of ETC components across RP mutants. While we did not observe significant changes in the expression of nuclear-encoded or mitochondrially-encoded ETC components at the RNA level (Fig. 4A, left, p>0.1, ROAST); a distinct pattern emerged at the TE level. Nuclear-encoded components remained unaffected (p>0.4), whereas mitochondrial-encoded ETC components showed reduced TE across all mutants (Fig. 4A, right, p<0.005 for rpl-33(0)/+ and rpl-5(0)/+ and p ≤ 0.16 for rps-10(0)/+ and rps-23(0)/+, ROAST). This discrepancy suggests that the impaired mitochondrial translation may indirectly affect the stoichiometry of ETC. Although proteomics-level measurements lacked sufficient coverage to quantify corresponding changes in protein abundance for ETC components, the mitochondrially encoded Complex-I component, NDUO-5, was notably underexpressed in rps-10(0)/+ animals (~70% reduction, p_adj = 0.2, Fig. S4A, Table S1).

Given the marked reduction in the TE of mitochondrial-encoded ETC components across all mutants, we investigated if there was a corresponding change in the abundance of mitochondrial ribosomes. Mitochondrial ribosomal protein mRNAs were generally mildly reduced (p < 0.05 for rpl-33(0)/+, rpl-5(0)/+ and rps-23(0)/+; p <0.2 for rps-10(0)/+, ROAST), however no consistent changes were observed at the TE levels (Fig. S4B, p > 0.5 for all mutants). Similarly, changes at the protein level were not consistent with a wider distribution of mitochondrial ribosomal proteins in rps-23(0)/+ animals (Fig. S4B). These results suggest that the observed variations in TE across ETC components may be due to impaired mitochondrial translation rather than differences in mitochondrial ribosome abundance.

The trend of elevated expression of gst genes and SKN-1-mediated oxidative stress regulation, both indicative of increased ROS, alongside reduced TE of mitochondrially-encoded ETC components in RP mutants, pointed towards potential mitochondrial dysfunction. Previous studies on mitochondrial dynamics established a link between mitochondrial dysfunction and morphological changes, particularly under stress (Zemirli, Morel, and Molino 2018). Specifically, fission-induced mitochondrial fragmentation, characterized by round mitochondria as opposed to networked mitochondria, is associated with increased ROS and elevated oxidative stress (Yu, Robotham, and Yoon 2006; Qi et al. 2011).

To investigate the impact of RP haploinsufficiency on mitochondrial morphology, we introduced a body-wall-specific nuclear and mitochondrial GFP marker into the backgrounds of RP haploinsufficient mutants (Ahnn and Fire 1994). No differences were observed at 16°C up to the L4 stage, but upon transferring the L4 animals to 23°C and imaging by day three of adulthood, we detected partially penetrant increases in mitochondrial fragmentation across all mutants (Fig. 4B). To quantify fragmentation, we measured the convexity (degree to which shape differs from its convex hull), defect area (area outside of convex hull), and skeleton branch length of each individual mitochondria. Our results reveal that on average, mitochondria from rps-10(0)/+, rps-23(0)/+, and rpl-5(0)/+ are significantly more convex, or less networked, compared to wild-type control animals (Fig. 4C, first plot, p<0.001, Student’s t-test). Similarly, all four mutants had a significant average decrease in mean defect area and branch length relative to wild-type control animals (Fig. 4C, second and third plots, p<0.001, Student’s t-test). Together, these results suggest that RP mutants have increased mitochondrial fragmentation that may indicate underlying mitochondrial dysfunction that is both temperature-sensitive and age-related. These results suggest that RP mutants struggle to maintain mitochondrial homeostasis, particularly under moderately higher temperatures, pointing towards a vulnerability in their ability to adapt to environmental conditions.

The changes in mitochondrial morphology among RP mutants prompted us to investigate mitochondrial function. Given the link between mitochondrial structure and metabolism (Westermann 2010), we next evaluated mitochondrial membrane potential using MitoTracker Red CMXRos staining (Sarasija and Norman 2018) (Fig. S4C) and analyzed the overall energy status of the animals by measuring their relative ADP/ATP ratios (Palikaras, Lionaki, and Tavernarakis 2015).

The rps-10(0)/+ mutants exhibited significant decreases in mitochondrial membrane potential, as indicated by significant MitoTracker accumulation (p<0.001, independent Student’s t-test) and elevated ADP/ATP ratios (p=0.003, paired Student’s t-test), compared to stage-matched controls (Fig. 5A and 5B). These results reveal that the rps-10(0)/+ mutants display a significant disruption in cellular energy homeostasis in addition to compromised mitochondrial membrane potential.

