Human plasma-like medium facilitates metabolic tracing and enables upregulation of immune signaling pathways in glioblastoma explants

Purpose Metabolism within the tumor microenvironment (TME) represents an increasing area of interest to understand glioma initiation and progression. Stable isotope tracing is a technique critical to the study of tumor metabolism. Cell culture models of this disease are not routinely cultured under physiologically relevant nutrient conditions and do not retain cellular heterogeneity present in the parental TME. Moreover, in vivo, stable isotope tracing in intracranial glioma xenografts, the gold standard for metabolic investigation, is time consuming and technically challenging. To provide insights into glioma metabolism in the presence of an intact TME, we performed stable isotope tracing analysis of patient-derived, heterocellular Surgically eXplanted Organoid (SXO) glioma models in human plasma-like medium (HPLM). Methods Glioma SXOs were established and cultured in conventional media or transitioned to HPLM. We evaluated SXO cytoarchitecture and histology, then performed spatial transcriptomic profiling to identify cellular populations and differential gene expression patterns. We performed stable isotope tracing with 15N2-glutamine to evaluate intracellular metabolite labeling patterns. Results Glioma SXOs cultured in HPLM retain cytoarchitecture and cellular constituents. Immune cells in HPLM-cultured SXOs demonstrated increased transcription of immune-related signatures, including innate immune, adaptive immune, and cytokine signaling programs. 15N isotope enrichment from glutamine was observed in metabolites from diverse pathways, and labeling patterns were stable over time. Conclusion To enable ex vivo, tractable investigations of whole tumor metabolism, we developed an approach to conduct stable isotope tracing in glioma SXOs cultured under physiologically relevant nutrient conditions. Under these conditions, SXOs maintained viability, composition, and metabolic activity while exhibiting increased immune-related transcriptional programs.


Introduction
A hallmark of cancer biology is metabolic reprogramming to provide energy and critical components for cell growth [1][2][3]. There has recently been a resurgence of interest in the metabolism of glioma cells and a recognition of the importance of metabolic adaptations that enable cells to grow and proliferate. Metabolic studies of glioma can provide valuable insight into tumor metabolism and unravel new treatment targets [2]. Intracellular metabolic networks are flexible and sensitive to the concentrations of environmental metabolites [4][5][6]. The current gold standard for investigating tumor metabolism dynamics in the presence of an intact TME entails in vivo stable isotope tracing in mouse models of cancer [7]. These experiments, however, are technically challenging, costly, and time-consuming. More tractable methods for interrogating cancer cell metabolism, such as cell culture models, fail to capture vital contributors to tumor biology, including heterocellular interactions of the tumor microenvironment (TME) and physiologic nutrient levels.
Surgically eXplanted Organoids (SXOs) have emerged as an attractive model to study glioma metabolism. SXOs are efficiently created without mechanical or enzymatic single-cell dissociation of the resected tumor tissue, allowing for maintenance of local cytoarchitecture and native cell-cell interactions [8]. These models recapitulate parental tumor features such as cellular heterogeneity, cell-cell and cell-stroma interactions, gene expression and mutational profiles, and treatment response [8][9][10]. Glioma SXOs are traditionally cultured in the nutrientrich, non-physiologic medium glioma organoid complete (GOC). SXO metabolism has not been interrogated in settings that more closely mirror the parental tumor metabolic milieu.
Recently, several studies have described the use of media designed to better reflect the physiologic environment in various cancer models [1,6,[11][12][13]. Among these is human plasmalike medium (HPLM), which contains physiologically relevant concentrations of typical media components such as glucose, amino acids, and salt ions, as well as metabolites absent from commonly used culture media.
In other tumor subtypes, culture in HPLM has significant effects on cellular metabolism [6], and analysis using stable isotope labeled metabolites in HPLM can reveal tumor metabolic pathway activity [14]. Measuring the contribution of isotopically labeled nitrogen from glutamine to diverse metabolic pathways has emerged as a valuable tool to study cancer cell metabolism [15,16].
Proliferating glioma cells use glutamine as a substrate for the synthesis of nucleosides, nonessential amino acids, and glutathione, among other pathways. Additionally, evidence has emerged that culture in HPLM also has immunomodulatory effects, including enhancing T cell activation following antigen stimulation through diverse transcriptional responses [17].
Here, we cultured SXOs in SXO-adapted HPLM to enable stable isotope tracing (Fig. 1). We found that SXOs cultured in HPLM maintain histologic characteristics and TME cell constituents, increase immune cell signaling, and display stable metabolic activity during prolonged HPLM culture. Formalin-fixed paraffin-embedded tissue slides were deparaffinized and stained with fluorescent antibodies and antibodies coupled to DNA-barcoded oligos (SYTO13 to identify DNA, GFAP to identify tumor cells, and CD45 to identify immune cells). Fluorescent section images were collected, then spatially resolved regions of interest (ROIs) were selected for analysis. Oligotagged probes for each ROI were collected in 96-well plates and subjected to next-generation sequencing (NGS) preparation. Stocks of GOC were used up to 1 week after preparation. Plates were rotated at 120 rpm in a humidified incubator at 37°C, 5% CO2, and 5% oxygen. GOC was refreshed in SXO cultures every 48 hours. All SXOs were cultured for a minimum of eight weeks prior to experimentation.

