Beyond Stability: Rethinking Housekeeping Genes as Contributors to Rheumatoid Arthritis Pathogenesis

Abstract

The concept of housekeeping genes has evolved considerably since the term first appeared in the 1970s, when it was used primarily to describe genes encoding transfer RNA and ribosomal RNA. Over the subsequent five decades, housekeeping genes became defined as constitutive genes required for the maintenance of basic cellular functions, expressed in all cells of an organism under normal and pathophysiological conditions [1-3]. Their products participate in fundamental processes such as protein synthesis, energy metabolism, and cytoskeletal organization, making them essential for cell survival regardless of tissue type, developmental stage, or cell cycle state [4]. This presumed universality and stability led to their widespread adoption as reference controls for normalizing gene expression data in quantitative PCR and transcriptomic studies. It holds an implicit assumption that these genes resist pathological perturbations impacting tissue-specific genes [5]. In routine experimental practice, a limited set of genes, such as GAPDH, ACTB, and 18S rRNA, has been widely adopted as internal controls. However, these genes were chosen largely for experimental convenience rather than for evidence of constitutive expression, and studies have shown that they can vary considerably across tissues [6-8]. With the advent of RNA-seq, systematic identification of housekeeping genes has become possible with unprecedented accuracy. In a landmark study using data from 16 normal human tissues, Eisenberg and Levanon (2013) applied three stringent criteria: expression in all tissues, low variance across tissues, and no exceptional expression in any single tissue to define 3804 high-confidence human housekeeping genes [9] (Table S1). This systematically defined set, rather than the limited set of conventional internal controls, serves as the foundation for our analysis.

However, this long-held view is increasingly being questioned. Recent advances in high-throughput sequencing have revealed that the expression patterns of housekeeping genes are more complex than previously appreciated. Studies in cancer biology have documented aberrant expression of several housekeeping genes in malignant tissues, suggesting that their dysregulation may contribute to disease progression [10-12]. In the context of autoimmune diseases, emerging evidence points to the involvement of housekeeping genes in inflammatory pathways [13], yet systematic investigations have remained scarce. Rheumatoid arthritis (RA), with its complex molecular architecture characterized by the collective dysregulation of hundreds of genes [14], provides an ideal context to examine whether these fundamental genes might play previously unrecognized roles in disease pathogenesis. The global burden of rheumatic diseases, exacerbated by the COVID-19 pandemic and limited access to specialist care in low-income countries, further underscores the need for deeper molecular insights [15].

To address this question systematically, we analyzed 15 independent transcriptome datasets from the Rheumatoid Arthritis Bioinformatics Center (RABC) [16], encompassing baseline RA patients and healthy controls (Figure S1, Tables S2 and S3, Supporting Information). These datasets were carefully curated to exclude treatment effects and ensure tissue comparability, resulting in a robust collection of gene expression profiles spanning diverse patient populations and experimental platforms.

Differential expression analysis between RA patients and healthy controls revealed a striking and unexpected pattern. When we applied cumulative hypergeometric tests to assess whether housekeeping genes were overrepresented among differentially expressed genes (DEGs), significant enrichment was observed in 11 of the 15 datasets (Table S4). This enrichment was not marginal: the overlap between DEGs and housekeeping genes consistently exceeded what would be expected by chance alone (Table 1). To quantify this effect across studies, we calculated relative enrichment factors for each dataset. Meta-analysis of these 15 studies using a random effects model yielded a pooled logarithmic relative enrichment estimate of 0.48 (95% confidence interval: 0.17–0.80), which deviated significantly from null model expectations (p = 0.0012) (Figures 1 and S2–S4). The consistency of this signal across independent datasets, despite differences in patient characteristics, sample sizes, and microarray platforms, argues strongly against the possibility of technical artifact [17].

The designation “housekeeping gene” carries with it an implicit expectation of expression stability, yet our analysis challenges this very foundation. When we compared expression variability between RA patients and healthy controls, a clear pattern emerged. In two-thirds of the datasets examined, housekeeping genes exhibited significantly greater variance, standard deviation, and coefficient of variation in patient samples compared with controls (Figures S5–S7). This increased fluctuation was not observed for non-housekeeping genes to the same degree, suggesting that the disease state imposes a generalized transcriptional instability that disproportionately affects genes traditionally considered constitutive.

