Glucose-6-phosphate isomerase, or GPI, sits near the top of the glycolytic pathway, catalyzing the interconversion of glucose-6-phosphate and fructose-6-phosphate. Every cell in the body depends on glycolysis to some degree, but mature red blood cells are uniquely exposed. Once they extrude their nuclei during development, erythrocytes lose the ability to manufacture replacement enzymes, so their entire energy supply hangs on whatever GPI molecules they started with. When the GPI gene is knocked out on both copies of chromosome 19, the result is hereditary nonspherocytic hemolytic anemia, a condition whose severity ranges from mild, compensated hemolysis to life-threatening hydrops fetalis and, in some families, neurological impairment. More than 40 pathogenic variants have been described, yet clinicians still cannot reliably predict how sick a given patient will become from their genotype alone.

That genotype-phenotype discordance is precisely what motivated the new study. Traditional diagnostic workflows lean heavily on identifying coding-sequence mutations and assessing their likely structural damage to the encoded protein. But occasionally, patients with two defective alleles fare better than those with one, or genetically similar patients diverge sharply in clinical course. Several research groups have proposed that layers of regulation beyond the protein sequence, including allele-specific expression, cis-regulatory elements, and epigenetic modifications, might modulate disease severity in red cell enzymopathies. Direct evidence, however, has been scarce, largely because standard short-read sequencing struggles to resolve these features. Short reads often cannot phase variants across long stretches of DNA, and they cannot simultaneously capture the native methylation marks that decorate the genome.

The Polish-led team behind the new report turned to PacBio HiFi long-read sequencing to break through that barrier. Their patient was a 47-year-old woman with transfusion-dependent hemolytic anemia. Short-read Illumina whole-genome sequencing had already identified two rare heterozygous missense variants in the GPI gene: c.572A>G, which substitutes arginine for histidine at position 191, and c.1414C>T, which substitutes cysteine for arginine at position 472. Both variants are vanishingly rare in the European population, with frequencies below 0.0001 in the gnomAD database. Critically, short reads alone could not determine whether the two mutations sat on the same chromosome or on opposite ones, a distinction that matters enormously for interpreting residual enzyme function.

The long-read data settled the question directly. By phasing the variants across their shared haplotypes, the researchers showed that the mutations sit in trans, one on each parental copy of the gene. Haplotype 1 carries p.His191Arg, while Haplotype 2 carries p.Arg472Cys, and no wild-type allele remains. The team also catalogued 15 non-coding variants within the GPI locus, though deep-learning splice predictions from SpliceAI found no disruption of canonical splicing motifs above the standard reporting threshold. In parallel, full-length isoform sequencing of blood RNA, performed in two independent technical replicates, assembled the GPI transcriptome in unprecedented detail and revealed two previously undescribed minor isoforms alongside the canonical transcript, a descriptive bonus that the authors deliberately kept out of their mechanistic model.

The most provocative observation came from layering transcript counts and methylation calls onto the phased haplotypes. Quantifying allele-resolved reads at the discriminating c.1414 position, the researchers found that Haplotype 2, the one bearing p.Arg472Cys, was modestly over-represented in the transcript pool in both replicates: roughly 62 percent in the first and 55 percent in the second, pooling to about 60 percent overall. The direction was consistent, but the effect was statistically non-significant, and the authors are explicit that this constitutes a trend toward allele-specific expression, not an established one. Isoform-level allelic quantification was underpowered, with too few full-length reads to draw any conclusion at that resolution.

Methylation added a second intriguing thread. Using the polymerase kinetics of PacBio HiFi sequencing, the team obtained native 5-methylcytosine calls across the GPI locus without the bisulfite conversion that conventional methylation mapping requires. The wild-type cytosine at c.1414, preserved on Haplotype 1, sits within a CpG dinucleotide, the genomic context most prone to methylation in human cells. In the small number of informative reads spanning that position, the site was called as methylated on every read carrying the C allele. On Haplotype 2, the C-to-T transition physically destroys the CpG, so the methylated state there is abolished by definition. The co-occurrence of CpG loss on the same allele that shows a directional excess of transcripts is exactly the kind of correlation that fuels a hypothesis, and the authors are careful to label it as precisely that, noting that only a handful of reads support the methylation call.

What makes the model worth testing is what is already known about the protein on Haplotype 2. Prior biochemical work has established p.Arg472Cys as a thermally unstable variant that degrades rapidly, a property consistent with its comparatively low AlphaMissense pathogenicity score of 0.257, compared with 0.914 for p.His191Arg, since structure-based predictors do not necessarily capture instability effects. If the same allele that produces an unstable protein also carries a CpG whose loss slightly lifts transcriptional repression, then the cell might partially compensate by simply making more of a fragile enzyme from that allele. That compensation would depend on a nucleus and would be lost when red cells enucleate and stop transcribing, potentially explaining the age-dependent decline in GPI activity that earlier kinetic studies of hemolytic anemia have proposed. In nucleated tissues, by contrast, continuous transcription from the trans allele might explain why this patient, carrying a p.His191Arg variant previously linked to neurological features, presented with an exclusively hematological phenotype.

