Abstract

Background

   Organoids are approved by the US FDA as an alternative to animal
   experiments to guide drug development and for sensitivity screening.
   Stable organoids models of gastric cancer are desirable for
   personalized medicine and drug screening.

Methods

   Tumor tissues from a primary cancer of the stomach and metastatic
   cancer of the lymph node were collected for 3D culture. By long-term
   culture for over 50 generations in vitro, we obtained stably growing
   organoid lines. We analyzed short tandem repeats (STRs) and karyotypes
   of cancer cells, and tumorigenesis of the organoids in nude mice, as
   well as multi-omics profiles of the organoids. A CCK8 method was used
   to determine the drugs sensitivity to fluorouracil (5-Fu), platinum and
   paclitaxel.

Results

   Paired organoid lines from primary cancer (SPDO1P) and metastatic lymph
   node (SPDO1LM) were established with unique STRs and karyotypes. The
   organoid lines resulted in tumorigenesis in vivo and had clear genetic
   profiles. Compared to SPDO1P from primary cancer, upregulated genes of
   SPDO1LM from the metastatic lymph node were enriched in pathways of
   epithelial-mesenchymal transition and angiogenesis with stronger
   abilities of cell migration, invasion, and pro-angiogenesis. Based on
   drug sensitivity analysis, the SOX regimen (5-Fu plus oxaliplatin) was
   used for chemotherapy with an optimal clinical outcome.

Conclusions

   The organoid lines recapitulate the drug sensitivity of the parental
   tissues. The paired organoid lines present a step-change toward living
   biobanks for further translational usage.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12967-024-05512-0.

   Keywords: Gastric cancer, Living biobank, Organoid models, Drug
   sensitivity, Translational usage

Introduction

   Gastric cancer is malignancy with high incidence and mortality in East
   Asia [[50]1, [51]2], and the poor prognosis is attributed to gastric
   cancer metastasis [[52]3–[53]5]. Lymphatic metastasis is the most
   common metastatic site in advanced gastric cancer [[54]6]. The regimen
   of platinum plus fluorouracil (S-1 or capecitabine) is often used in
   gastric cancer treatment. Regimens of two or multi-drugs are generally
   accepted in different medical centers, such as the SOX (S-1 plus
   oxaliplatin), DCF (docetaxel, cisplatin and 5-fluorouracil), AS
   (albumin-bound paclitaxel and S-1), ECF (epirubicin, cisplatin, and
   5-fluorouracil), and FLOT (5-fluorouracil, leucovorin, oxaliplatin, and
   docetaxel) regimens [[55]7–[56]10]. However, there is no conclusion
   about which regimen is the best balance between efficacy and toxicity.
   The therapeutic response may relate to phenotypic features of
   histological types and the degree of differentiation [[57]10, [58]11].
   A prospective study demonstrated that there is only a 35% of complete
   responsive rate to FLOT or FOLFOX (5-fluorouracil plus leucovorin and
   oxaliplatin) regimens [[59]12]. If the sensitivity of cancer cells to
   drugs is known in advance, personalized treatments can be used to kill
   cancer cells to the maximum extent.

   Since the successful cultivation of intestinal organoids from mouse
   intestinal mucosa in 2009 [[60]13], organoid technology has developed
   rapidly. Organoid is a 3D cellular model in vitro, which can simulate
   the characteristics of parental organs in terms of tissue structure,
   cell function, and genomic profile. Organoid has been widely used in
   biomedical research, especially in cancer research, such as gastric
   cancer [[61]14], lung cancer [[62]15–[63]17], pancreatic cancer
   [[64]18], colorectal cancer [[65]19–[66]21], bladder cancer [[67]22],
   and neuroendocrine tumors [[68]23]. Recently, the FDA Modernization Act
   2.0 was approved by the US government, which allows organoids to be
   used as models for drug screening and sensitivity assay as an
   alternative to animals [[69]24]. Li and colleagues constructed a
   signet-ring cell organoid line from colon cancer, which recapitulated
   the histology of the original cancer tissue [[70]25]. Hoshi and
   colleagues constructed an organoid line of acinar cell carcinoma from
   the pancreas, which could tolerate cryopreservation and recovery and
   showed functions similar to the original tissue [[71]26]. However,
   there have been no reports on the construction of stable organoid lines
   from gastric cancer.

   Recently, we constructed an organoid living biobank for gastric cancers
   [[72]27]. In this study, a pair of organoid lines derived from a
   poorly-differentiated adenocarcinoma and its metastatic lymph node were
   successfully cultured for over fifty generations. We characterized
   parameters including cell doubling time, culture medium requirements,
   tumorigenesis in vivo, STRs, karyotypes, genetic profiles, and drug
   sensitivity. Importantly, based on the drug sensitivity analyses, the
   patient received a better therapeutic regimen and achieved a sustained
   response.

Materials and methods

Ethics statement

   Protocols for fresh organoid culture of gastric cancer were approved by
   the Ethics Committee for Scientific Research of Ruijin Hospital, and
   informed consents were obtained from the patients. An additional number
   was used to register the samples in the database with no link to
   patient names or personal information.

Establishment of organoid lines

   The primary cancer and lymphatic metastatic cancer of the station six
   were collected after radical gastrectomy within 30 min ex vivo and
   stored in advanced DMEM/F12 medium (12,634,010, Thermo Fisher
   Scientific). The tissues were washed with PBS containing
   1×penicillin-streptomycin (C0222, Beyotime) and 1×puromycin (A1113802,
   Thermo Fisher Scientific) at least four times, and then minced to 1–2
   mm^3 for further digestive incubation in buffer including 1 mg/ml
   collagenase(40507ES60, Yeasen), 1 mg/ml collagenase IV (40510ES60,
   Yeasen), and 1 U/ml dispase II (40104ES60, Yeasen) for 1.5 h at 37℃
   with shaking. The incompletely digested tissue pieces were discarded
   and then cells were centrifuged at 1000 rpm for 5 min. The cell
   precipitates were mixed with DMEM and matrigel (356,231, Corning) at a
   1:1 ratio and seeded in a 24-well plate (50 µl/well). Once the matrigel
   was polymerized after 30 min incubation at 37 °C, complete medium
   (Table [73]S1) was added (500 µl/well). The organoid culture was
   performed at 37 °C, 5% CO[2], and the medium was replaced every 5 days.
   The well-grown organoids were obtained after 7 to 10 days of
   cultivation. The organoids were digested with TrypLE Express enzyme
   (12,605,010, Thermo Fisher Scientific) for 0.5–1 h. The cell
   suspensions were collected and centrifuged at 1000 rpm for 5 min. The
   cell precipitates were mixed with matrigel at a 1:1 ratio and seeded in
   a 24-well plate (50 µl/well, 5 × 10^4 cells) again for passaging or
   subsequent study.

