Patient samples
Two primary (UH-SCC-17A, UH-SCC-18A) and their corresponding metastatic (UH-SCC-17B, UH-SCC-18B) oral tongue squamous cell carcinomas (OTSCC) and one metastatic (UH-SCC-23B) buccal squamous cell carcinoma (SCC) were collected from three consenting patients under the ethical permission 31/2016 handed by The Ethical Committee of North Ostrobothnia's Hospital District (PPSHP). Clinical and histopathological features of the patients are presented in Figure 1 and detailed further in Supplementary Table 1.
Histopathological slides were collected from the Helsinki University Hospital. After the surgical removal of the primary tumour and the metastatic lymph nodes, patients received the appropriate treatment (radiotherapy, chemotherapy, targeted therapy, and immunotherapy) based on the oncologist’s recommendation.
Fresh tumour samples were received in 50 ml falcon tubes containing ice-cold Hanks’ Balanced Salt solution (HBSS) (Thermo Fisher Scientific, Waltham, Massachusetts, United States), supplied with 100 U/ml penicillin (Thermo Fisher Scientific), 100 μg/ml streptomycin (Thermo Fisher Scientific) and 250 ng/ml amphotericin B (Thermo Fisher Scientific). Tissue samples were kept on ice in a Petri dish containing HBSS while the necrotic tissue was removed using a scalpel. Living tissue pieces were moved into a new Petri dish containing HBSS and chopped into 1–2 mm pieces with a scalpel. Tissue pieces were placed in a 15 mL falcon tube and centrifuged for 5 min at 1000 rpm (200× g) at 4 °C. After centrifugation, the supernatant was discarded and fresh HBSS buffer was added before another round of centrifugation. The supernatant was discarded, and the pellet containing tissue pieces was suspended in 5 ml HBSS buffer containing 1 mg/ml collagenase type I from Clostridium histolyticum (Sigma-Aldrich, St. Louis, Mo, USA) and placed on a rocker platform at 37 °C. After 1 h of incubation, the tube was centrifuged and the supernatant was discarded and replaced with fresh HBSS buffer before another round of centrifugation. The digested sample was suspended in HBSS buffer filtered using a 100 μm cell strainer (Falcon™ Cell Strainer, Fisher Scientific, NH, USA), and the flow-through (single cells) was collected and centrifuged. The supernatant was discarded, and the cell pellet was suspended in a minimal essential medium (MEM) supplemented with L-glutamine (2 mmol/l), 10% fetal bovine serum, nonessential amino acid solution, penicillin (100 U/ml), streptomycin (100 μg/ml) and 250-ng/ml amphotericin B (all from Thermo Fisher Scientific).
Peripheral blood mononuclear cells (PBMNCs) were isolated from the peripheral blood of a tongue tumour patient. A density gradient technique was followed using Ficoll–Paque PLUS (GE Healthcare, Piscataway, NJ, USA).
Luminescent cell viability assay on Myogel-coated wells
PerkinElmer 96-TC plates were coated using 50 µl of Myogel (0.5 mg/ml) per well and incubated in a 37° C, 5% CO2, 95% humidity incubator overnight. On the next day, we aspirated the excess Myogel and seeded 100 ul of cell suspension containing 1000 cells in each well. 100 ul of cell culture media was added into blank wells, and plates were transferred to the incubator. Twenty four hours after cell seeding, the appropriate drugs (cisplatin: 10 uM, cetuximab: 5 ug/ml, carboplatin: 104 nM, fluroracil: 103 nM) were diluted and added to the corresponding wells (4-10 replicates). For radiotherapy testing, plates were irradiated with 2 Gy/day for three days, using a Gamma irradiator OB29/4 (STS, Braunschweig, Germany). The source of radiation was Caesium, isotope 137. On day 5, we added 100 ul of CellTiter-Glo® to all plate wells; plates were then shaken for 5 min at 450 rpm using a plate shaker and centrifuged for 5 min at 1000 rpm. We used BMG Pherastar FS-plate reader to detect viability.
Zebrafish microinjection
Two days post-fertilization (dpf), wild-type zebrafish (Danio rerio) larvae (from AB strain) were used for the in vivo assays. Fish were dechorionated, anesthetized with 0.04% tricaine, and microinjected into the perivitelline space with 4 nl of cell suspension (1500 cells per fish). Fish were then transferred into a 24-well plate (five fish per well) containing 1 mL of fresh embryonic medium and randomly distributed into different groups based on the treatment options. Drugs at a similar concentration as in the in vitro assay were added to the fish water. For radiotherapy testing, fish were subjected to 2 Gy/day for three days. All larvae were incubated for three days at 34°C.
Quantitative PCR
An RNeasy Mini Kit (Qiagen, Düsseldorf, Germany) was used to extract RNA from the fish according to the manufacturer’s instructions. An iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA) was used to synthesize cDNA from a total of 400 ng. 10 μl iQ SYBR green, 7 μl nuclease-free water and 1 μl of a primer solution were added to 2 μl of a cDNA sample. Human glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used to detect the signal from human tumour cells using a primer with the sequences: Forward: 5’-AAGGTCATCCCTGAGCTG-3’, Reverse: 5’-TGCTGTAGCCAAATTCGTTG-3’. Zebrafish GAPDH was used as the housekeeping gene with a primer having the following sequences: Forward 5’-AGTGTCAGGACGAACAGAGGCT-3’ Reverse 5’-GCCAATGCGACCGAATCCGTTA-3’.
