Search bioRxiv⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/273763

Perivascular signals alter global genomic profile of glioblastoma and response to temozolomide in a gelatin hydrogel

Harley, B.·Ngo, M.
View original source ↗Export metadata ↓

Abstract

Glioblastoma (GBM) is the most common primary malignant brain tumor, with patients exhibiting poor survival (median survival time: 15 months). Difficulties in treating GBM include not only the inability to resect the diffusively-invading tumor cells but also therapeutic resistance. The perivascular niche (PVN) within the GBM tumor microenvironment contributes significantly to tumor cell invasion, cancer stem cell maintenance, and has been shown to protect tumor cells from radiation and chemotherapy. In this study, we examine how the inclusion of non-tumor cells in culture with tumor cells within a hydrogel impacts the overall gene expression profile of an in vitro artificial perivascular niche (PVN) comprised of endothelial and stromal cells directly cultured with GBM tumor cells within a methacrylamide-functionalized gelatin hydrogel. Using RNA-seq, we demonstrate that genes related to angiogenesis and remodeling are upregulated in the PVN model compared to hydrogels containing only tumor or perivascular niche cells, while downregulated genes are related to cell cycle and DNA damage repair. Signaling pathways and genes commonly implicated in GBM malignancy, such as MGMT, EGFR, PI3K-Akt signaling, and Ras/MAPK signaling are also upregulated in the PVN model. We describe the kinetics of gene expression within the PVN hydrogels over a course of 14 days, observing the patterns associated with tumor cell-mediated endothelial network co-option and regression. We finally examine the effect of temozolomide, a frontline chemotherapy used clinically against GBM, on the PVN culture. Notably, the PVN model is less responsive to TMZ compared to hydrogels containing only tumor cells. Overall, these results demonstrate that inclusion of cellular and matrix-associated elements of the PVN within an in vitro model of GBM allows for the development of gene expression patterns and therapeutic response relevant to GBM.

Source connections

Explore related subjects

bioengineeringPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Harley, B., Ngo, M.. 2018-03-03. Perivascular signals alter global genomic profile of glioblastoma and response to temozolomide in a gelatin hydrogel. https://doi.org/10.1101/273763

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

N-GLYcanyzer monitoring of a microperfusion process reveals critical parameters' impact on Trastuzumab biosimilar N-glycoform time evolution

Controlling monoclonal antibody (mAb) N-glycosylation as a critical quality attribute (CQA) remains challenging in continuous and intensified biomanufacturing, where conventional end-point analyses obscure dynamic pathway responses. Here, we deployed the N-GLYcanyzer process analytical technology (PAT) for automated monitoring of a microperfusion process that reveals critical parameters' impact on Trastuzumab biosimilar N-glycoform time evolution across a 20-day culture duration of Chinese Hamster Ovary (CHO) cells in a ~2 mL scale-down microbioreactor perfusion platform. Using offline PAT assays combined with our integrated at-line PAT workflow, we systematically screened six perfusion process parameters: carbon source, viable cell density (VCD), perfusion rate, temperature, and pH. While nutrient limitation and thermal/acidic stress significantly disrupted N-glycans maturation, notably triggering high-mannose (Man5) accumulation and severe productivity penalties under pH 6.7 and 39 {degrees}C operation, targeted feed supplementation with galactose and MnCl2 consistently shifted glycoform profiles away from G0F toward galactosylated species (G1F, G1F', G2F). Crucially, coupling this combined feed strategy with an intensified perfusion rate (1.2 VVD) achieved a sustained, desired galactosylation shift without a productivity cost. These findings demonstrate that nutrient feed composition and perfusion rate are primary levers for steering continuous intensified microperfusion processes for biologics production with desired N-glycoform states and highlight the utility of automated at-line PAT like N-GLYcanyzer for enabling real-time attribute-based control in continuous biomanufacturing.

bioengineering↗
bioRxiv2026

Differential sensing with imprinted and non-imprinted organic electrochemical transistors reduces calibration drift in wearable sweat cortisol sensors

The output of a wearable cortisol sensor drifts over days for reasons unrelated to cortisol, so a calibration made on the first day soon stops being valid. In our differential organic electrochemical transistor, a PEDOT:PSS transistor coated with a molecularly imprinted polymer (MIP) for cortisol is paired with one coated with a non-imprinted polymer (NIP) of the same composition, and the NIP response is subtracted. Over four calibrations in artificial sweat across 14 days, the response of the MIP alone shifted by 15% to 22% of its range on day 1, whereas the differential output shifted by 1% to 10%, and the variation between days at 1 to 10nM fell by 65% to 70% on average for individual pairs. In sweat from one participant, read against the calibration from day 1, both outputs tracked the direction of visible changes in salivary cortisol, while the MIP alone diverged from the differential output, consistent with the laboratory drift.

bioengineering↗
bioRxiv2026

A cytotoxicity dataset for data-driven design of ionizable lipid nanoparticles

Ionizable lipid nanoparticles (LNPs) are widely used to deliver nucleic acid therapeutics, but the design of safe LNPs remains challenging due to the complex activity relationships among structure, composition, and physicochemical properties. Here, we present a curated dataset, the Lipid Nanoparticle Cytotoxicity Database (LNPCD), which aggregates 1,413 LNP cytotoxicity measurements across 1,288 unique ionizable lipids drawn from 11 high-throughput combinatorial studies. Lipid structures were generated computationally from the published libraries and manually verified against the original publications. Each record pairs a machine-readable SMILES representation of the ionizable lipid with its measured viability endpoint and the associated experimental descriptors, including cell line, dose, and cargo type. The dataset provides a traceable resource for examining how ionizable lipid structure and experimental conditions relate to in vitro cell viability, and for developing and evaluating computational models of lipid nanoparticle cytotoxicity.

bioengineering↗