Hematoxylin and eosin (H&E) staining

   Well-grown organoids (diameter > 100 μm) were blown thoroughly at least
   10 times and centrifuged at 1500 rpm for 5 min. The organoid spheres
   were collected and fixed in 4% paraformaldehyde for 30 min, put in 10%
   agarose, and embedded in paraffin. The paraffin block was cut into 8-µm
   sections. The sections were treated in xylene twice for 10 min, and
   immersed in pure ethanol for 5 min, 95% ethanol for 5 min, and 75%
   ethanol for 5 min. The sections were heated with 1×antigen repair
   reagent at 100℃ for 20 min, and washed with clean water. After washing,
   the sections were stained in hematoxylin for 10 min, immersed in
   hydrochloric acid and ammonium hydroxide for several seconds, and
   washed in running water for 1 h. The sections were stained in alcohol
   eosin for 2 min (G1005, Servicebio), and dehydrated in 70% and 90%
   alcohol for 10 min, and further immersed in 95% ethanol and pure
   ethanol for 5 min, respectively. The sections were transparent in
   xylene for 10 min twice, and sealed with neutral resin.

Immunohistochemistry (IHC) and western blotting

   In immunohistochemical assays, the sections were prepared in the same
   way as in H&E stain. The staining methods were performed as previously
   described [[74]28]. The primary antibodies included E-cadherin
   (GB11082, Servicebio, 1:200), N-cadherin, [75]GB111273, Servicebio,
   1:200), and Ki67 ([76]GB111141, Servicebio, 1:200). The western blot
   assay was performed as previously described [[77]29]. The primary
   antibodies included p-mTOR (5536, CST, 1:1000), mTOR (2972, CST,
   1:1000), p-AKT (4060, CST, 1:200), AKT(4691, CST, 1:1000),
   HRP-conjugated goat anti-rabbit IgG (H + L) (SA00001-2, Proteintech,
   1:3000), and HRP-conjugated anti-GAPDH (HRP-60,004, Proteintech,
   1:3000).

Cell growth analysis

   The well-grown organoids were digested with TrypLE Express enzyme for
   0.5–1 h. The single-cell suspension was centrifuged at 1000 rpm for
   5 min. The cell precipitate was mixed with matrigel at a 1:1 ratio and
   seeded in a 96-well plate (6 µl /well, 6000 cells, n = 3). After
   polymerization of the matrigel for 30 min at 37 °C, the complete medium
   (60 µl/well) was added. Cellular vitality of the organoids was measured
   by a CCK8 assay (CK04, DOJINDO) every 48 h for 8 days. The growth
   curves were plotted using time (hour) as the x-axis and cell vitality
   from the 450-nm OD value as the y-axis.

Short tandem repeats (STRs) analysis

   The well-grown organoids were digested with TrypLE Express enzyme for
   15 min and centrifuged at 1500 rpm for 5 min to collect the organoid
   spheres. DNA was extracted with a genome extraction kit (Ap-mn-p-500,
   Axygen), amplified by the 21-STR amplification protocol, and detected
   on an ABI 3730XL. We focused on eight core STR loci (D5S818, vWA,
   D7S820, D16S539, TH01, D13S317, TPOX and CSF1PO) and the sex chromosome
   Amelogenin. The STR results were matched with the STR data library in
   DSMZ tools, which covers STR data of 2455 cell lines from the ATCC,
   DSMZ, JCRB and RIKEN databases. When the STR matching rate was above
   the cut-off value (≥ 75%), the two matched lines were judged to be
   homologous, and otherwise, unique [[78]30].

Karyotypes analysis

   Colchicine solution (0.2 µg/ml) was added to well-grown organoids and
   incubated for 1.5 h at 37℃. Organoids were then digested with TrypLE
   Express enzyme for 0.5–1 h. The single-cell suspensions were
   centrifuged at 1500 rpm for 5 min. The cell precipitates were preheated
   with 6 ml of KCL hypotonic solution (0.075 mol/L) at 37℃ in a water
   bath for 30 min with gentle agitation and were then thoroughly dried.
   The fixative solution (1 ml of methanol and glacial acetic acid at a
   ratio of 3:1) was slowly added for 30 min, and then the cells were
   centrifuged at 1500 rpm for 5 min. The cell precipitates were further
   fixed by the above fixative solution (5 ml) for 30 min and then
   centrifuged at 1500 rpm for 5 min. The cells were resuspended in 0.5 ml
   fixative solution. One or two drops of the cell suspension were added
   to a cooled slide and fixed by an alcohol flame. After drying at room
   temperature, the cells were stained with Giemsa for 10 min, the dye was
   washed and the slide was dried at room temperature. The cells in the
   division phase were observed and captured using the ZEISS karyotyping
   system (Ikaros, ZEISS).

Whole-exome sequencing (WES)

   The paired organoid lines (at passage 18) and cryopreserved primary
   cancer tissues were used for DNA extraction (DP304, Tiangen). The DNA
   was fragmented by enzyme digestion and adenine deoxynucleotide and a
   sequencing adaptor were added. By PCR amplification, the enriched
   library was hybridized by Agilent SureSelect Human All Exon V6 to
   capture exonic regions. Genomic sequencing was performed using an
   Illumina PE150.The FASTQ format of sequencing data was mapped to the
   human reference genome (hg38) by the BWA Tool. Single nucleotide
   polymorphisms (SNPs; Sentieon DNAseq 201808.05), copy number variation
   (CNV; CNVkit, 0.9.6), somatic single nucleotide variants (SNVs;
   Sentieon DNAseq 201808.05), and insertions and deletions (InDel;
   Sentieon DNAseq 201808.05) were analyzed. The driver gene mutations
   were screened based on the CGC513, Bert Vogelstein125, SMG127 and
   Comprehensive435 databases. The raw data (PRJNA1015608) were uploaded
   to SRA database (PRJNA1015608).

