Yinshui Chang

Cancer cells show highly heterogeneous and plastic states in glioblastoma, an incurable brain tumour ^1–5 . However, how these malignant states arise and whether they follow defined cellular trajectories across tumours is poorly understood. Here we generated a deep single-cell and spatial multi-omic atlas of human glioblastoma that pairs transcriptomic, epigenomic and genomic profiling of 12 tumours across many regions. We identify that glioblastoma heterogeneity is driven by spatially patterned transitions of cancer cells from developmental-like states towards those defined by a glial injury response and hypoxia. This cellular trajectory regionalizes tumours into distinct tissue microenvironments and manifests in a molecularly conserved manner across tumours as well as genetically distinct tumour subclones. Consistently, using a new deep-learning framework to map cancer cell states jointly with clones in situ, we show that tumour subclones are finely spatially intermixed through glioblastoma tissue microenvironments. We validate this cellular trajectory by using somatic mutation-based hierarchies in vivo and an experimental patient-derived organoid model ex vivo. Finally, we show that this cancer cell trajectory is intimately linked to myeloid heterogeneity and unfolds across regionalized myeloid signalling environments. Our findings define a stereotyped trajectory of cancer cells in glioblastoma and unify glioblastoma tumour heterogeneity into a tractable cellular and tissue framework.

A bstract: Molecular assays such as spatial transcriptomics facilitate fine-grained identification of cell types for analyzing tumor microenvironments. However, they are expensive to acquire in routine clinical practice compared to hematoxylin and eosin (H&E) images. With molecular guidance, such images can distinguish cell lineages, yet often fail to resolve closely related subtypes. To overcome this, we introduce CellBridge , a multi-modal contrastive learning framework pretrained on 16M cells with paired H&E and spatial profiles, for transferring molecular information into morphological representations. CellBridge further incorporates a Cell-Ontology-aware conformal prediction module that assigns labels at different levels of specificity: leaf-level cell types when morphology alone provides sufficient evidence, and broader ancestor categories when distinguishing among more specific cell types would require molecular information. Across in-domain and out-of-domain benchmarks, CellBridge outperforms leading pathology foundation models in macro F1 by up to 16 percentage points, with the largest gains on the most fine-grained label spaces. It consistently achieves smaller prediction sets, while reducing class-wise coverage gaps. We envision CellBridge as a step toward extracting cellular information that currently requires costly molecular assays from routinely acquired pathology images, while explicitly identifying where such assays remain necessary. Our code is available at https://github.com/Lotfollahilab/cellbridge .

Conventional dendritic cells (cDC) play key roles in immune induction, but what drives their heterogeneity and functional specialization is still ill-defined. Here we show that cDC-specific deletion of the transcriptional repressor Bcl6 in mice alters the phenotype and transcriptome of cDC1 and cDC2, while their lineage identity is preserved. Bcl6-deficient cDC1 are diminished in the periphery but maintain their ability to cross-present antigen to CD8⁺ T cells, confirming general maintenance of this subset. Surprisingly, the absence of Bcl6 in cDC causes a complete loss of Notch2-dependent cDC2 in the spleen and intestinal lamina propria. DC-targeted Bcl6-deficient mice induced fewer T follicular helper cells despite a profound impact on T follicular regulatory cells in response to immunization and mounted diminished Th17 immunity to Citrobacter rodentium in the colon. Our findings establish Bcl6 as an essential transcription factor for subsets of cDC and add to our understanding of the transcriptional landscape underlying cDC heterogeneity.

T follicular helper (T _FH ) cells are essential for effective antibody responses, but deciphering the intrinsic wiring of mouse T _FH cells has long been hampered by the lack of a reliable protocol for their generation in vitro. We report that transforming growth factor–β (TGF-β) induces robust expression of T _FH hallmark molecules CXCR5 and Bcl6 in activated mouse CD4 ⁺ T cells in vitro. TGF-β–induced mouse CXCR5 ⁺ T _FH cells are phenotypically, transcriptionally, and functionally similar to in vivo–generated T _FH cells and provide critical help to B cells. The study further reveals that TGF-β–induced CXCR5 expression is independent of Bcl6 but requires the transcription factor c-Maf. Classical TGF-β–containing T helper 17 (T _H 17)–inducing conditions also yield separate CXCR5 ⁺ and IL-17A–producing cells, highlighting shared and distinct cell fate trajectories of T _FH and T _H 17 cells. We demonstrate that excess IL-2 in high-density T cell cultures interferes with the TGF-β–induced T _FH cell program, that T _FH and T _H 17 cells share a common developmental stage, and that c-Maf acts as a switch factor for T _FH versus T _H 17 cell fates in TGF-β–rich environments in vitro and in vivo.