The authors are refreshingly transparent about the limits of a single-patient, correlative dataset. They did not measure GPI protein abundance, enzymatic activity, or stability in this patient, and the allelic imbalance never reached statistical significance. To convert their candidate dual-effect variant into an established mechanism, they lay out a demanding experimental agenda: allele-specific transcript quantification at far greater depth with formal correction for mapping bias, statistically powered haplotype-resolved methylation comparison across cell fractions, a direct test of whether the disrupted CpG carries regulatory activity through reporter assays or targeted demethylation, and direct measurement of GPI protein levels and enzymatic activity comparing reticulocyte-enriched and mature erythrocyte fractions. Only such evidence could show that transcriptional output, epigenetic state, and protein stability genuinely interact to shape disease.

Even as a hypothesis, the case marks a turning point in how rare enzymopathies can be examined. One sequencing workflow simultaneously phased compound heterozygous variants, mapped native methylation without chemical conversion, resolved novel transcript isoforms, and generated a specific, testable account of why one patient’s disease unfolded as it did. Short-read exomes and genomes, the workhorses of modern diagnostics, simply cannot deliver that integrated view of a locus. As long-read multi-omics becomes cheaper and more routine, the field will be able to ask systematically whether dual-effect variants, single changes that sabotage both a protein and its own regulatory context, recur across other patients and other genes. If they do, the humble one-letter mutation may turn out to be a far more layered act of molecular sabotage than anyone suspected.

Subject of Research: Integrative long-read multi-omics characterization of a compound heterozygous GPI deficiency patient revealing a candidate dual-effect coding and cis-regulatory variant

Article Title: Integrative Long‐Read Multi‐Omics of a Patient With GPI Deficiency: A Molecular Case Study of a Candidate Dual‐Effect GPI Variant

Article References: Stolarek, I., Delimata‐Raczek, J., Koralewska, N., Sikora, K., Rakoczy, M., Marcinkowska‐Swojak, M., Handschuh, L., Czyż, J., & Figlerowicz, M. (2026). Integrative Long‐Read Multi‐Omics of a Patient With GPI Deficiency: A Molecular Case Study of a Candidate Dual‐Effect GPI Variant. Journal of Cellular and Molecular Medicine, 30(17), Article e71338. https://doi.org/10.1111/jcmm.71338

Image Credits: AI Generated

DOI: 10.1111/jcmm.71338

Keywords: GPI deficiency, long-read sequencing, PacBio HiFi, hemolytic anemia, allele-specific expression, DNA methylation, glycolysis, red blood cells, compound heterozygosity, multi-omics, Iso-Seq, rare disease

The technical achievement at the heart of the study is considerable. Building a truly complete genome has long been one of genomics’ most stubborn challenges, because repetitive DNA sequences near the centers and ends of chromosomes resist standard sequencing methods. The research team overcame this by combining PacBio HiFi long-read sequencing, which produces highly accurate reads spanning tens of thousands of DNA letters, with Hi-C technology, which captures the physical contacts between distant stretches of chromosomes and allows the sequences to be anchored into their correct chromosomal positions. The final assembly spans 2.42 gigabases, with a scaffold N50 of 72.35 megabases and a contig N50 of 66.56 megabases, meaning that half of the assembled genome resides in stretches longer than those figures. The guanine-cytosine content of the assembly sits at 42.36 percent. In practical terms, the contiguity of this assembly approaches the ideal of a single gap-free sequence per chromosome, giving researchers an unprecedented view of the bear’s genome architecture.

With that high-resolution reference in hand, the team turned to comparative genomics, aligning the Tibetan brown bear’s chromosomes against those of its closest living relative, the polar bear. The analysis revealed high chromosomal synteny between the two species, meaning that large blocks of genes retain the same order and orientation across both genomes despite millions of years of separate evolution. This structural conservation is scientifically valuable for two reasons. It confirms that the assembly is accurate, since genuine chromosome-scale structure should be preserved between closely related ursids, and it provides a stable framework for pinpointing the regions that have instead diverged, which are precisely the locations where adaptation to very different environments, Arctic sea ice versus high-altitude steppe, is most likely to leave its mark.

The genomic record also preserves a vivid portrait of the species’ demographic turbulence. By reconstructing changes in effective population size through time, the researchers found that the Tibetan brown bear expanded during the early Pleistocene, a period of dynamic climate and habitat change. That expansion, however, was followed by two severe population bottlenecks. The first occurred approximately two million years ago, coinciding with the pre-Poyang Glaciation, and the second struck around one hundred thousand years ago during the Last Glacial Period. Together, these crashes reduced the bear’s cumulative effective population size by nearly eighty percent. For a species that today persists only on the Qinghai-Xizang Plateau, this history of repeated near-collapses carries a sobering message about the genetic fragility that may still lurk beneath its wild, unbroken landscape.