Single-cell transcriptome sequencing (scRNA-seq)

Cell dissociation and single-cell capture

   The well-grown organoids were digested with TrypLE Express enzyme for
   0.5–1 h. The single-cell suspensions with cell vitality greater than
   70% were collected into nuclease-free water.

Library preparation and sequencing

   The Chromium Single Cell 3ʹ Gel Beads-in-emulsion (GEM) Library &Gel
   Bead Kit v3 was used for library construction, and sequencing was
   performed on a Novaseq 60,000 (Illumina). The raw data (GSE242893) were
   updated to GEO database.

Quality control and data preprocessing

   ScRNA-seq data analysis was performed using the 10×Genomics software
   package (Cell Ranger version 2.1.1) for data quality control and
   alignment. SCfastp was used for sequence filtering and sequence quality
   assessment. CellrangerCounts was used to count the number of cells. A
   total of 7662 cells and 6257 cells were collected from SPDO1LM and
   SPDO1P organoid lines, respectively. Gene expression matrixes were
   generated using unique molecular identifiers (UMIs). The alignment of
   transcriptome data to the GRCh38 reference genome
   (GRCh38_Ensembl_Ensembl104) was performed by CellrangerAggr. The UMIs
   were mapped to genes. Low-quality cells that matched the following were
   removed: (1)expressed gene number of cells are < 200 or > 7500.2
   expressed mitochondrial genes or ribosomal genes of cells were > 25%. A
   total of 10,578 cells from SPDO1P and SPDO1LM were used for analysis.

Data normalization

   The Seurat package (4.1.1) was used for data normalization, including
   “NormalizeData” and “LogNormalize”. The log-transformed UMI counts were
   computed to obtain the normalized gene expression value.

Dimension reduction

   A subset of 2500 highly variable genes (HVGs) was selected for analysis
   by “FindVariableFeatures” and “ScaleData”. The “RunPCA” of the Seurat
   package was used for principal components analysis (PCA). The
   “JackStraw” was used to determine 20 PCA. The “ScoreJackStraw” was used
   to obtain principal components scores.

Cell clustering and gene set enrichment analysis (GSEA)

   The “FindClusters” was used for cell clustering analysis. The linear
   dimensionality reduction and plotting of tSNE was performed by
   “RunTSNE” with a resolution of 0.2. The “FindAllMarkers”
   (min.pct = 0.25, logfc.threshold = 0.25) was used to obtain
   differential genes of the subclusters. The GSEA was performed to obtain
   enriched pathways.

Transcriptome sequencing (RNA-seq)

   Tissues (5 × 5 mm) and organoid cells (1.5 × 10^6) were collected for
   RNA-seq assays. Total RNA was extracted with TRIzol Reagent. The
   RNA-seq libraries were prepared using the Fast RNA-seq Lib Prep Kit
   V2(RK20306, ABclonal, China) according to the manufacturerˊs
   instructions. The sequencing was performed by the NovaSeq xplus. Gene
   counts were normalized to transcripts per million (TPM) for data
   analysis.

3D invasion assay

   The well-grown organoids were digested with TrypLE Express enzyme for
   15 min and then centrifuged at 1000 rpm for 5 min. The precipitates
   were mixed with matrigel (354,231, Corning) at a ratio of 1:1 and
   seeded in a 24-well plate (50 µl/well, 1000 organoid spheres). After
   matrigel polymerization for 30 min at 37℃, complete medium
   (500 µl/well) was added. Images were captured at 0 and 12 h using an
   inverted microscope.

2D migration assay

   The well-grown organoids were digested with TrypLE Express enzyme for
   15 min and centrifuged at 1000 rpm for 5 min. Cells (300 µl) were
   resuspended in serum-free DMEM (2 × 10^5 cells) and added to the upper
   chamber of the transwell chamber. Complete medium (600 µl) was added to
   the lower chamber. The cells were incubated at 37℃ and 5% CO[2] for
   72 h, then stained with 1% crystal violet for 30 min. The cells in the
   upper chamber were wiped off and dried at room temperature. The cells
   were counted, and images were captured under a microscope.

2D invasion assay

   The well-grown organoids were digested with TrypLE Express enzyme for
   15 min, and centrifuged at 1000 rpm for 5 min. Matrigel (10% 100 µl)
   was added to the upper chamber of the transwell chamber and polymerized
   for 30 min at 37℃. Cells (2 × 10^5) were resuspended in 200 µl of
   serum-free DMEM (cells) and added to the upper chamber of the transwell
   chamber. Complete medium (600 µl) was added to the lower chamber. The
   cells were incubated at 37℃ and 5% CO[2] for 96 h, then stained with 1%
   crystal violet for 30 min. The cells in the upper chamber were wiped
   off and dried at room temperature. The cells were counted and images
   were captured under a microscope.

Tubule formation assay

   Human umbilical vein endothelial cells (EA.HY926) were evenly seeded in
   a 96-well plate coated with matrigel. The supernatants of SPDO1LM and
   SPDO1P organoid lines were added for incubation for 48 h. The tubular
   length and branch nodes were counted under a microscope.

Gene knockdown by siRNA

   The siRNAs for FLNA were used for gene knockdown (Sangon Biotech,
   China). The knockdown was detected using western blotting. The
   antibodies included anti-FLNA (67,133, Proteintech, 1:1000) and
   HRP-conjugated goat anti-mouse IgG (H + L) (SA00001-1, Proteintech,
   1:3000).

