The study harnessed Mendelian randomisation, a statistical framework that has become one of the most powerful tools in modern genetic epidemiology. Rather than measuring correlations directly, which are easily distorted by confounding factors and reverse causation, Mendelian randomisation uses naturally occurring genetic variants as instrumental variables. Because these variants are randomly allocated at conception, in the same way that a clinical trial randomly assigns participants to treatment groups, they are largely shielded from environmental confounders. If a genetic variant that raises circulating levels of a protein also raises the risk of depression, the logic of the method suggests that the protein itself may causally influence depression risk, rather than the other way around.

The protein in question, CD40, sits at a critical junction of the immune system. It is a receptor found on the surface of immune cells, most notably B lymphocytes and antigen-presenting cells, and it interacts with a partner molecule called CD40 ligand expressed on activated T cells. When that interaction occurs, it triggers cascades of intracellular signalling that drive B-cell maturation, antibody class switching, and the production of inflammatory cytokines. CD40 signalling is essential for a properly functioning adaptive immune response, but when the pathway is overactive, it can fuel chronic inflammation, autoimmunity and tissue damage. Aberrant CD40 signalling has previously been implicated in conditions including multiple sclerosis, rheumatoid arthritis, inflammatory bowel disease and atherosclerosis, which is precisely why the pathway has attracted attention as a therapeutic target.

The locus studied by the team sits in a genomically crowded neighbourhood. The CD40 gene, which encodes the receptor protein, lies close to SLC12A5, a gene encoding KCC2, a potassium–chloride cotransporter that is crucial for neuronal function. KCC2 is one of the best-known regulators of neuronal inhibition: it maintains the low intracellular chloride concentration that allows the neurotransmitter GABA to exert its inhibitory effects in the brain. Loss of KCC2 function shifts the neuronal chloride balance, weakens inhibition, and is linked to epilepsy and altered neuronal excitability. The physical proximity of CD40 and SLC12A5 on the genome creates an analytical challenge that the researchers were careful to confront: genetic variants in this region could influence disease risk through either gene, and standard genetic association studies cannot, on their own, distinguish between the two possibilities.

To resolve that ambiguity, the team turned to colocalisation analysis, a computational technique designed to determine whether two genetic associations, for instance an association with circulating CD40 protein levels and an association with depression, are driven by the same causal variant or by different variants that merely happen to lie in the same region. This distinction matters enormously. If the same variant drives both signals, that supports a shared causal mechanism running through a single gene or protein. If two separate variants are responsible, the apparent overlap may be illusory, a phenomenon known as spatial confounding. Colocalisation methods evaluate the probability that a shared variant explains both traits, and the technique has become a standard safeguard against false positives in protein-based drug target studies.

The results of the analysis, as the study’s title indicates, revealed pleiotropic effects of the CD40/SLC12A5 locus extending across three domains at once: the measurable abundance of CD40 protein in the blood, the risk of depression, and the risk of immune disease. Circulating protein levels are increasingly measured at scale in population biobanks through proteomic platforms that quantify thousands of proteins in plasma. These protein quantitative trait loci, or pQTLs, provide the raw material for studies like this one, allowing researchers to ask whether genetically determined differences in protein abundance translate into differences in disease risk. The CD40/SLC12A5 locus is one of the reported signals for circulating CD40, and the new work tested whether that genetic signal carries consequences for psychiatric and immune outcomes.

The connection between immunity and depression is one of the most actively pursued frontiers in psychiatric research. People with chronic inflammatory diseases experience depression at elevated rates, and large epidemiological studies have repeatedly linked high blood levels of inflammatory markers such as C-reactive protein and interleukin-6 with depressive symptoms. Clinical observations add further intrigue: some patients treated with inflammatory cytokines such as interferon-alpha for hepatitis C or cancer develop severe depressive episodes, and a subset of depression patients shows elevated peripheral inflammation. These converging lines of evidence have spawned the field sometimes called immuno-psychiatry, which seeks to identify biologically defined subtypes of depression that may respond to anti-inflammatory treatment. Genetic studies offer a way to test whether the immune–depression link is genuinely causal or merely a by-product of confounding by lifestyle, socioeconomic factors or reverse causation, where illness and distress alter immune function rather than the reverse.