This observation raises fundamental questions about the nature of housekeeping gene regulation in disease. The increased variability may reflect disruption of the epigenetic mechanisms that normally maintain stable expression. Indeed, when we examined matched DNA methylation and transcriptome data from RA patients, housekeeping genes demonstrated significantly stronger correlations between promoter methylation status and expression levels compared with non-housekeeping genes (Table S5, Figures S8–S17). This association implicates epigenetic dysregulation as a potential mechanism underlying housekeeping gene perturbations in RA, consistent with the well-documented role of DNA methylation changes in autoimmune pathogenesis [18].

Our findings in RA align with emerging evidence from other autoimmune and inflammatory conditions. Importantly, our research team has been at the forefront of systematically investigating housekeeping gene dysregulation in autoimmune diseases. In a recent study published by our group, we investigated housekeeping genes in psoriasis and identified 34 housekeeping genes associated with disease pathogenesis through integrated multi-cohort and single-cell transcriptomic analysis [12]. Using weighted gene co-expression network analysis combined with differential expression methods, we demonstrated that the co-expression relationships between six genes (APOL2, DCUN1D3, UBE2F, HIGD1A, PPIF, and STAT3) and known psoriasis-related genes differed significantly between psoriasis patients and healthy individuals. Single-cell transcriptomic profiling further revealed that these housekeeping genes were differentially expressed primarily in basal, spinous, supraspinous, and proliferating keratinocytes. Ligand-receptor interaction analysis demonstrated significant activation of the interleukin-17, interleukin-6, and midkine pathways within keratinocyte subpopulations, which led to the upregulation of STAT3, EIF5A, and RAN, thereby promoting keratinocyte hyperproliferation and enhancing immune reactivity [19].

Our team has systematically revealed for the first time that housekeeping genes, long considered mere bystanders in disease processes, exhibit consistent and reproducible dysregulation across multiple autoimmune conditions. The convergence of evidence from RA and psoriasis suggests a broader paradigm that housekeeping genes may function as dual regulators in disease contexts, participating in pathogenic mechanisms while retaining their essential cellular functions. The interleukin-17 pathway, central to both psoriasis and RA pathogenesis, appears to intersect with housekeeping gene regulation through mechanisms involving metabolic reprogramming and epigenetic modification [20]. The mechanistic target of rapamycin pathway, which integrates signals from growth factors, nutrients, and immune stimuli, has been shown to regulate both housekeeping and inflammatory gene expression, providing a potential link between basic cellular metabolism and immune activation in RA [21]. These mechanistic insights are complemented by recent advances in understanding trained immunity, where metabolic and epigenetic reprogramming of innate immune cells shapes chronic inflammatory responses [22, 23].

The evidence presented here challenges the conventional partitioning of genes into “housekeeping” and “tissue-specific” categories in disease research. Our finding that housekeeping genes are recurrently and significantly enriched among RA-associated DEGs suggests that these fundamental cellular components may represent previously overlooked contributors to autoimmune pathogenesis. The increased expression variability observed in patient samples and the strong correlation with DNA methylation changes point toward epigenetic dysregulation as a plausible mechanism.

To explore whether the observed expression variations are epigenetically or genetically driven, we performed mQTL-SMR and eQTL-SMR analyses. These identified hundreds of housekeeping genes whose promoter methylation or expression showed causal associations with RA risk, with substantial overlap between the two layers (Tables S6 and S7). Mediation analysis on genes sharing the same causal variant further supported direct epigenetic–transcriptional cascades for RNPEP and PET100 (Table S8). For RNPEP, methylation appears to promote expression, which in turn increases RA risk, with approximately 77.5% of the methylation effect mediated through expression. This suggests that epigenetic activation of RNPEP encoding aminopeptidase B, a zinc-dependent metallopeptidase involved in peptide processing [24], may represent a novel RA risk mechanism. Conversely, PET100 showed a protective mediation pattern where methylation suppresses its expression, and lower expression is associated with reduced RA risk, with a mediation proportion of about 50.5%. Given PET100's role in mitochondrial respiratory chain assembly [25, 26], this finding points to a potential link between mitochondrial dysfunction and RA [27], with epigenetic silencing of PET100 possibly acting as a compensatory protective mechanism. Collectively, these findings indicate that housekeeping gene dysregulation in RA involves both genetic and epigenetic influences, with DNA methylation acting as a mediator in the pathway from genetic variation to altered expression and disease risk.