Drug sensitivity assays using CCK8 method

   The well-grown organoids were digested with TrypLE Express enzyme for
   15 min and centrifuged at 1000 rpm for 5 min. After cell counting, the
   cell suspension was mixed with matrigel at a ratio of 1:1, and seeded
   in a 96-well plate (6 µl/well, 6000 cells). After matrigel
   polymerization for 30 min at 37℃, complete medium (60 µl/well)
   containing various concentrations of drugs was added (Table [79]S2).
   Traditional gastric cancer cell lines (AGS and NCI-N87) were used as
   controls. The cells were digested by TrypLE Express Enzyme into single
   cell suspension and 60 µl/well cells were seeded in 96-well plates for
   CCK8 analysis. Concentration gradients of drugs (5-Fu, 0, 0.5, 0.75, 1,
   2, 5, and 10 μm), oxaliplatin (OXA, 0, 2, 5, 10, 20, 30, and 40 μm),
   and paclitaxel (PTX, 0, 0.1, 0.5, 1, 2, 5, and 10 nm) were added. After
   5 days incubation at 37℃, a 10% CCK8 solution (CK04, DOJINDO) was
   added, and the matrigel was blown off with tips to make the organoids
   fully exposed to the reagent. After incubation at 37℃ for 2 h, the
   OD450 values were measured using a microplate reader to calculate cell
   vitality and IC50 of drugs.

Hypoxia and normoxia culture

   The well-grown organoids were digested with TrypLE Express enzyme for
   0.5–1 h. The cell precipitate was mixed with matrigel at a 1:1 ratio
   and then seeded in a 96-well plate (6 µl /well, 6000 cells, n = 3).
   After the matrigel polymerization, complete medium (60 µl/well) was
   added and cells were cultured for 72 h in hypoxic (37℃, 94% N[2], 5%
   CO[2,] and 1% O[2]) and normoxic (37℃, 20% O[2,] and 5% CO[2])
   conditions. The cellular vitality of organoids was measured using a
   CCK8 assay.

Lactate assay upon inhibition of glycolysis

   The well-growth organoids were digested with TrypLE Express enzyme for
   0.5–1 h. Organoid cells were seeded in a 96-well plate (6 µl /well,
   6000 cells, n = 3). The chemical 2-deoxy-D-glucose (1 mM, M5140,
   AbMole, China) was added for incubation at 37℃, 20% O[2] and 5% CO[2]
   for 72 h. The culture medium was collected and tested by the lactate
   assay using the Cedex Bio Analyzer (06343759001, Roche, Switzerland)
   following the manufacturer’s instructions.

Tumorigenesis of organoid lines in nude mice

   Organoid cells of SPDO1P or SPDO1LM (100 µl, 5 × 10^5 cells) were
   injected into the armpit of male BALB/c-nu mice (4 weeks of age, 4
   mice/group). After 4 weeks, the mice were sacrificed and the tumors
   were removed. The tumor volume was calculated by: V = π/6 × L (long
   diameter) × W (short diameter). The xenograft tumors (PDO-X) were fixed
   in 4% paraformaldehyde overnight, embedded in paraffin, and stained
   with H&E. This study was approved by the Institutional Animal Care &
   Use Committee of Shanghai Jiao Tong University School of Medicine
   (B-2022-010).

Statistics

   GraphPad Prism 6.0 (GraphPad Prism, RRID: SCR_002798) was used for
   statistical analysis. The Shapro-Wilk test was used for normal test.
   The student’s t-test (homogeneity of variance) or unpaired t-test with
   Welch’s correction (heterogeneity of variance) was used for analysis of
   two group data, respectively. ANOVA or Brown-Forsythe ANOVA was used
   for data analysis for the above two groups. Each experiment was
   repeated at least three times independently. P < 0.05 was considered
   statistically significant.

Results

Clinicopathological characteristics of organoid cell lines

   The cancer tissues were collected from a 68-year-old female patient who
   presented with upper abdominal pain for several months. Under
   endoscopy, an ulcerated mass of 3.5 × 3.5 cm was found at the greater
   curvature of the antrum. By enhanced computed tomography (CT)
   examination, localized thickening of the gastric antrum was found,
   resulting in an unclear three-layer structure of the gastric wall with
   abnormal enhancement in the arterial phase. An enlarged lymph node near
   the right gastroepiploic artery was identified (Fig. [80]1A). After
   radical distal gastrectomy, an ulcerative mass (Borrmann II) and an
   enlarged lymph node were observed (Fig. [81]1B). The pathological
   reports showed a poorly-differentiated adenocarcinoma (intestinal type
   by the Lauren classification) of stage A (T3N2M0) without H. pylori
   infection.

Fig. 1.

   [82]Fig. 1
   [83]Open in a new tab

   Clinicopathological characteristics of organoid lines (A) A thickened
   gastric antrum wall is shown by CT scanning. The red box marks the
   enlarged image. (B) An ulcerated mass is observed at the antrum in the
   gastrectomy sample with the enlarged lymph node near the right
   gastroepiploic artery (T, primary tumor; LN, lymph node). (C) From left
   to right, histology of poorly-differentiated adenocarcinoma of original
   tissues by H&E staining, organoid morphology by light- field
   observation, and organoid morphology by H&E staining. The histology is
   compatible with poorly-differentiated adenocarcinoma with irregular
   lumens. The upper panel indicates SPDO1P and the lower panel indicates
   SPDO1LM. Scale bar, 100 μm. (D) Immunohistochemical staining of
   E-cadherin, N-cadherin, and Ki67 for SPDO1P and SPDO1LM. Scale bar,
   80 μm. (E) SPDO1P and SPDO1LM maintain cell vitality in continuous
   culture in vitro. Both organoid lines expanded for over 50 generations.
   Note: P1 = passage 1, P15 = passage 15, and P50 = passage 50. Scale
   bar, 100 μm

   Organoids were successfully established from the primary cancer
   (SPDO1P) and the metastatic lymph node (SPDO1LM). By H&E staining, both
   organoid lines showed poorly-differentiated adenocarcinoma with cell
   disarrangement, and matched well with their original tissues
   (Fig. [84]1C). Organoid lines grew in nests with lost polarity.
   Irregular lumens were occasionally observed. By immunohistochemistry,
   we examined the expression of E-cadherin, N-cadherin and Ki67
   biomarkers for SPDO1P and SPDO1LM (Fig. [85]1D). E-cadherin was
   expressed on SPDO1P organoids and significantly downregulated on
   SPDO1LM organoids. N-cadherin was weakly expressed on SPDO1P organoids
   and significantly upregulated on SPDO1LM organoids. The cell
   proliferation marker Ki67 was upregulated in SPDO1LM organoids compared
   to the SPDO1P organoids. The cell vitality of SPDO1P and SPDO1LM was
   maintained in vitro for the long-term culture (over 600 days and 50
   generations) and tolerated repeated freezing and recovery
   (Fig. [86]1E).