Having observed disruptions in mitochondrial function in rps-10(0)/+ but not large subunit RP mutants, we assayed oxygen consumption rates in small subunit RP mutants. rps-10(0)/+ mutants showed a reduction in oxygen consumption in comparison to stage matched controls (p<0.05, paired Student’s t-test), a trend not observed in rps-23(0)/+ (p>0.1, paired Student’s t-test) (Fig. 5C). To ensure decreased oxygen consumption is not due to smaller body size, we quantified the body length and width of rps-10(0)/+ animals, relative to controls, and found no significant difference between them (Fig. S4D, width p=0.48, length p=0.080)

Given that energy and mitochondrial functionality changes were specific to the rps-10(0)/+ mutants, we sought to determine if they were caused by a background mutation introduced during the CRISPR-mediated early stop integration or possibly due to the maternal inheritance of defective mitochondria. To address these possibilities, rps-10(0)/+ hermaphrodites were used to reintroduce the wild-type rps-10 sequence, leveraging the unique SuperPAM (GGNGG) sequence inserted during the creation of the mutation (Fig. 5D). This procedure yielded two independent wild-type rescue strains from the F1 generation, each carrying two copies of the reverted wild-type rps-10 gene. These strains displayed MitoTracker intensities and oxygen consumption rates comparable to control groups, suggesting the observed mitochondrial defects are specific to the single-copy loss of the rps-10 gene and not related to mitochondrial biogenesis or maternal inheritance (Fig. 5E and 5F, p>0.4, independent and paired Student’s t-test respectively).

Additionally, we investigated whether mitochondrial defects observed in rps-10(0)/+ mutants stemmed from changes in mitochondrial biogenesis or overall abundance. Specifically, we measured mitochondrial DNA levels and observed no differences in mitochondrial DNA content when compared to stage matched wild-type controls (Fig. S4E). Furthermore, we integrated fluorescent reporters for outer and inner mitochondrial membrane components (TOMM-20::Tag-RFP and COX-4::GFP (Raiders et al. 2018)) into rps-10(0)/+ mutants and detected no significant differences in fluorescence intensity compared to stage-matched controls (Fig. 5G, p>0.3, independent Student’s t-test). Taken together, these results indicate that the observed mitochondrial respiration deficits in rps-10(0)/+ animals are due to defects in mitochondrial functionality rather than decreased abundance.

Considering the established connections between eIF2α, the integrated stress response (ISR), and mitochondrial function (Baker et al. 2012; Raini et al. 2017; Back 2020), we investigated the role of GCN1, a key factor in the ISR that activates eIF2α kinase GCN2 (Garcia‐Barrio et al. 2000; Hinnebusch 2005). GCN1’s activation of GCN2 requires its interaction with ribosomes (Marton et al. 1997; Sattlegger and Hinnebusch 2005) and a cryo-EM study has elucidated that GCN1 binds to collided ribosomes at the interface of the small and large subunits for its role in ribosome quality control (RQC) (Pochopien et al. 2021). Although this structure does not show a physical interaction between RPS10 and GCN1, such a link was implicated through yeast two-hybrid interactions, and correlation between decreased levels of RPS10 and reduced eIF2α phosphorylation, implying a compromised activation of GCN2 (S. J. Lee, Swanson, and Sattlegger 2015). Decreasing gcn-1 expression, either through RNAi or a loss-of-function mutation (n4857) (Hirose and Horvitz 2014), resulted in modestly lower oxygen consumption rates (~30% and 20% reduction respectively with p-values 0.12 and 0.06, paired Student’s t-test) (Fig. 5H, left plot). RNAi-mediated knockdown of rps-10 resulted in approximately a 50% reduction in oxygen consumption in a wild-type background (Fig. 5H, middle plot, p=0.097). However, in gcn-1(n4857) mutants, rps-10 RNAi did not exacerbate the decrease in oxygen consumption (Fig. 5H, right plot, p=0.5, paired Student’s t-test), despite causing developmental delays, thereby validating the effective knockdown of rps-10. As a control, we used RNAi against gas-1, a gene encoding an ETC component that could lower oxygen consumption even more in the gcn-1(n4857) mutants suggesting a floor level of oxygen consumption was not obtained in gcn-1(n4857) mutants (Fig. 5H, right plot, p=0.02, paired Student’s t-test). These results overall suggest that (i) disruption in gcn-1 could lead to reduced mitochondrial function, (ii) in the absence of functional GCN-1, RPS10’s effect on mitochondrial function might be minimized or that a compensatory mechanism is activated.

The similarity of symptoms between DBA and Pearson syndrome (Gagne et al. 2014), the role of mitochondria in hematopoiesis (Fontenay et al. 2006), and the observed lack of coordination in the expression of mitochondrial components in DBA patients (Panici et al. 2021) led us to re-examine the RNA expression and translation efficiency of genes upon knockdown of the two most frequently mutated RPs (shRPS19 and shRPL5) in DBA patients using hematopoietic cells (Khajuria et al. 2018). Specifically, we investigated the conserved orthologs and expression differences in rpl-5(0)/+ C. elegans mutants and RPL5 knockdown in human hematopoietic cells (Table S7). We particularly wondered if there is conserved unidirectional or bidirectional regulation at the RNA and TE levels impacted by reduced levels of RPL5 across C. elegans and human cells. Similar to our previous results, which suggested translational control to maintain ribosome numbers (Fig. 3B, S3D), we observed significant functional GO term enrichments for both cytoplasmic large and small ribosomal subunits (6.5 and 5.3 fold enrichment, respectively, with p-values < 0.01), characterized by increased RNA levels while TE was decreased, when data from C. elegans and humans were combined (Fig. S5A, Table S7).