Microfluidic chip assay
3D microfluidic chips were designed and obtained from Probiont™ as in our previous publication [15]. Freshly isolated tumour cells were stained with Celltrace Far Red (Invitrogen, Carlsbad, CA, USA) and embedded in a Myogel-fibrin mixture. Myogel/fibrin was prepared using the following concentrations: 2.4 mg/ml Myogel, 0.5 mg/ml fibrinogen (Merck, Darmstadt, Germany), 0.3 U/ml thrombin (Sigma-Aldrich), and 33.3 μg/ml aprotinin (Sigma-Aldrich). Caspase 3/7 marker (2.5-μM; Sartorius, Göttingen, Germany) was added to detect apoptotic cells. Next, 2 uL of cell suspension (containing 500 cells) in gel was loaded in each small channel of the chip. Immune cells isolated from the patient’s peripheral blood were stained with CellTrace Violet (Invitrogen) and suspended in DMEM/F12 media (Thermo Fisher Scientific) supplied with 10% serum. The mix was then divided into three groups: tumour cells only, tumour and immune cells, and tumour and immune cells in the presence of pembrolizumab (186 ug/mL). 100 uL of immune cell suspension (100 000 cells) containing caspase 3/7 marker (2.5-μM) was injected into the large channel. After injection, chips were kept in the incubator for three days and imaged daily.
Imaging of microfluidic chips
Chips were imaged daily under a fluorescent microscope using Nikon Ti-E with Alveole Primo (Nikon, Tokyo, Japan) connected to a Hamamatsu Orca Flash 4.0 LT B&W camera (Hamamatsu Photonics, Hamamatsu, Japan) and Lumencor Sola SE II 365 (Lumencor, Beaverton, Oregon, USA). Images for the following transmitted light and fluorescent filters were acquired: DAPI (Semrock 5060C, excitation 377/50, emission 447/60), GFP (Semrock 3035D-NTE, excitation 472/30, emission 520/35), and Cy5 (excitation 640/20, emission 700/75). Chips were placed on a microscope slide adapter. The NIS-Elements Advanced Research program was automated to image at 20x magnification and to scan 10 images horizontally to form a complete representation of the channel producing a multichannel composite image.
Semi-automated counting of cells
Before analysis, multichannel images were cropped to contain only the tumour cell channel. Using the built-in algorithms of Image J (NIH, National Institutes of Health, USA), we coded a semi-automated positive cell counter to quantify intensities in the three different channels of the composite image. The algorithm separated tumour cells from the background based on intensity and morphology. All detected red tumour cells were analysed for positive green intensity to calculate the apoptotic tumour cells with a minimum threshold to exclude artefacts.
DNA sequencing
Tumour-derived DNA for whole exome sequencing was extracted from fresh-frozen tissue of four patient samples, using the ALLPrep DNA/RNA Mini Kit (Qiagen) according to the manufacturer’s instructions.
Genomic DNA concentrations were measured with Qubit and quality was assessed using Genomic DNA ScreenTape Assay (Agilent TapeStation 4200, California, United States). 100 ng of good quality Genomic DNA (DIN > 6) was then fragmented using mechanical sheering (Covaris, Woburn, United States) and converted into Illumina-compatible sequencing libraries using the KAPA HyperPrep kit (Roche, Basel, Switzerland). Each library had its own set of Unique Dual Indexes (UDI).
The four separate sequencing libraries were combined and captured (4-plex) using Roche’s KAPA HyperExome probes (target size 43 Mb) according to the manufacturer’s instructions.
The captured 4-plex library pool was then sequenced at a concentration of 1.3 pM using a NextSeq Mid Output 300 cycle flow cell on the NextSeq 500 to produce 2 x 150 bp reads.
Conversion Software (bcl2fastq v2.20) was used to convert BCL files to FASTQ file format and demultiplex the samples. Sequenced reads were trimmed for adapter sequences and quality filtered with Trimmomatic (v. 0.39) to remove reads that were shorter than 36 nucleotides.
Data pre-processing and variant calling was performed following Genome Analysis Toolkit (GATK version 4.2.0.0, Broad Institute) best practices workflow for somatic short variant discovery. The raw sequencing reads were aligned to the Grch38 (from GATK resource bundle, annotations from gencode human v38) human reference genome using the BWA-mem algorithm (version 0.7.17), and duplicated reads were identified and tagged with the MarkDuplicates tool from Picard. Finally, base quality score recalibration was performed on samples using high confidence variant data obtained from the GATK resource bundle. Somatic single nucleotide variants (SNV) and indels were detected by the GATK Mutect2 and subsequently filtered by the FilterMutectCalls tool. Additional annotations were added for variants/indels with Funcotator.
The sequencing, read processing and subsequent variant calling was performed by the Biomedicum Functional Genomics Unit at the Helsinki Institute of Life Science and Biocenter Finland at the University of Helsinki.
Statistics
In vitro luminescent cell viability assays on Myogel-coated wells were done in 4–10 replicates. The zebrafish xenograft experiments were conducted using 3–12 groups of fish per treatment, each group containing eight fish pooled together. Three chips were used for each condition in the 3D microfluidic chip experiment. SPSS software program version 26.0 (IBM SPSS Statistics, SPSS, IL, USA) was used for statistical analyses. We used non-parametric Kruskal Wallis one-way ANOVA to determine the statistical significance; P-values ≤ 0.05 were regarded as significant. As zebrafish experiments are known for producing outlier values, we ran Grubbs’ test to detected and remove outliers. P < 0.01 was regarded as a significant outlier.