Culture conditions

   Both SPDO1P and SPDO1LM organoid lines proliferated in 3D culture with
   complete medium (Fig. [87]2A). The doubling times of SPDO1P and SPDO1LM
   were 120 h and 72 h, respectively (Fig. [88]2B). In addition, the
   SPDO1LM organoid line could proliferate in traditional 2D culture.
   During the first 24 h, a portion of cells adhered to the wall in a
   semi-floating state. At 48 h, SPDO1LM cells were completely adhered to
   the wall and could be passaged continuously (Fig. [89]2C). The doubling
   time of SPDO1LM in 2D culture was 96 h (Fig. [90]2D).

Fig. 2.

   [91]Fig. 2
   [92]Open in a new tab

   Culture conditions of SPDO1P and SPDO1LM. (A) The SPDO1P and SPDO1LM
   lines grow well in complete medium in 3D culture (light field). Scale
   bar, 100 μm. (B) The growth curves of SPDO1P (red) and SPDO1LM (blue)
   in 3D culture show that the doubling time of SPDO1P (120 h) is longer
   than that of SPDO1LM (72 h). (C) The SPDO1LM cells can adhere to the
   wall in 2D culture with 10% FBS-DMEM medium (light field). Scale bar,
   100 μm. (D) The growth curve of SPDO1LM in 2D culture with a doubling
   time of 96 h. (E) Comparison of cell vitality of the SPDO1LM line based
   on the stepwise elimination of growth factors or chemicals. (F)
   Comparison of cell vitality of the SPDO1P line based on stepwise
   elimination of growth factors or chemicals. (G) The growth of SPDO1P
   (upper panel) and SPDO1LM (lower panel) cultured in complete medium
   (left), without Y27632 (middle), and 10% FBS-DMEM (right) under
   light-field observation. The SPDO1LM line grows well in complete
   medium, without Y27632, and 10% FBS-DMEM, while the SPDO1P line shows
   lower cell vitality without Y27632 or 10% FBS-DMEM. Scale bar, 200 μm,
   ^nsP > 0.05, ^*P < 0.05, ^***P < 0.001. Abbreviation: DMEM, Dulbecco’s
   modified Eagle medium; FBS, fetal bovine serum

   To clarify the requirement for growth factors or cytokines of SPDO1P
   and SPDO1LM lines, we examined the growth status by a stepwise
   elimination method. SPDO1LM could grow well even in DMEM medium
   supplemented only with 10% FBS and no growth factors or chemicals. The
   cell vitality of SPDO1LM in DMEM medium supplemented with 10% FBS was
   about 60% of that in complete medium (Fig. [93]2E, Table [94]S1, Figure
   [95]S1) However, after the removal of growth factors or chemicals, the
   cell vitality of SPDO1P line was only 30% of that in complete medium
   and was difficult to passage (Fig. [96]2F, Figure [97]S2). These data
   indicate that the SPDO1P line was highly dependent on growth factors
   and chemicals such as R-spondin 1, noggin, EGF, SB202109, and Y27632.
   Among them, Y27632, which is a selective ROCK1 and ROCK2 inhibitor, was
   crucial to SPDO1P growth (Fig. [98]2G). Therefore, the medium
   requirement of the SPDO1P line was as follows: advanced DMEM/F12,
   HEPES, B27, N2 additive, GlutaMAX, nicotinamide, N-acetyl-L-cysteine,
   gastrin, noggin, Y27632, SB202109, EGF, and R-spondin 1 (Table [99]S1).
   We note that both SPDO1P and SPDO1LM lines could grow well without
   WNT/β-catenin signaling pathway activator WNT3a or CHIR-99,021.

STRs and karyotypes analysis

   By STRs analysis, both SPDO1P and SPDO1LM lines did not match the STRs
   of 2455 cell lines in the public database, which suggests that SPDO1P
   and SPDO1LM lines were newly created organoid cell lines. By comparison
   of eight core STRs loci and the sex gene amelogenin between SPDO1P and
   SPDO1LM lines, the above loci were consistent (Fig. [100]3A). Allele
   drift at the D16S539 and CSF1PO loci was observed between the two lines
   (Fig. [101]3B, Figure [102]S3). Overall, the matching rate between
   SPDO1P and SPDO1LM lines was 88.9%, which means that SPDO1P and SPDO1LM
   were homologous derivatives.

Fig. 3.

   [103]Fig. 3
   [104]Open in a new tab

   STRs and karyotypes analysis, as well as tumorigenesis of SPDO1P and
   SPDO1LM organoid lines. (A) The crucial STRs loci are compared between
   SPDO1P and SPDO1LM lines. Both organoid lines show XX sex alleles. STRs
   loci are matched well at D5S818, vWA, D7S820, TH01, D13S317, and
   TPOXSTR. There is allele drift at the loci of D16S539 and CSF1PO. (B)
   The chromatograms of the D5S818 locus (red) and D16S539 locus (green)
   from SPDO1P and SPDO1LM lines. The D5S818 locus is matched well, and
   one allele is shifted at the D16S539 locus. (C) The karyotype of SPDO1P
   line is 46, XX, + 8, -21 (upper), while that of the SPDO1LM line is 48,
   XX, + 8, +12, r(21)(p13q22.3) (lower). (D) The tumor volume of the
   SPDO1LM xenograft is larger than that of the SPDO1P xenograft
   (385.76 ± 112.16 vs. 63.46 ± 10.55 mm^3). (E) By H&E staining, both
   xenografts of SPDO1P and SPDO1LM are poorly-differentiated
   adenocarcinoma. Scale bar, 100 μm.;^**P < 0.01

   The karyotypes of SPDO1P and SPDO1LM lines were 46, XX, + 8, -21 and
   48, XX, + 8, +12, r(21)(p13q22.3), respectively (Fig. [105]3C). The
   number of chromosomes from SPDO1P was 46, including an extra chromosome
   8 and a loss of chromosome 21(Fig. [106]3C, upper; Figure [107]S4). The
   number of chromosomes from SPDO1LM was 48, with extra chromosomes 8 and
   12, and a circular chromosome 21(Fig. [108]3E, lower; Figure [109]S4).