Among unidirectional GO enrichments, we observed ~2 fold enrichment in the categories of DNA unwinding and mitochondrial matrix among genes that were underexpressed both at the RNA and TE level in C. elegans and human cells in response to RPL5 reduction (Fig. S5A, Table S7). Furthermore, distinct GO categories related to mitochondrial components, especially those associated with the ETC and mitochondrial ribosomes, showed an increase in RNA levels coupled with a decrease in TE (Fig. S5A). This pattern is consistent with our results in C. elegans, suggesting that the effects on the ETC are indirect, and primarily due to impaired mitochondrial translation. Particularly, significant enrichment of the categories of complex I of the ETC (NADH dehydrogenase activity, Fig. 6A) and the large subunit of mitochondrial ribosomes (both categories are 2.8 fold enriched, p-values < 0.01, Fig. 6B) indicate a potentially conserved translational buffering mechanism for mitochondrial ribosomes and ETC components in the face of reductions in cytoplasmic ribosomal machinery.

While we observed significantly reduced TE for nuclear ETC components, and both subunits of the mitochondrial ribosome following reductions in RPS19 and RPL5 (p < 0.05, ROAST, Fig. 6C; Fig. S5B), mitochondrially encoded ETC components were increased at the translation efficiency level (Fig. 6C, p=0.14, p=0.02, for RPL5 and RPS19 reduction respectively, ROAST). These findings overall highlight a broadly consistent effect of RPL5 reduction on the RNA and TE of critical mitochondrial components in both C. elegans and humans, pointing towards a conserved regulatory mechanism.

To elucidate the gene expression regulatory mechanisms linking mitochondria and ribosomes in human cells, we performed an unbiased co-expression analysis at the transcription and translation levels across lymphoblastoid cells derived from 13 individuals. Specifically, we quantified the similarity of expression patterns across all genes using a compositional proportionality metric (Thomas P. Quinn et al. 2017; T. P. Quinn et al. 2019). This comprehensive analysis unveiled a significant correlation between ribosomal and mitochondrial membrane genes, evidenced by over 1000 significant interactions (Fig. 7A, Table S8). These findings suggest a highly coordinated regulation of ribosomal and mitochondrial gene expression in human cells (Fig. S5C), highlighting the interplay between these essential cellular components.

Building upon these insights into the coordinated expression of ribosomal and mitochondrial genes, we investigated how reductions in cytoplasmic ribosomal proteins affect mitochondrial function, a subject that has been relatively unexplored in human cells. We used K562 leukemia cell line to examine the impacts of reduced levels of specific ribosomal protein transcripts, RPS10, RPL35A (the ortholog of C. elegans rpl-33), RPL5, and RPS23. Using siRNA knockdowns, we achieved an approximately 50% reduction in their transcript levels (Fig. S5D). The mitochondrial membrane potential, assessed using MitoTracker, demonstrated significant decreases in activity following RPS10 reduction (Fig. 7B, p = 5e-4, independent Student’s t-test), accompanied by an increase in the ADP/ATP ratios (Fig. 7C, p<0.05 for RPS10 siRNA, paired Student’s t-test), highlighting the critical role of ribosomal proteins in supporting mitochondrial energy metabolism.

These observations collectively emphasize a balance between ribosomal and mitochondrial gene expression, which is crucial for cellular energy production and metabolic health. The differential regulation of translation efficiency for mitochondrially versus nuclear-encoded ETC components suggests adaptability to counteract the effects of ribosomal protein reduction, with a substantial impact of RPS10 reduction on mitochondrial activity and energy metabolism.

C. elegans strains used in this study were sourced from the Caenorhabditis Genetics Center (CGC), supported by the NIH Office of Research Infrastructure Programs (P40 OD010440). Standard cultivation practices involved growing the nematodes on Nematode Growth Medium (NGM) plates, which were seeded with the E. coli OP50 strain. To ensure genetic stability, particularly to minimize recombination and prevent the loss of balancer chromosomes, we incubated the animals at 16°C, with the exception of experiments investigating mitochondrial morphology under stress (see Fig. 4B–C). In these experiments, we did not initially notice differences in the mitochondrial structures of young adult worms maintained at 16°C. To further explore mitochondrial dynamics under stress, we allowed the animals to mature into three-day-old adults at 16°C, then transferred adults to a stress-inducing environment at 23°C for two days prior to the imaging procedures.

Balancer chromosomes are denoted as follows: +¹ = tmC20, +² = tmC5, +³ = mIn1, +⁴ = nT1. Balancer chromosomes were used in all control strains, except for the analysis of mitochondrial morphology (Fig. 4B–C). For the C. elegans experiments depicted in Figures 1–4 and Figure 6, the following balancers were used for each haploinsufficient ribosomal protein strain, along with corresponding wild type/balancer controls: rps-10(0)/+¹, rps-23(0)/+², rpl-5(0)/+³, and rpl-33(0)/+³. For Fig. 5C, rps-23(0)/+⁴ was used, as +⁴ (nT1) is homozygous lethal and facilitates the collection of a large sample of animals.

Strains ESC614 and ESC615 were derived from heterozygous animals carrying the genotype rps-10[cc2557,T8X]/tmC20, [unc-14(tmIs1219) dpy-5(tm9715)] I. Young adult heterozygotes were injected with a CRISPR injection mix, which included a 2.5 μM homologous recombination template (ESC-AS-130), 50 ng/μL guide RNA plasmid (pAS14), and 50 ng/μL Cas9-expressing plasmid (pDD132), adapting the co-conversion method (Arribere et al. 2014). Rescue mutations were initially selected by identifying balancer chromosome, tmC20-free adult animals, characterized by the absence of pharyngeal GFP markers and the avoidance of both the developmental arrest associated with homozygous rps-10(0) and the uncoordinated phenotype of tmC20, [unc-14(tmIs1219) dpy-5(tm9715)] progeny. These potential rescue mutations were subsequently confirmed through PCR amplification and Sanger sequencing.