Liquid Chromatography-Mass Spectrometry (LC-MS). For metabolomic and stable isotope
tracing experiments, SXOs and GSCs were snap-frozen and stored at -80°C until analysis. For preparation of SXOs, accurate masses of snap frozen SXOs were obtained using an analytical balance. 100µL 80% LC-MS grade acetonitrile (Fisher Scientific A9554) prepared in LC-MS grade water (Fisher Scientific W6500) per mg of tissue was added to snap frozen SXOs on ice, after which SXOs were homogenized by manual agitation. Homogenized samples were vortexed at 4°C for 20 minutes, then centrifuged for 10 minutes at 21,100xg at 4°C. The supernatant was transferred to a fresh microcentrifuge tube and centrifuged again for 10 minutes at 21,100xg at 4°C. For GSCs, neurospheres were harvested from 6-well plates, followed by the addition of 4°C saline to quench metabolic activity. Samples were transferred to microcentrifuge tubes and centrifuged for 1 minute at 21,100xg at 4°C. The resulting supernatant was aspirated, with the remaining cell pellets snap frozen in liquid nitrogen and stored at -80°C. Metabolites were extracted in 80% Acetonitrile at a concentration of 1,000 cells/µL, vortexed for 20 minutes at 4°C, and centrifuged (21,100xg for 10 minutes at 4°C). The supernatant was transferred to a fresh microcentrifuge tube and centrifuged again (21,100xg for 10 minutes at 4°C). The supernatant was transferred to a glass vial for LC-MS analysis. 10µL of SXO sample or 20uL of GSC sample was injected and analyzed with an Q Exactive™ HF-X orbitrap mass spectrometer (ThermoFisher) coupled to a Vanquish ultra-high performance liquid chromatography system (ThermoFisher). Peaks were integrated using El-Maven software (0.12.0, Elucidata). Total ion counts were quantified using TraceFinder software (5.1 SP2, ThermoFisher). Peaks were normalized to total ion counts using the R statistical programming language. Correction for the natural abundance of 15 N isotopes was accomplished using the R script Accucor [21]. Total fractional enrichment was calculated by summing fractional enrichment of isotopically labeled nitrogen in each isotopologue of the metabolite. Quantification and Statistical Analysis. SXOs were allocated to experiments randomly and samples were processed in an arbitrary order. All statistical tests were two-sided, where applicable. Student's t-test was used to assess the statistical significance of a difference between the two groups. Linear regression analysis and Pearson's correlation coefficient were used to assess correlation between the two groups. Statistical analyses were performed with GraphPad Prism (9.2.0.332, GraphPad Software, LLC) and included both descriptive statistics as well as tests of statistical significance. All data are plotted as mean ± standard deviation. For all tests, p-values less than 0.05 were considered statistically significant.

Results
We established a SXO model, SXO210, from a 68-year-old male patient diagnosed with a right temporal glioblastoma, IDH wild-type, grade 4 (Supplementary Table 1). Clinical pathology was consistent with glioblastoma, highlighted by GFAP immunostaining, a high mitotic index, areas of pseudopalisading necrosis, geographic necrosis, and microvascular proliferation. From this tumor, we successfully generated SXOs, which were cultured in GOC medium for 8 weeks prior to initiating these studies.

SXO cellular architecture is preserved after culture in SXO HPLM
We assessed SXO architecture and viability after culture in SXO HPLM, compared to in GOC, by performing histological analyses, which were reviewed by a board-certified neuropathologist.

Each SXO was independently stained with hematoxylin and eosin (H&E). SXOs cultured in SXO
HPLM maintained typical histologic features of glioblastoma, including necrosis, microvascular proliferation, high cellularity with abundant mitosis, and nuclear atypia (Fig. 2e-f). Quantitative analysis of cellular nuclei demonstrated that SXOs cultured in SXO HPLM for either 24 or 120 hours maintained similar cell density to those cultured in GOC, indicating limited impact of shortand medium-term HPLM culture on cellularity (Fig. 2g). SXOs were subsequently stained for cellular proliferation using Ki67 (Fig. 2h-m). Quantitative bioimage analysis of the proportion of cells with positive staining for Ki67 between the GOC and both SXO HPLM-cultured groups were not significantly different, indicating a similar degree of cell proliferation upon short-and medium-term exposure to HPLM (Fig. 2n). Finally, SXOs were stained for the marker of the stem-like cell fraction, Sox2 (Fig. 2o-2t). The percentage of cells positive for Sox2 staining was equivalent between SXOs cultured in GOC and in SXO HPLM, suggesting that the stem-like cell population was not significantly altered upon culture in HPLM (Fig. 2u).