Tumorigenesis of SPDO1P and SPDO1LM lines in vivo

   After inoculation of SPDO1P and SPDO1LM cells into the armpit of nude
   mice, tumorigenesis was observed in 100% of the experimental animals
   (n = 4 per group). The tumor volumes of the SPDO1LM xenograft were
   larger than that of the SPDO1P xenograft (SPDO1LM-X vs. SPDO1P-X,
   385.76 ± 112.16 vs. 63.46 ± 10.55 mm^3; Fig. [110]3D). By H&E staining,
   the histology of the SPDO1P and SPDO1LM xenografts were consistent with
   their original poorly-differentiated adenocarcinoma (Fig. [111]3E).

Genetic features by whole-exome sequencing

   To identify genomic variation and driver mutations of SPDO1P and
   SPDO1LM, WES was performed for organoid lines and matched original
   tissues. Both SPDO1P and SPDO1LM showed genomic instability with copy
   number variation in some chromosomes. Among them, SPDO1P showed a gain
   of chromosome 8 and a loss of chromosome 21 (Fig. [112]4A), consistent
   with the results found in the karyotypes analysis. The SPDO1LM showed a
   gain of chromosome 8 and 12 and a loss of chromosome 21 (Fig. [113]4B),
   consistent with the result found in the karyotypes analysis. Both
   SPDO1P and SPDO1LM lines and their matched tissues showed similar
   proportions of SNPs (Fig. [114]4C).

Fig. 4.

   [115]Fig. 4
   [116]Open in a new tab

   Genetic features of SPDO1P and SPDO1LM by whole-exome sequencing. (A)
   By CNV analysis, the SPDO1P shows a chromosome 8 gain (marked by red)
   and chromosome 21 loss (marked by blue). (B) By CNV analysis, the
   SPDO1LM line shows a chromosome 8 gain (marked by red), chromosome 12
   gain (marked by red), and chromosome 21 loss (marked by blue). Copy
   ratio > 0 indicats a gain; copy ratio < 0 indicates a loss. (C) The bar
   plot of SNP ratios in SPDO1P, SPDO1LM, and their matched original
   tissues. There is no significant difference in the ratio of SNPs
   between SPDO1P and its original tissue (^nsP = 0.8832) or SPDO1LM and
   its original cancer tissue (^nsP = 0.7003) by the Chi-square test. (D)
   By mutation analysis, 25 somatic driver mutations are shared in both
   SPDO1P and SPDO1LM lines, as well as their original cancer tissues.
   There are specific driver mutations in the SPDO1P line and the matched
   cancer tissue (red box) and SPDO1LM line and matched cancer tissue
   (blue box)

   By gene mutation analysis, both SPDO1P and SPDO1LM lines, as well as
   matched original tissues showed wild-type for TP53. A total of 25
   somatic driver mutations were shared in both organoid lines and their
   matched original tissues, including frameshift insertion mutations of
   CIC, missense mutations of SETD1B, LAMA2, IL21R, FAT4, KAT6B, ERBB3,
   CEP89, LRP1B, DLG1, MUC16, FGFR2, MAGI2, SVEP1, SPEG, EPHA1, PAX3,
   PREX2, PTPRT, ZNF132, CACNA1D, and PRPF8, and nonsense mutations of
   SYNE1, LRRK2, and CTCF (Fig. [117]4D). In addition, there were specific
   driver mutations in SPDO1P or SPDO1LM (Fig. [118]4D). Among the 63
   somatic driver mutation genes in the SPDO1LM line, 34 mutated genes
   (53.97%) were the same as that found in the original cancer tissue,
   covering missense mutations of EPHA7, FAT3, EPPK1, and FLNA, a
   frameshift deletion mutation of CDKN1B, and a frameshift insertion
   mutation of KMT2B. Among the 62 somatic driver mutation genes in the
   SPDO1P line, 32 mutated genes (51.61%) were the same as those found in
   the original cancer tissue, including missense mutations of KRAS,
   PIK3C2B, BRAF, AKT1, FAT1, and CLASP2. A missense mutation of ERBB2 was
   identified in the SPDO1P line, but not in its original cancer tissue
   (Table [119]S3). In addition, we compared the gene expression
   correlation of transcriptomic data between SPDO1P or SPDO1LM and the
   original tissues and found the gene expression profiles of SPDO1P and
   SPDO1LM correlated well with their original tissue (r = 0.9079 and
   0.7992, respectively).

ScRNA-Seq and cell functions

   In tSNE mapping, the gene expression patterns of SPDO1P and SPDO1LM
   were different, suggesting significant functional differences between
   the two organoid lines (Fig. [120]5A). Both organoid lines expressed
   epithelial markers KRT8, KRT18 and EPCAM, indicating their epithelial
   origin (Fig. [121]5B). By GSEA pathway enrichment analysis, the highly
   expressed genes of SPDO1P were enriched in the oxidative
   phosphorylation and MYC targets V1 pathways (Fig. [122]5C,Table
   [123]S4), and the highly expressed genes of SPDO1LM were enriched in
   hypoxia, epithelial-mesenchymal transition (EMT), glycolysis,
   apoptosis, coagulation, angiogenesis, IL2-STAT5 signaling, and
   PI3K-AKT-mTOR signaling pathways (Fig. [124]5D, Table [125]S4). We
   compared the gene expression consistency between the scRNA-seq and
   RNA-Seq of the two organoid lines and found that 95% of the top 20
   upregulated genes (19/20) in the scRNA-seq of SPDO1P were highly
   expressed in the RNA-seq analysis. The 90% of the top 20 upregulated
   genes (18/20) in the scRNA-seq of SPDO1LM were also highly expressed by
   RNA-seq (Figure [126]S5).