We used CRISPR-Cas9 to generate a strain expressing TOMM-20::TagRFP in an otherwise N2 background. We annealed and inserted sense and anti-sense guide DNA oligos (ESC-RR-21 and ESC-RR-22) into pRB1017, a vector that expresses guide RNA with U6 promoter. Cas9 protein was expressed from eft-3 promoter using the plasmid pDD162. The 5′ and 3′ homology arms were amplified ~500 bp upstream of tomm-20 stop codon using oligos ESC-RR-17 and ESC-RR-18, and ~500 bp downstream of stop codon using ESC-RR-19 and ESC-RR-20. C. elegans genomic DNA was used for these amplifications. These homologous arms were used to replace the ccdB in pDD284(TagRFP-TŜEĈ3xFlag). All plasmids for microinjection were purified using the PureLink HiPure Plasmid Miniprep Kit (Invitrogen, #K210002). Oligo sequences used to generate these plasmids are provided below as well as in Table S9. The downstream injection and selection process was done exactly as explained in the supplementary methods section utilizing a self-excising hygromycin selection-based CRISPR-Cas9 protocol (Dickinson et al. 2015).

All strains and DNA sequences that were generated or used in this study are provided below in the methods section as well as in Table S9.

C. elegans animals were bleach-synchronized (adult animals were incubated with 0.5M NaOH and 1% NaClO for 6 minutes to extract eggs) and subsequently grown on NGM plates until they reached the L4 stage. At this point, the animals were collected using 50 mM NaCl. To ensure the removal of bacteria and prepare for proteomic analysis, the collected animals underwent a series of serial centrifugations at 300 × g for 10 minutes with 3% sucrose in 50 mM NaCl. The animals were then resuspended in Laemmli Buffer (Bio-Rad, 1610737), supplemented with PMSF (Thermo Fisher Scientific, 36978) and BME (Sigma Aldrich, 11411446001), and immediately flash-frozen. Next, the samples were subjected to mechanical disruption via manual bead-beating to ensure thorough digestion of the protein content. The resulting protein fractions were then loaded onto NuPAGE Bis-Tris Gels (4–12%) (Thermo Fisher Scientific, NP0335BOX) and ran using MES SDS Running Buffer (Thermo Fisher Scientific, B0002). The gel was briefly stained by Coomassie Staining (Bio-Rad, #1610786). Post-electrophoresis, the gels within the top stacking portion were cut into sections using a razor for further processing. The excised gel sections underwent trypsin digestion before the peptides were desalted. These prepared samples were then analyzed using a Dionex Liquid Chromatography system coupled with an Orbitrap Fusion 2 mass spectrometer. The analytical run was conducted over a 120-minute period to ensure comprehensive peptide identification and quantification. Protein identification was provided by the UT Austin Center for Biomedical Research Support Biological Mass Spectrometry Facility (RRID:SCR_021728).

For data analysis, raw outputs, including Label-Free Quantification (LFQ) values and peptide counts, were processed using Proteome Discoverer version 2.5 (Orsburn 2021). This software facilitated the mapping of the data against the C. elegans reference database. Further quantitative analysis was performed utilizing the DEP (Differential Expression Proteomics) package in R, accessible via the Bioconductor project (https://rdrr.io/bioc/DEP/man/DEP.html). Proteomics analysis by DEP is provided (Table S1).

For the body area and length assays, animals were synchronized through a two-hour egg-laying period and subsequently grown until L4 development. Prior to imaging, animals were anesthetized at room temperature using 10mM Levamisole. The animals were imaged using a Leica Stellaris 8 confocal microscope with a Leica K5 microscope camera. Leica Application Suite X was used for image acquisition. Images were acquire through differential interference contrast (DIC) with a 20x (NA 0.4) objective. ImageJ software was used to measure the body area by drawing segmented lines along the length of each animal from head to tail (Fig. 1D). Body length was quantified by drawing segmented lines from head to tail, and body width was quantified by drawing segmented lines across the midsection of the body (Fig. S4D). Body area (Fig. 1D) and body length and width (Fig. S4D) were measured using ImageJ software. All measurements were normalized using the median body area of the respective wild type/balancer control group. Statistical differences between the average normalized body area of mutant and control animals were measured using a two-tailed Welch’s two-sample t-test.

Animals were synchronized using a two-hour egg-laying window. Heterozygous animals, either carrying a ribosomal protein gene mutation and a balancer chromosome or a wild-type gene with the corresponding balancer, were individually transferred to fresh NGM plates. A single animal was moved to a new plate roughly every 24 hours. Hatched progeny from each animal were counted (Fig. 1E and S1B). Differences in the mean brood size of each mutant, compared to its respective control, were quantified using a two-tailed Student’s t-test.