SXOs cultured in HPLM maintain tumor microenvironment cellular composition and upregulate immune transcriptional programs
To assess the TME of SXOs cultured in SXO HPLM or GOC, we performed spatial transcriptomics with a NanoString GeoMx digital spatial profiler (DSP). We used immunofluorescent probes for the marker SYTO13 to identify all cells, GFAP to specifically mark tumor cells [23], and CD45 to specifically identify immune cells [24] on fixed SXO sections cultured in either GOC, HPLM for 24 hours, or HPLM for 120 hours (Fig. 3a). We then scanned these regions to construct digital maps of cellular content and selected regions of interest (ROIs) containing diverse heterocellular populations representative of the whole SXO or regions exclusively containing immune cells.
Individual ROIs were collected for independent transcriptional analysis by next-generation sequencing. Using next-generation sequencing of representative heterocellular SXO ROIs, we classified cell types present using the CIBERSORT digital cytometry platform [25]. We found that SXOs contained a diverse, spatially variable population of constituent cells, consistent with prior work [10,26]. 20-30% of cells in each SXO were identified as non-tumor, including CD4+ T-cells, CD8+ T-cells, and macrophages/microglia (Fig. 3b). SXOs cultured in SXO HPLM and GOC maintained similar proportions of tumor cells (Fig. 3c), CD4+ T-cells (Fig. 3d), CD8+ T-cells (Fig.   3e), macrophages/microglia (Fig. 3f), endothelial cells (Fig. 3g), and other tumor immune microenvironment cells (TIMEC), including B-cells, cancer-associated fibroblasts (CAFs), and natural killer (NK) cells (Fig. 3h). Activation of CD4+ T-cells can be marked by increased expression of CD69 [27]. We observed increased expression of CD69 in CD4+ T cells of SXOs cultured in HPLM for 120 hours, but not 24 hours (Fig. 4a). TGF-β signaling in bulk immune cell populations and T regulatory (T reg ) cell signaling in T reg cells were significantly enriched in SXOs cultured in HPLM for 120 hours relative to SXOs cultured in GOC (Fig. 4b). Gene set enrichment analysis revealed increased expression of pathways related to the innate immune system, adaptive immune system, and cytokine signaling in these cells (Table 1). These data indicate that culture of SXOs in HPLM retains TME heterogeneity while inducing immunologic transcriptional programs.

Measurement of isotopic labeling by liquid chromatography-mass spectrometry (LC-MS) in SXOs cultured in HPLM
We utilized LC-MS to quantify metabolite levels and paired 15 N isotopically labeled metabolite levels for a diverse set of metabolites in SXOs cultured in HPLM, utilizing three GSC lines (UTSW63, TS516, and HK157) as controls for label accumulation. We selected metabolites expected to be enriched for isotopic labeling from glutamine for further analysis. We first confirmed intra-organoid and intracellular accumulation of the isotopic label in glutamine, which was present in culture conditions containing the 15 N2 glutamine stable isotope tracer and not in unlabeled conditions (Fig. 5a). We then determined the fractional enrichment of isotopic label derived from 15 N2 glutamine in the non-essential amino acids alanine ( Supplementary Fig. 1a), asparagine ( Supplementary Fig. 1b), and glutamate (Fig. 5c), the branched-chain amino acid isoleucine (Fig. 5d), the redox substrate glutathione (Fig. 5e), the nucleoside citicoline (Fig. 5f), the glutamate derivative N-acetylglutamate ( Supplementary Fig. 1c). We compared nitrogen- Analysis of metabolism in SXOs may provide insights into how gliomas alter typical neural cell metabolic programs to foster growth and proliferation. This approach also captures features of whole tumor metabolism driven by microenvironmental components that are lacking in pure tumor cell cultures. Evaluating the biochemical networks active in these primary models under physiological nutrient levels may aid in the identification of novel tumor-specific metabolic phenotypes and associated targetable vulnerabilities.

Limitations
We acknowledge that our study comes with limitations. The number of SXOs utilized was limited (n = 9 for both histology and stable isotope tracing), and all experimental replicates derive from a single patient, thus potentially underpowering this study to detect true differences in organoid viability and metabolic activity. We evaluated culture of SXOs and GSCs in HPLM for a maximum of 120 hours. This limited length may fail to identify the impact of longer-term adaptation on SXO metabolic activity or viability.

Conclusion
We demonstrated ex vivo stable isotope tracing under conditions that recapitulate the in vivo TME using SXOs grown in HPLM. The provision of nutrients at levels that mimic in vivo settings triggered an activation-related transcriptional response in immune cells present in SXOs. Our work offers an approach for evaluating whole tumor metabolism that addresses key challenges associated with conventional methods.