Fig. 5.

   [127]Fig. 5
   [128]Open in a new tab

   ScRNA-Seq and biological functions assays. (A) The tSNE plot of 10,578
   cells from SPDO1P and SPDO1LM. The highly expressed genes of the SPDO1P
   and SPDO1LM organoid lines are separated into two clusters. (B) The
   violin plots show high expression of KRT8, KRT18, and EPCAM in SPDO1P
   and SPDO1LM. (C) The highly expressed gene set of SPDO1P is enriched in
   oxidative phosphorylation, MYC targets V1, and E2F target pathways. (D)
   The highly expressed gene set of SPDO1LM is enriched in hypoxia, EMT,
   glycolysis, apoptosis, coagulation, angiogenesis, IL2-STAT5 signaling,
   PI3K-AKT-mTOR signaling, protein secretion, and P53 pathways. (E)
   Images of the tubule formation analysis (light field). The supernatant
   of SPDO1LM shows stronger proangiogenic potential than that of SPDO1P
   with longer branch lengths and increased tubular junctions, n = 5. (F)
   Images of the organoid spheres in the 3D invasion assay of SPDO1P and
   SPDO1LM (light field). The SPDO1P line grows stably in a sphere
   structure, but the SPDO1LM line extends pseudopodia around the spheres.
   (G) Images of migration and invasion cells. The SPDO1LM line shows more
   migrated cells than the SPDO1P line after 72 h of incubation and more
   invading cells than SPDO1P after line for 96 h. n = 6. (H) The
   correlation analysis shows that the gene expression levels of the top
   10 genes in the oxidative phosphorylation pathway of SPDO1P and the
   original tissue are consistent (r = 0.7542). (I) The lactate
   concentrations of both organoids significantly declined upon
   2-deoxy-D-glucose (DG, 1 mM, 72 h) treatment. (J) The Δlactate
   concentration (lactate concentration of control group - lactate
   concentration of DG- treated group) is different between the two
   organoids. (K) According to western blot analysis, the protein
   expression levels of p-mTOR, mTOR, p-AKT, and AKT in the SPDO1LM line
   are higher than those of the SPDO1P line. ^*P < 0.05, ^**P < 0.01,
   ^***P < 0.001

   To verify the functional changes of upregulated pathways of SPDO1LM
   line, we examined proangiogenic ability, migration and invasion of the
   two organoid lines. We collected the conditioned media of the SPDO1LM
   organoids and SPDO1P and incubated the human endothelial cell line
   EA.HY926 in the media for 48 h. The conditioned media of SPDO1LM
   promoted angiogenesis with longer branch lengths (6831 ± 927 vs.
   2304 ± 671 μm, ^***P = 0.001) and increased the number of tubular
   junctions (15.20 ± 4.83 vs. 2.40 ± 1.50, ^***P = 0.001), compared to
   that of SPDO1P (Fig. [129]5E). In the 3D invasion assay, the SPDO1P
   organoids grew in a spherical structure, while SPDO1LM organoids
   extended pseudopodia around spheres, indicating stronger invasion
   characteristics (Fig. [130]5F). The migration and invasion abilities of
   the SPDO1P and SPDO1LM lines were identified in 2D cultivation. The
   migrated cells of SPDO1LM were more than that of SPDO1P for 72 h
   (110 ± 18 vs. 16 ± 4, ^***P < 0.0001). The invaded cells of SPDO1LM
   were more than that of SPDO1P at 96 h (105 ± 30 vs. 5 ± 2,
   ^***P < 0.0001)(Fig. [131]5G).

   To verify the difference in the metabolic pathways of SPDO1P and
   SPDO1LM, we cultured the SPDO1P and SPDO1LM in normoxic and hypoxic
   conditions. The expression correlation of the top 10 oxidative
   phosphorylation pathway genes of SPDO1P and the original tissue was
   consistent (Fig. [132]5H, Figure [133]S6A). The hypoxia- and
   EMT-related genes MXI1 and ENO2 were enriched in SPDO1LM, compared with
   SPDO1P (Figure [134]S6B). SPDO1P was more oxygen-dependent than SPDO1LM
   (Figure [135]S6C). The cell proliferation rate in SPDO1P decreased in
   hypoxia, compared with that of normoxic, while the SPDO1LM was not
   (Δorganoid cell vitality SPDO1P vs. SPDO1LM: 102.27 ± 24.26 vs.
   63.53 ± 24.98, ^*P = 0.0107) (Figure [136]S6D). Hexokinase 2 (HK2) is a
   glycolysis-related gene and was highly expressed in SPDO1LM (Figure
   [137]S6E). We examined the lactate contents of the two organoid lines
   upon treatment with 2-deoxy-D-glucose (DG), a hexokinase inhibitor. The
   lactate content of SPDO1LM was significantly higher than that of
   SPDO1P, but significantly decreased upon DG treatment (SPDO1LM:
   366.62 ± 7.93 vs. 160.21 ± 8.05 mg/L; SPDO1P: 297.23 ± 7.76 vs.
   162.32 ± 11.57 mg/L; Fig. [138]5I). Upon inhibition of hexokinase by DG
   treatment, the reduction of lactate content in SPDO1LM was considerably
   different, compared to SPDO1P (Fig. [139]5J). These data indicate that
   the glycolysis of SPDO1LM was greatly inhibited by the hexokinase
   inhibitor.

   Regarding to the highly expressed genes of PI3K-AKT-mTOR signaling
   pathways in SPDO1LM by scRNA-Seq, we verified the protein expression of
   p-mTOR, mTOR, p-AKT, and AKT of SPDO1LM by western blot (Fig. [140]5K).
   In addition, we found that the top 100 upregulated genes intersected
   with EMT pathway genes, including ENO1, LGALS1, and FLNA (Figure
   [141]S7A). Then, the FLNA gene was knocked down by siRNA in SPDO1LM
   (Figure [142]S7B). In 2D migration assay, the FLNA knockdown
   significantly inhibited SPDO1LM migration (Figure [143]S7C).