The lifespan assay was conducted on solid Nematode Growth Media, both with and without the addition of Fluorodeoxyuridine (FUDR)(Van Raamsdonk and Hekimi 2011; Sutphin and Kaeberlein 2009). Following synchronization via egg-laying, animals were placed on NGM plates with or without 0.5 μM Fluorodeoxyuridine (FUDR) (Van Raamsdonk and Hekimi 2011; Sutphin and Kaeberlein 2009) (Fig. S1C and 1F, respectively). The animals were grown at 16°C, and the survival of each nematode was assessed every 24 hours. Animals on the NGM plates without FUDR were transferred every two days. For FUDR plates, concentrated OP50 spun at 2000 rpm was used as a food source during the lifespan assay. Statistical analysis was conducted using the Log-rank test with Bonferroni correction for multiple comparisons (Yang et al. 2011).

Puromycin pulse labeling was performed on three replicates. Animals were bleach-synchronized and collected at the L4 stage. Plates with L4 worms were supplemented with OP50 and 0.5mg/ml puromycin dihydrochloride from Streptomyces alboniger (Sigma-Aldrich, P8833). These worms were incubated at 20℃ and pulse-labeled for 4 hours (Weaver et al. 2020). After labeling, worms were washed twice with 5% sucrose and twice with 50 mM NaCl before being flash frozen with 5μl glass beads.

For worm lysis, Laemmli buffer supplemented with PMSF and β-mercaptoethanol (BME) was added to each sample before bead beating for 30 seconds. Samples were spun down through centrifugation for 2 minutes at 15,000 g. The samples were then boiled at 95℃ for 5 minutes. An additional 2-minute centrifugation was performed to clear debris prior to measuring protein concentration with Qubit Protein Assay Kit (Thermo Scientific, Q33211).

Approximately 16–17 μg of protein per sample was loaded on Invitrogen^™ NuPAGE^™ Bis-Tris Mini Protein Gels (4–12%, 1.0–1.5 mm, Invitrogen, NP0335) and separated at 80–120 V in 1X MES SDS running buffer. Proteins were transferred onto a PVDF membrane (Thermo Scientific, 88518) at 30V for 1 hour. Membranes were blocked in 1X TBS-T containing 5% milk. Puromycin and actin were probed with primary antibodies: Anti-Puromycin Antibody, clone 12D10 (Mouse, 1:2000, Sigma-Aldrich, MABE343) and Anti-Actin Monoclonal Antibody [clone: C4] (Mouse, 1:500, VWR, 10221–880), respectively. A secondary antibody, Goat Anti-Mouse IgG H&L (HRP) (1:10,000, Abcam, ab6789) was used in conjunction with a chemiluminescent substrate (Thermo Scientific, 34579) for visualization.

All replicates were quantified by recording the average intensity in the range of approximately 20–35 kDa using ImageJ, with the quantified actin band used for normalization.

For the acute oxidative stress and heat assays (Fig. 2C, S2A, S2B, S2C and S2E), C. elegans were grown at 16°C. To induce proteotoxicity, translation or TORC1 inhibition, bleached eggs were pre-grown on NGM plates with the following drugs prior to the stress assay: final concentrations of 100 μM rapamycin (MedChemExpress, HY-10219), 25 nM bortezomib (MedChemExpress, HY-10227), and 25 nM cycloheximide (Sigma, 01810–5G).

For the oxidative stress assay, L4 animals were transferred to fresh OP50-seeded NGM plates spiked with 0.2 M paraquat (Thermo Fisher Scientific, AC227320050). For heat stress, L4 animals were subjected to a temperature of 37°C to induce heat stress.

For both assays, the survival rate was evaluated at room temperature across various time points, determining mortality based on the lack of movement within 15 seconds after being gently prodded with a platinum pick. Statistical analysis was conducted using the Log-rank test with Bonferroni correction to compare the survival distributions of for multiple groups (Yang et al. 2011).

For Ribo-seq and RNA-seq procedures, L4 staged heterozygous larvae were collected, and the remaining bacteria was cleaned up using a 5% sucrose solution with 50 mM NaCl. The animals were stored in 300 μl of lysis buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, and 5 mM MgCl2, 1 μl of 1M DTT, 10 μl of 10% Triton-X, 100 ug/ml cycloheximide) and flash frozen in liquid nitrogen. These samples were then stored at −80°C until further use.