To explore the clinical relevance of housekeeping gene expression fluctuations in RA, we examined whether their expression tracks with disease activity in a large patient cohort. Using four validated activity measures and well-established clinical thresholds, nearly one-third of housekeeping genes showed expression patterns that consistently paralleled disease activity (Tables S9 and S10). These findings suggest that housekeeping gene fluctuations are not random but reflect the clinical state of RA patients. A smaller subset of housekeeping genes was also observed to distinguish seropositive from seronegative RA (Tables S11 and S12), raising the possibility that certain genes within this conventional class may contribute to subtype-specific pathways. While these observations are preliminary and require further validation, they underscore the potential clinical relevance of housekeeping gene dysregulation in RA and point toward avenues for future biomarker and stratification research.

From a practical perspective, these observations have immediate implications for experimental design. The use of housekeeping genes as normalization controls in RA transcriptomic studies may require careful reevaluation, as their expression stability cannot be assumed a priori in diseased tissues. More fundamentally, our findings invite a conceptual shift in how we think about gene function in disease. Rather than viewing housekeeping genes as passive bystanders, we might consider them as integral components of disease-associated transcriptional networks, whose perturbations could propagate effects throughout the system [28].

Future research directions about housekeeping genes are numerous. Functional studies examining the consequences of housekeeping gene knockdown or overexpression in relevant cell types could establish causal relationships with inflammatory phenotypes. Longitudinal studies tracking housekeeping gene expression patterns during disease progression and in response to therapy might reveal whether these changes are reversible and whether they correlate with clinical outcomes. The development of computational approaches that account for housekeeping gene variability could improve the accuracy of differential expression analyses in RA and other complex diseases. Furthermore, integrating housekeeping gene signatures with existing clinical and molecular data may enhance risk stratification and guide personalized treatment decisions, particularly in resource-limited settings where comprehensive immunomonitoring is not feasible.

The unexpected involvement of housekeeping genes in RA pathogenesis, together with our team's pioneering work in psoriasis, invites us to look beyond the usual suspects in autoimmune research. Housekeeping genes, long relegated to methodological footnotes, may yet prove to be important characters in the story of disease. The recognition that housekeeping genes can actively participate in autoimmune pathogenesis opens new avenues for understanding disease mechanisms and may ultimately lead to novel therapeutic strategies targeting these fundamental cellular components. As we refine our understanding of the molecular architecture of rheumatic diseases, it is time to give housekeeping genes the attention they deserve.

Wenhua Lv and Hao Tang contributed equally as first authors, responsible for study conceptualization, data analysis, and manuscript drafting. Ruijie Zhang, Mingming Zhang, and Yongshuai Jiang contributed equally as corresponding authors, providing oversight of the research, critical revision of the manuscript, and final approval of the submitted version.

This work was supported by the National Natural Science Foundation of China [Grants 31970651, 92046018]; the Excellent Youth Support plan of Education Department of Heilongjiang Province [Grant YQJH2023036]; Marshal Initiative Funding [Grant HMUMIF-22010]; Mathematical Tianyuan Fund of the National Natural Science Foundation of China [Grant 12026414]; XingLian Outstanding Talent Support Program 2024 and Provincial Universities in Heilongjiang Province (2024 Wenhua Lv).

The authors declare no conflicts of interest.

The data that support the findings of this study are available from public domain resources. Additional detailed results and information can be found in Supporting Information methods, figures, and tables.