The new findings carry implications in both directions. For immunology, they suggest that variation at this locus influences susceptibility to immune-mediated disease partly through the same mechanisms that affect circulating CD40 abundance, reinforcing CD40 as a credible drug target for conditions in which the pathway is overactive. Therapeutic agents aimed at the CD40 pathway, including monoclonal antibodies designed to block CD40–CD40 ligand interaction, are already in clinical development for autoimmune conditions. For psychiatry, the genetic link to depression raises the possibility that part of depression’s biological architecture runs through immune signalling, and that individuals with certain genetic profiles might represent a subgroup whose depressive illness is inflammation-associated. That, in turn, hints at a future in which depression is stratified genetically and treated with therapies matched to the underlying biology, an approach that remains aspirational but is increasingly grounded in data of this kind.

The study also illustrates the strengths and the necessary cautions of the modern causal inference toolkit. Mendelian randomisation is only as reliable as its instruments, and genetic variants can exert effects through multiple pathways, a problem known as horizontal pleiotropy that can bias results if not properly modelled. Sensitivity analyses, including weighted median and mode estimators and MR-Egger regression, are typically deployed to detect and adjust for such bias. Meanwhile, colocalisation guards against the distinct pitfall of linked but distinct causal variants. By combining both approaches, the authors of the new study were able to make a stronger case than either method alone would permit, while the close linkage of CD40 and SLC12A5 demanded particularly careful dissection. The region is, in effect, a natural experiment in how human geneticists untangle the contributions of neighbouring genes.

What the work does not yet establish is the precise biological pathway by which CD40 signalling or its modulation influences mood and behaviour, and whether the effects observed at the population level in genetic data will translate into meaningful benefit for individual patients. Animal models of CD40 manipulation have largely focused on autoimmune and neuroinflammatory outcomes, and the connection to depressive-like behaviour remains only partially explored. Human intervention studies targeting the CD40 pathway in depression have not been carried out, and any therapeutic ambition remains speculative. Nevertheless, the convergence of genetic evidence linking a single immune locus to depression risk provides exactly the kind of causal support that has been missing from purely correlational studies, and it gives researchers a concrete molecular hypothesis to test in the laboratory and the clinic.

The research, published in Translational Psychiatry in 2026, represents a collaboration involving Laattoe, Hyppönen and Stacey and colleagues, and forms part of a broader movement to mine biobank-scale genetic and proteomic data for clues about disease mechanisms. As datasets grow and proteomic assays expand, studies of this kind are expected to multiply, systematically screening hundreds of circulating proteins for causal links to psychiatric and immune outcomes. For now, the CD40/SLC12A5 locus stands out as a case study in the power of that approach, and a reminder that the boundaries between immunology and psychiatry are far less rigid than the traditional organisation of medicine would suggest. Depression, long understood primarily through the lens of brain neurotransmitters, increasingly appears also to be a condition in which the immune system’s genetic wiring plays a measurable and potentially modifiable role.

The study detected a total of 12,129 independent associations with phenotypic traits, distinguishing 11,243 linked to common variants and unveiling 886 rare-variant associations (RVAs). These findings illuminate how much of the elusive rare-variant heritability can now be directly mapped to single loci, particularly through the lens of large-scale WGS, a technique offering unprecedented resolution across the entire genome compared to previous array-based genotyping or imputation methods.

Intriguingly, rare variants associated with traits were only found in 30 of the 34 studied phenotypes, revealing an inherent challenge: the limited statistical power to detect associations involving variants with very low minor allele frequencies (MAF). Most RVAs (64%) had a MAF exceeding 0.1%, indicating that while WGS exponentially expands variant discovery, the rarer spectrum remains difficult to dissect with current cohort sizes.

One of the study’s critical insights pertains to pleiotropy, the phenomenon where a single genetic variant impacts multiple traits. Approximately 8% of genome-wide significant variants influenced more than one phenotype. Notably, the SLC39A8 missense variant rs13107325, with a MAF of 7.5%, exhibited associations with an impressive 14 different traits. Equally compelling, a rare insertion-deletion variant within the ASGR1 gene intron showed pleiotropic effects across nine distinct traits, underscoring that rare variants can exert broad biological influence.