The frozen animal pellets were then ground to a fine powder to break cuticles in liquid nitrogen using a mortar and pestle, and the powder was collected in a 1.5 ml tube. The powder was allowed to thaw on ice before 20 units of Turbo DNAse (Thermo Fisher Scientific, AM2238) were added. Each lysate sample was divided into two parts for RNA-seq and Ribo-seq, and 1 ml Trizol (ThermoFisher Scientific, 15596026) was added to the RNAseq aliquot. After a brief vortex and incubation on ice for 15 minutes, the RNA concentration was measured using a Qubit RNA BR assay. RNAse I (Thermo Fisher Scientific, EN0601) was added to each Ribo-seq sample at a ratio of 150 units per 30 μg of RNA and incubated for 30 minutes at room temperature. The RNAse I reaction was stopped with a final concentration of 25 mM ribonucleoside vanadyl complexes (Sigma, R3380). The samples were loaded onto the 34% sucrose cushion prepared in lysis buffer and spun at 70,000 rpm for 4 hours at 4°C using a TLA 100.3 Rotor in an Optima Ultracentrifuge (Beckman, 361889). After centrifugation, the supernatant was removed to isolate the pellet, which was then dissolved in 1 ml Trizol. The samples with Trizol added for RNA-seq and Ribo-seq were briefly vortexed. Following a 5-minute room temperature incubation, 200 μl of chloroform were added, and the samples were spun at 15,000 rpm for 10 minutes, then the aqueous layer was transferred to a new tube. The final concentration of 50 mM 3 M NaAcetate (pH 5.5), 5 mM MgCl2, and 2 μl Glycoblue coprecipitant (ThermoFisher Scientific, AM9515) was added along with 500 μl isopropanol, and the samples were incubated overnight at −20°C. The next day, the samples were spun at 15,000 rpm, at 4°C for 60 minutes and the pellets were washed with 80% ethanol. The pellets were dissolved in DEPC treated water before the subsequent steps. The Ribo-seq samples were treated with T4 PNK (ThermoFisher Scientific, EK0031) in T4 PNK buffer for 30 minutes at 37°C, and were run on a 15% TBE urea gel. After staining SYBR Gold (ThermoFisher Scientific, S11494) and imaging, ribosome footprints between 26 and 34 bases were cut out for further processing with D-Plex Small RNA sequencing kit with unique molecular identifiers (UMI) (Diagenode, C05030001, C05030021) for library preparation. For RNA-seq library preparation, SMARTer Stranded RNA-Seq Kit (TakaraBio, 634837) was used. Three independent biological replicates were performed. Each wild-type control included two samples that are time-matched and stage-matched to RP mutants to avoid any gene expression changes that are due to the observed developmental delay in RP mutants. Transcriptome mapping reads for the third replicate were not sufficient (less than 300K reads), therefore the third replicate was removed from further analyses.

To analyze human shRPS19 and shRPL5 knockdown experiments in hematopoietic cells, raw data was downloaded from NCBI GE (GSE89183, (Khajuria et al. 2018)).

For read mapping and further processing of the data, Riboflow nextflow pipeline was utilized (Ozadam, Geng, and Cenik 2020) (https://github.com/ribosomeprofiling/riboflow). Before mapping the unique identifier barcodes that were added in Ribo-seq, libraries were collapsed to count the number of unique RNA molecules and adapters which were removed. Reads were mapped against the C. elegans transcriptome (Ensembl, WBCel235) and human transcriptome (Gencode, GRCh38.p14). The Ribo-seq and RNA-seq counts obtained from the pipeline were analyzed using the edgeR pipeline in R (Robinson, McCarthy, and Smyth 2010). To investigate differences in translation efficiency (TE), ribosome-bound RNA (Ribo) levels, and RNA expression levels across the samples, three specific contrasts were constructed. Subsequently, a quasi-likelihood F-test (glmQLFTest) was employed to assess these contrasts and adjust for multiple testing errors using a false discovery rate (FDR) approach, setting a p-value threshold of 0.05 for significance. The contrast specifically designed to evaluate translation efficiency was defined as `TE_RPmutantvsControl = (RP.Ribo - RP.RNA) - (Control.Ribo - Control.RNA)`. This approach allowed for comparison of the TE between ribosomal protein (RP) mutants and their respective controls. When examining significance across all RP mutants, we categorized samples into two groups: RP mutants versus controls. However, when analyzing individual mutants, each was compared to both stage matched and time matched controls, within the same genetic background.

RNA-seq, Ribo-seq and TE analysis results by EdgeR are provided in Table S2, S4, and S6.

Mitochondrial genes were extracted using Mitocarta 3.0 (Rath et al. 2021). C. elegans and human gene orthologs were extracted using BioMart (Smedley et al. 2015). Gene expression values of all conserved orthologs between C. elegans and humans are provided in Table S7.

For ROAST multivariate gene expression analysis, raw counts for both RNA and TE were normalized using the edgeR package and subsequently transformed with respect to the mean-variance trend using the voom function from the limma package (Ritchie et al. 2015). Rotation gene set tests were conducted on transformed counts for RNA and TE using the roast function (Wu et al. 2010) from limma.

For mitochondrial imaging, RP mutants carrying balancers and N2 worms were crossed with male PD4251 (ccIs4251 [myo-3p::GFP::LacZ::NLS + myo-3p::mitochondrial GFP + dpy-20(+)] I; vsm-1(ok1468) IV) worms. From these crosses, L4 animals carrying a single copy of the myo-3p::GFP mitochondrial marker were recovered from the transient F1 generation for use in mitochondrial imaging and analysis. These L4 animals were transferred to 23⁰C where they were incubated until day three of adulthood.

Animals were then immobilized using 10 mM Levamisole and imaged using a Leica Stellaris 8 Confocal Microscope. Leica Application Suite X was used for acquiring images. (Fig. 4B). Images were acquired at room temperature with a 63x (NA 1.4) objective. At least nine nematodes were imaged for each mutant, spanning three biological replicates, with 5–10 cells being imaged in each nematode.

Image analysis was performed using the ilastik v1.4.0 “Pixel Classification + Object Classification” pipeline (Berg et al. 2019). Briefly, raw images were input for pixel classification to separate mitochondria and nuclei from the background based on GFP signal intensity. Nuclear-localized GFP was excluded for the remaining analyses, and only mitochondrial-localized GFP was used. The object classification portion of the pipeline was used to extract various features of each individual mitochondrial object, including convexity, defect area, and branch length. For each condition more than 10,000 data points were collected. Statistical comparison of mean features, specifically convexity, mean defect area, and mean branch length, between control and mutant samples was carried out using a Student’s t-test, performed with the ť.tesť function in R (https://www.R-project.org/) to compare differences in average feature measurements across mutant and control animals. Visualization of feature distributions was performed using the `ggploť package in R (Fig. 4C, (Villanueva and Chen 2019)). The pipeline and code used in this analysis is available at https://github.com/raqmejtru/mito_image_analysis/.