Yet survival through those bottlenecks suggests the species carries more than vulnerability. Enduring glacial cycles on a plateau whose average elevation exceeds 4,000 meters demands a physiology unlike that of lowland relatives, and the most striking discovery of the study concerns exactly that. When the researchers examined candidate hibernation-related pathways, they found that the Tibetan brown bear displays a distinctly carbohydrate-dominated metabolic signature. This stands in sharp contrast to the hibernation strategy documented in North American brown bears, which rely primarily on lipid, or fat-based, metabolism to fuel their long winter dormancy. In other words, two populations of the same widespread species appear to have evolved fundamentally different biochemical approaches to the same survival problem.

The implications of that finding extend well beyond bears. Fat is the standard fuel of hibernation across most studied mammals, because it stores more energy per gram and spares blood sugar during months of fasting. A sugar-first strategy on the Tibetan Plateau hints at selective pressures unique to that environment, where the short growing season, the composition of available foods, and the metabolic demands of life at low oxygen levels may have favored a different balance of carbohydrate and lipid pathways. The authors emphasize that this distinct metabolic adaptation highlights the bear’s unique mechanisms for surviving the extreme plateau environment, and it gives physiologists a natural experiment in alternative hibernation biochemistry encoded within a single species.

For conservation biologists, the new genome arrives at a critical moment. The Tibetan brown bear sits at the top of the plateau’s food web, playing a crucial ecological role, yet high-quality genomic resources for the species have been scarce until now. A chromosome-level reference genome transforms what conservation managers can do: it enables accurate estimates of genetic diversity and inbreeding, allows the tracking of gene flow between populations, and provides the resolution needed to identify locally adapted lineages that deserve special protection. Given the roughly eighty percent historical reduction in effective population size documented in the study, such tools are not a luxury. They are the baseline data upon which any serious plan for the species’ long-term management on the plateau must rest.

The study also strengthens the broader evolutionary picture of the bear family. Ursids have long fascinated evolutionary biologists because the family contains species with radically different ecologies, from the omnivorous brown bear to the hypercarnivorous, ice-bound polar bear. A near telomere-to-telomere assembly for the Tibetan brown bear adds a critical high-quality data point for reconstructing the phylogenomic relationships within Ursus and for dating the divergences that produced today’s brown bear lineages. Because structural variants, gene duplications, and regulatory regions can now be examined in their full chromosomal context rather than through fragmented assemblies, questions about how bears colonized and adapted to some of Earth’s harshest habitats can be addressed with far greater precision than before.

The research was carried out by Muran Zhao, Anmin Wang, Hai Liu, Chenxing Yu, Yanlin Liu, Nan Sun, and Guogang Li of Qinghai Normal University, with fieldwork, sample collection, and laboratory analysis conducted under permits granted by the university. The work was supported by the National Natural Science Foundation of China. The team notes that the new assembly is intended as a resource for the wider community, facilitating future studies of ursid evolutionary history and supporting conservation and management of the plateau’s wildlife. Because the genome is derived from a female bear, it also enables improved analysis of the sex chromosomes, an area where earlier fragmented assemblies often fell short.

What began as a technical sequencing project has ended with a discovery that may echo well beyond the Qinghai-Xizang Plateau. If a close relative of the brown bears studied across North America and Eurasia can hibernate on a carbohydrate-dominated metabolic program, then hibernation is not a single fixed solution that evolution produced once and reused everywhere. It is a flexible toolkit, and different populations have assembled it from different parts. As climate change alters the length and severity of winters across the world’s mountains, understanding that flexibility, and the genes that underpin it, could prove essential not only for protecting the Tibetan brown bear but for predicting which hibernating species can adjust and which cannot. The near-complete genome of this high-altitude survivor is now available as the roadmap for answering those questions.

Subject of Research: Chromosome-level genome assembly and hibernation-related evolution of the Tibetan brown bear (Ursus arctos pruinosus)

Article Title: Near telomere-to-telomere genome reveals the phylogenomics and hibernation-related evolution in Tibetan brown bear (Ursus arctos pruinosus)

Article References: Zhao, M., Wang, A., Liu, H., Yu, C., Liu, Y., Sun, N., & Li, G. (2026). Near telomere-to-telomere genome reveals the phylogenomics and hibernation-related evolution in Tibetan brown bear (Ursus arctos pruinosus). BMC Genomics. https://doi.org/10.1186/s12864-026-13324-3

Image Credits: AI Generated

DOI: 10.1186/s12864-026-13324-3

Keywords: Tibetan brown bear, telomere-to-telomere genome, genomics, hibernation, Qinghai-Xizang Plateau, population bottlenecks, PacBio HiFi, Hi-C sequencing, phylogenomics, conservation genomics, carbohydrate metabolism, Ursidae evolution