Quantitatively, after adjusting for statistical biases such as the winner’s curse, the average phenotypic variance explained by RVAs was slightly higher (0.027%) than that explained by common variant associations (0.023%). Yet, when considering the overall heritability, common variants account for approximately 31% of the common-variant heritability portion, whereas rare variants explained about 11% of rare-variant heritability. This remarkable data reflects the complementary roles both variant classes play within the genomic landscape of human traits.

Lipid traits featured prominently in the rare variant analysis, with 18% of RVAs linked to dyslipidemia, triglycerides, LDL, or HDL cholesterol, despite these phenotypes comprising only 12% of the total traits examined. This 1.5-fold enrichment suggests that rare variants impacting lipid metabolism are more likely to have substantial effect sizes, emphasizing the biological significance of rare genetic architecture in cardiovascular-related traits.

Further validation emerged from an independent European ancestry cohort of approximately 67,000 individuals from the Alliance for Genomic Discovery (AGD). Here, RVAs identified in the UK Biobank explained roughly a third of the rare-variant heritability for LDL and HDL, reinforcing the robustness and replicability of findings across populations. Moreover, alkaline phosphatase (ALK) stood out as the sole non-lipid phenotype for which rare variants explained more than one-third of its estimated rare variant heritability, highlighting unique rare variant effects beyond lipid metabolism.

The study also revealed that 41% of RVAs were situated within genomic loci accessible to whole-exome sequencing (WES), predominantly in coding regions. Nonetheless, many of the most impactful RVAs were located outside these WES target zones, underscoring the limitation of exome-focused studies and reinforcing the importance of comprehensive WGS approaches for capturing the full spectrum of functionally significant variants.

A striking example includes the rare indel rs754165241 in ASGR1, associated with a 1.43 standard deviation increase in ALK levels and explaining about 3% of phenotypic variance—the largest variance explained among all associations detected. This variant has been validated in several large population cohorts, further proving the power of WGS to uncover rare variants with outsized biological effects.

Technological insights from the study highlight the immense advantages of WGS over imputation-based approaches for locus detection and fine-mapping. The authors document improved fine-mapping resolution, revealing previously undetectable haplotypes in European ancestry populations that are absent from existing imputation reference panels. This technological advancement promises to sharpen precision medicine strategies by offering a more comprehensive inventory of genetic variation.

Collectively, this expansive GWAS using WGS data exemplifies the evolving landscape of genetic discovery, where integration of rare and common variants brings researchers closer to the elusive goal of fully characterizing heritable contributions to complex traits. The results hold significant implications for genetic epidemiology, risk prediction, and the future design of genomic studies aimed at personalized health interventions.

The study also underscores the persistent gaps in rare variant detection and heritability mapping, advocating for even larger cohorts and diverse ancestries to enhance power and generalizability. Such efforts will be critical to unravel the full genetic underpinnings of human health and disease, potentially revealing novel therapeutic targets.

By demonstrating that significant portions of rare-variant heritability are now accessible via WGS-based GWAS and corroborating these associations across independent datasets, the research sets a new gold standard. This heralds a new era in human genetics research, where the intricate fabric of rare and common variation can be more fully appreciated and leveraged.

Researchers and clinicians alike can anticipate a paradigm shift as these findings illustrate the essential need for whole-genome approaches in capturing the genetic complexity of human traits—pushing beyond the lists of common SNPs and spotlighting the critical impact of rare genetic variation in shaping biology.

Strongly rooted in rigorous data and innovative methodology, this landmark study charts a path toward a comprehensive genetic map of human traits and diseases, bringing the scientific community steps closer to decoding the extensive yet hidden architecture of heritability.

Subject of Research:
Deciphering the contributions of rare and common genetic variants to human phenotypes using extensive whole-genome sequencing in a large population cohort.

Article Title:
Estimation and mapping of the missing heritability of human phenotypes.

Article References:
Wainschtein, P., Zhang, Y., Schwartzentruber, J. et al. Estimation and mapping of the missing heritability of human phenotypes. Nature (2025). https://doi.org/10.1038/s41586-025-09720-6

Image Credits:
AI Generated

DOI: https://doi.org/10.1038/s41586-025-09720-6