The Mitotracker staining of C. elegans was adapted from the protocol (Sarasija and Norman 2018). Following bleach-synchronization, L4 animals and OP50 bacteria were collected using M9 buffer. Samples were incubated with 1 μg/mL Mitotracker CMXRos (Thermo Fisher Scientific, M7512) for 6 hours at 20°C. To remove excess dye, the nematodes were then washed twice with M9 buffer. After washing, the animals were transferred to fresh NGM plates seeded with OP50 and allowed approximately one hour for foraging, which helps in clearing any dye that might have been nonspecifically accumulated in the gut. For quantitative imaging, the stained nematodes were mounted on slides prepared with 3% agarose in M9 buffer, ensuring no anesthetics were used that could potentially interfere with the fluorescence. Imaging was performed at room temperature using a 20x (NA 0.4) objective on a Leica SPE microscope using fixed fluorescent exposure and a Leica K5 microscope camera. Leica Application Suite X was used for image acquisition. The intensity of the Mitotracker accumulation per area for each animal was quantitatively measured using ImageJ software. Staining specificity was assessed through co-localization studies using a C. elegans strain with a CRISPR-engineered knock-in of cox-4 gene tagged with GFP (cox-4::GFP), serving as a marker for mitochondrial inner membranes. These co-localization analyses were conducted using a Leica Stellaris 8 confocal microscope with a 63X (NA 1.4) objective and Leica Application Suite X. For visualization and confirmation of the staining pattern’s specificity, we compared the images to the known localization of the COX-4::GFP signal within the mitochondria (Fig. S4C).

Bleach-synchronized L4 animals were transferred to bacteria-free NGM plates to ensure cleanliness for the assay. A selection of 50 animals was made to specifically exclude those with homozygous balancer chromosomes, and these selected worms were placed in 50 μL of M9 buffer. The prepared samples were flash frozen in liquid nitrogen and stored at -80°C until further analysis. The method for measuring the ADP/ATP ratio in C. elegans was adapted from the protocol (Palikaras, Lionaki, and Tavernarakis 2015). The ADP/ATP ratio was quantitatively measured from lysis supernatant using the Sigma ADP/ATP Assay Kit (Sigma-Aldrich, MAK135–1KT), following the manufacturer’s instructions. Measurements were carried out on a Glomax Luminescence Microplate Reader (Promega).

To measure oxygen consumption in C. elegans, bleach-synchronized animals were grown on NGM plates seeded with OP50 or HT115 E. coli expressing specific or non-target siRNAs at 16°C until the L4 stage. The animals were collected using a 50 mM NaCl solution and underwent a cleaning process to remove bacteria. This involved centrifugation at 800 × g for 1 minute, repeated three times with M9 buffer to ensure thorough cleansing. After cleaning, the nematodes suspended in the M9 buffer were transferred into the sealed and pre-calibrated chamber of O2k-FluoRespirometer (Oroboros), for the determination of the oxygen consumption rate. The oxygen consumption rate value was then normalized based on the exact number of animals that were introduced into the measurement chamber, to reflect the metabolic rate per individual animal.

To measure fluorescence intensity, L4 stage-matched animals expressing TOMM-20::TagRFP and COX-4::GFP, markers for mitochondrial localization, were imaged using a Leica SPE Fluorescence DIC microscope with a 20x (NA 0.4) objective lens and a Leica K5 microscope camera. Images were acquired using Leica Application Suite X. For quantifying the fluorescence intensity per animal, the captured images were analyzed using ImageJ software. Specifically, for the measurements of COX-4::GFP fluorescence intensity, care was taken to exclude the pharyngeal area from analysis. This precaution was necessary to avoid interference from the mVenus marker present in both the rps-10(0)/tmC20 and wild type/tmC20 control strains.

We conducted a detailed analysis of gene expression samples of 13 human individuals from diverse genetic backgrounds. The samples were selected to have Ribo-seq paired with RNA-seq data from the study GSE65912 (C. Cenik et al. 2015), utilizing the RiboFlow toolkit (Ozadam, Geng, and Cenik 2020). To refine the ribosome profiling dataset further, we applied a winsorization technique to adjust for potential PCR duplication artifacts, capping nucleotide counts at the 99.5th percentile to address over-amplified outliers. We excluded 166 human genes identified as lacking polyA tails to ensure the analysis focused on high-quality gene counts. Following this exclusion, both RNA-seq and Ribo-seq data were normalized using counts per million (CPM), selecting genes with a CPM greater than 1 in over 70% of the samples for further analysis. This process resulted in the retention of 10,145 human genes.

