Changes in immune-cell composition identified in human liver tissue

In the livers of patients with metabolic dysfunction-associated steatohepatitis (MASH), resident immune cells known as Kupffer cells decreased, while GPNMB-positive macrophages involved in lipid metabolism and inflammatory responses increased. Published in Nature Genetics, this study jointly analyzed gene expression, proteins, and cell locations in human liver tissue and found that this population exhibits different features depending on location and disease stage.

Metabolic dysfunction-associated steatotic liver disease (MASLD) encompasses simple steatosis through MASH, which involves inflammation and hepatocyte injury. Macrophages clear lipids and cellular debris while also participating in inflammatory responses, but conventional broad cell classifications could not readily explain the diversity observed in patient tissue. Applying observations from animal models directly to human disease has also remained challenging.

The researchers applied single-nucleus RNA sequencing (snRNA-seq) to samples from 18 individuals: 4 non-obese and 5 obese individuals with normal liver histology, 4 with simple steatosis, and 5 with MASH. They analyzed 176,150 high-quality nuclei and corroborated the findings using spatial transcriptomics and multiplex protein staining in separate tissue samples. The study compared samples from different disease stages; it did not longitudinally follow the same patients as they progressed to end-stage disease.

Macrophage features differ between portal areas and liver parenchyma

MASLD samples showed a decrease in Kupffer cells and an increase in metabolically active macrophages (MetMacs). In the single-nucleus analysis, GPNMB-positive cells accounted for 5.7% of all myeloid cells in the combined non-MASH group, comprising normal and simple-steatosis samples, compared with 12% in the MASH group. The GPNMB-positive population also included subgroups with high LPL or HS3ST2 expression, making it difficult to regard these cells as a population with a single function.

GeoMx spatial transcriptomics on biopsies from 8 patients with MASH and stage F3 fibrosis compared 77 regions that passed quality control to examine macrophage features by location. Separate CosMx spatial molecular imaging and MILAN multiplex protein staining detected GPNMB-positive macrophages both in portal areas, which contain blood vessels and bile ducts, and in the liver parenchyma. In particular, CosMx analysis showed high expression of the lipid-metabolism-related genes LPL and FABP5 in GPNMB-positive cells in areas containing lipogranulomas and cellular debris. The findings suggest a population whose metabolic and immune features vary by tissue location, rather than one confined to portal areas.

Functional experiments revealed both lipid clearance and inflammatory responses. Treating human liver slices with lipids for 24 hours led to aggregation of GPNMB-positive cells and early lipogranuloma formation. In a separate experiment using macrophages derived from the human THP-1 cell line, interleukin 32 (IL32) treatment increased lipid-particle uptake and intracellular endolysosomal degradation activity. Under palmitic-acid stimulation, IL32 increased expression of the inflammation-related gene IL1B and decreased expression of the anti-inflammatory gene IL10. These findings provide functional support for the interaction between IL32-expressing hepatocytes and macrophages observed in tissue, but do not demonstrate that this interaction directly causes fibrosis in patients.

Candidate blood markers and validation needs

The researchers also analyzed liver transcriptomes from 206 patients with MASLD and 10 controls, as well as serum proteomes from 247 patients with biopsy-confirmed MASLD. Several macrophage markers were associated with patient classification by disease activity and stage, with significant differences particularly evident for soluble CD163, GPNMB, and MSR1 in serum. An analysis of 53,030 UK Biobank participants also identified associations between blood GPNMB and metabolic traits, including diabetes.

These findings suggest candidates for noninvasive assessment, but establish neither their performance as a substitute for biopsy nor therapeutic efficacy. Some omics analyses had small sample sizes, and limitations include RNA detection dropout in single-nucleus analysis and possible signal contamination from neighboring cells in spatial analyses. The histologically normal control tissues were also obtained from patients, which remains a limitation. Follow-up studies should distinguish macrophage lipid-clearance functions from inflammatory responses and validate the clinical utility of candidate markers in larger, independent cohorts.

Sources: Nature Genetics paper, Open full text at Europe PMC.