Constitutive T cell-intrinsic miRNA expression is required for the differentiation of naive CD4(+) T cells into Tfh cells, thus making it difficult to study the role of miRNAs in the maintenance of already established Tfh cells and ongoing germinal center (GC) responses. To overcome this problem, we here used temporally controlled ablation of mature miRNAs specifically in CD4(+) T cells during acute LCMV infection in mice. T cell-intrinsic miRNA expression was not only critical at early stages of Tfh cell differentiation, but also important for the maintenance of already established Tfh cells. In addition, CD4(+) T cell-specific ablation of miRNAs resulted in impaired GC B cell responses. Notably, miRNA deficiency also compromised the antigen-specific CD4(+) T cell compartment, Th1 cells, Treg cells, and Tfr cells. In conclusion, our results highlight miRNAs as important regulators of Tfh cells, thus providing novel insights into the molecular events that govern T cell-B cell interactions and Th cell identity.

Fingolimod is an effective treatment for relapsing-remitting multiple sclerosis. It is well established that fingolimod, a modulator of the sphingosine-1-phosphate pathway, restrains the egress of CCR7⁺ lymphocytes from lymphatic tissues into the blood, thus resulting in reduced lymphocyte counts in peripheral blood. CXCR5⁺ T follicular helper (Tfh) cells provide help to B cells, are essential for the generation of potent Ab responses, and have been shown to be critically involved in the pathogenesis of several autoimmune diseases. Besides lymphoid tissue-resident Tfh cells, CXCR5⁺ circulating Tfh (cTfh) cells have been described in the blood, their numbers correlating with the magnitude of Tfh cells in lymphoid tissues. Although the effect of fingolimod on circulating lymphocyte subsets has been established, its effect on cTfh cells remains poorly understood. In this study, we found that although fingolimod strongly and disproportionally reduced cTfh cell frequencies, frequencies of activated cTfh cells were increased, and the composition of the cTfh cell pool was skewed toward a cTfh1 cell phenotype. The circulating T follicular regulatory cell subset and CXCR5⁺ CD8⁺ T cell frequencies were also strongly and disproportionally decreased after fingolimod treatment. In contrast, relative frequencies of CXCR5^– memory Th cells as well as regulatory T and B cells were increased. In summary, these data provide new insights into fingolimod-induced compositional changes of lymphocyte populations in the blood, in particular cTfh cells, and thus contribute to a better understanding of the mechanism of action of fingolimod in multiple sclerosis patients. Key Points Circulating Tfh cell frequencies are reduced upon fingolimod treatment. Fingolimod reduces circulating Tfr cell and CXCR5⁺ CD8⁺ T cell frequencies. Blood CXCR5^– Tmem, Treg, and Breg cell frequencies are elevated by fingolimod.

Mycobacteria have been implicated in the pathogenesis of autoimmunity. To determine the potential effect of mycobacterial antigens on peripheral blood mononuclear cells (PBMC), we analyzed PBMC incubated with the acetone-precipitable fraction of Mycobacterium tuberculosis (APMT) for changes in cellular protein expression. Two-dimensional gel analysis showed induction of a 36-kD polypeptide identified as proliferating cell nuclear antigen (PCNA), a known autoantigen, after incubation with AP-MT. PCNA plays a role in cell proliferation and is expressed as a late growth regulated factor. However, its synthesis in response to AP-MT was induced as an early event. The early induction of PCNA was regulated at a posttranscriptional level and was restricted to T cells. Treatment of PBMC with known T cell mitogens, namely PHA, anti-CD3 antibodies, and staphylococcal superantigens failed to induce an early PCNA increase. The distinct characteristics of the AP-MT effect on PCNA expression suggest a separate mechanism of induction in response to AP-MT, compared with the late increase observed in response to mitogens. The induction of PCNA in response to mycobacterial antigens may represent a pathogenically relevant mechanism in autoimmunity.