TE and RNA value for each gene in each sample was calculated as explained in Ribo-seq analysis section. To assess coordinated expression and TE among genes, we used the ‘lr2rho’ function from the ‘propr’ R package (Thomas P. Quinn et al. 2017), inputting centered log ratio values of TE or RNA expression for 10,145 human genes. The resulting rho values, ranging from −1 to 1, facilitated the generation of CoTE and co-expression matrices. Gene sets were curated from the Gene Ontology and KEGG pathway databases, focusing on mitochondrial membrane, ribosomal, and cilia-associated genes. Performing overlap analysis with these gene lists identified 445 mitochondrial, 65 ribosomal, and 124 cilia-associated genes. For comparability, we selected a subset of 65 genes from the cilia-associated gene set to match the ribosomal gene count. The TE and RNA levels of these 575 genes were then analyzed, focusing on interactions where the absolute rho value exceeded 0.75. Gene interactions with a rho value exceeding 0.75 were visualized using Cytoscape version 3.9.1 (Shannon et al. 2003). All Ribo-Mito interactions that had rho values higher than 0.75 are provided (Table S9).

K562 Cells (ATCC) were maintained in RPMI (Thermo Fisher Scientific,11879020) supplemented with 10% Fetal Bovine Serum/FBS (Gemini Bio, 900-108) and Glucose 2g/L (Thermo Fisher Scientific, A2494001) and incubated at 37°C and 5% CO₂. K562 cells were cultured until reaching 80% confluency. The cell count was assessed using a hemocytometer and Trypan Blue (ThermoFisher Scientific, 15-250-061) staining. The cells were centrifuged at 600 × g for 5 minutes, after which the supernatant was replaced with Opti-MEM (Thermo Fisher Scientific, 31985070) to adjust the cell count to 10⁶ cells per mL. siRNA (IDT), at a concentration of either 12.5 nM or 6.25 nM, was diluted in Opti-MEM and combined with Dharmafect 1 Transfection Agent (Horizon Discovery, NC1308404) at a final concentration of 0.2% v/v. The concentration was pre-determined by titrating down siRNA concentrations with an initial qPCR for each siRNA sequence for an approximate 50% reduction in the target RP gene. Transfection Opti-MEM mix was incubated at room temperature for 20 minutes before addition to the cell suspension in Opti-MEM. The cells were then incubated at 37°C with 5% CO₂ for 4 hours. After this incubation period, the transfection medium was removed and replaced with RPMI supplemented with FBS and glucose. The cells were harvested 48 hours post-knockdown procedure with details provided in the next section.

After the knockdown procedure, cells were centrifuged at 600 × g for 5 minutes and washed with chilled PBS (Corning, 21–040-CM). An aliquot was flash frozen in liquid nitrogen for ADP/ATP assay analysis, and the cells were then subjected to a second centrifugation under the same conditions. The supernatant was discarded, and 350 μL of Trizol was added to the cell pellet. The cells were briefly vortexed before undergoing standard phenol-chloroform precipitation. Next, the cell lysate was treated with Turbo DNase (ThermoFisher Scientific, AM2238) following the manufacturer’s guidelines and then subjected to acidic phenol-chloroform extraction (ThermoFisher Scientific, AM9720) for RNA purification. The purified RNA was then converted into cDNA using Superscript III Reverse Transcriptase (ThermoFisher Scientific, 18080–093). Quantitative PCR (qPCR) was performed using the PowerUP SYBR Green Master Mix Solution (ThermoFisher Scientific, A25779) following the manufacturer’s protocol. The qPCR reactions were performed on a Fast 96-well plate (ThermoFisher Scientific, 4346907) to quantify gene expression levels and knockdown efficiency.

Measurement of relative expression was performed using comparative methods through ΔΔCt measurement. Ct (Cycle threshold) values were obtained from qPCR amplification of samples with either their targeted primer or housekeeping primer (GAPDH) in three technical replicates and averaged. The ΔCt value was calculated by subtracting the average Ct value of a targeted primer with the average Ct value of a housekeeping primer. The ΔΔCt was calculated through subtracting the ΔCt of the treatment group with ΔCt of the control group. The relative expression level was computed with the following equation: Expression level = 2^(-ΔΔCt) (Livak and Schmittgen 2001).

siRNA sequences used in this study are provided below and in Table S9.

Following the collection of cells in PBS as described, the ADP/ATP ratio was determined using the ADP/ATP Assay Kit (Sigma-Aldrich, MAK135–1KT). The assay was conducted on a Glomax Luminometer (Promega), following the guidelines provided by the manufacturer.

In the assay for mitochondrial membrane staining, K562 cells were quantitatively stained using 50 nM Mitotracker CMXRos (ThermoFisher Scientific, M7512) alongside 2 μM Hoechst (ThermoFisher Scientific, 62249) for nuclear staining, for a duration of 30 minutes. Post-staining, the cells were washed in PBS and centrifuged at 600 × g for 5 minutes. Subsequently, the cells were resuspended in LiveCell Imaging Solution (ThermoFisher Scientific, A14291DJ). Fluorescent imaging was performed using a 40x (NA 1.3) objective on a Leica SPE microscope with a Leica K5 microscope camera. Images were taken using Leica Application Suite X, and the intensity of MitoTracker staining per area within the cells was quantitatively analyzed using ImageJ software.

Data points or raw data images are all provided in Texas Data Repository, University of Texas at Austin Dataverse Collection under Sarinay Cenik Lab Dataverse. The link is as follows: https://dataverse.tdl.org/dataset.xhtml?persistentId=doi:10.18738/T8/AVZD3L

All high throughput datasets were uploaded to NCBI GEO (GSE280071).