Drug sensitivity and clinical association

   We compared the CCK8 and GTC 3D assays for drug sensitivity testing in
   3D cultured organoids and found that there was no significant
   difference between the two methods (Figure [144]S8). The improved CCK8
   method showed smaller batch-to-batch variation than the GTC 3D method.
   According to CCK8 assay, the SPDO1LM line was more sensitive than
   SPDO1P to 5-Fu (IC50: 0.1267 ± 0.0053 vs. 0.4125 ± 0.0632 µM,
   ^**P = 0.0031), while the SPDO1P line was more sensitive to OXA than
   SPDO1LM (IC50: 2.7940 ± 0.3993 vs. 5.6873 ± 0.4514 µM, ^**P = 0.0025).
   In addition, both SPDO1P and SPDO1LM lines were sensitive to PTX (IC50:
   1.7040 ± 0.2390 vs. 1.8463 ± 0.2614 nM, ^nsP = 0.6002) (Fig. [145]6A
   and B, and [146]6C). Since the SPDO1LM line can proliferate in both 2D
   and 3D culture conditions, we examined the drug sensitivity of the
   SPDO1LM line in 2D and 3D conditions and found that the IC 50 of
   SPDO1LM-2D and SPDO1LM-3D was 119.70 ± 5.61 nM and 126.67 ± 5.33 nM,
   respectively (P < 0.05, Figure [147]S9). The drug sensitivity of 5-Fu
   was consistent between 2D and 3D culture settings. We also examined the
   drug sensitivity of SPDO1P or SPDO1LM and traditional cancer cells in
   2D and 3D culture conditions. The IC50 of 5-Fu in AGS, NCI-N87-2D, and
   NCI-N87-3D spheroids was 1.0336 ± 0.0858, 1.2673 ± 0.2088, and
   2.3930 ± 0.1200 µM, respectively. The IC50 of OXA in AGS, NCI-N87-2D,
   and NCI-N87-3D spheroids was 25.7033 ± 1.1197, 4.6780 ± 0.3169, and
   7.8373 ± 1.1069 µM, respectively. The IC50 of PTX in AGS, NCI-N87-2D,
   and NCI-N87-3D spheroids was 113.1967 ± 20.3906, 33.0667 ± 1.8236, and
   65.9033 ± 10.8147 nM, respectively. Both SPDO1P and SPDO1LM were
   sensitive to 5-Fu and PTX than traditional cancer cell lines of AGS,
   NCI-N87-2D, and NCI-N87-3D spheroids.

Fig. 6.

   [148]Fig. 6
   [149]Open in a new tab

   Drug sensitivity of SPDO1P and SPDO1LM lines and its effect on 3D cell
   invasion. (A) Left: The morphological changes of SPDO1P (upper) and
   SPDO1LM (lower) upon 5-Fu treatment. Middle: Cell vitality analysis of
   organoids compared to cancer cell lines AGS, NCI-N87-2D, and NCI-N87-3D
   spheroids upon 5-Fu treatment. Right: The bar chart shows IC50 values
   of organoids and cancer cell lines AGS, NCI-N87-2D, and NCI-N87-3D
   spheroids. (B) Left: The morphological changes of SPDO1P (upper) and
   SPDO1LM (lower) upon OXA treatment. Middle: Cell vitality analysis of
   organoids compared to cancer cell lines AGS, NCI-N87-2D, and NCI-N87-3D
   spheroids upon OXA treatment. Right: The bar chart shows IC50 values of
   organoids and cancer cell lines AGS, NCI-N87-2D, and NCI-N87-3D
   spheroids upon OXA treatment. (C) Left: The morphological changes of
   SPDO1P (upper) and SPDO1LM (lower) upon PTX treatment. Middle: Cell
   vitality analysis of organoids compared to cancer cell lines AGS,
   NCI-N87-2D, and NCI-N87-3D spheroids upon PTX treatment. Right: The bar
   chart shows IC50 values of organoids and cancer cell lines AGS,
   NCI-N87-2D, and NCI-N87-3D spheroids upon PTX treatment. n = 3. Scale
   bar, 100 μm. (D) In 3D cell invasion assays, 5-Fu more significantly
   inhibits SPDO1LM invasion in SPDO1LM compared to SPDO1P. (E) OXA
   significantly inhibits SPDO1P invasion compared to SPDO1LM. (F) PTX
   inhibits both SPDO1P and SPDO1LM invasion compared to controls. ^nsP >
   0.05, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001

   We further examined the effect of drug 5-Fu, OXA, and PTX (1µM) on 3D
   invasion in SPDO1P and SPDO1LM lines. By the ratio of the invasion
   area, 5-Fu showed a significant inhibitory effect on 3D cell invasion
   in SPDO1LM (SPDO1P vs. SPDO1LM: 92.55 ± 4.89% vs. 55.38 ± 5.10%,
   ^***P < 0.001) after 24 h of drug treatment (Fig. [150]6D). Drug OXA
   showed a significant inhibitory effect on 3D cell invasion of SPDO1P
   (SPDO1P vs. SPDO1LM: 12.72 ± 6.26% vs. 55.29 ± 24.03%,^**P = 0.003;
   Fig. [151]6E). Drug PTX showed an obvious inhibitory effect on 3D cell
   invasion in both SPDO1P and SPDO1LM lines (SPDO1P vs. SPDO1LM:
   64.41 ± 14.07% vs. 67.64 ± 19.98%) compared to controls (P < 0.05;
   Fig. [152]6F). The drug sensitivity analysis of organoids provided an
   important reference for chemotherapy. The patient received systemic
   chemotherapy of S-1 plus oxaliplatin (SOX regimen: oral S-1, 40 mg/m^2,
   twice a day on days 1–14, and oxaliplatin 130 mg/m^2 intravenously on
   day 1; each treatment cycle was 21 days) for totally six cycles after
   radical surgery. The patient has been stable for 24 months without
   recurrence.

Discussion

   Paired organoid lines from primary gastric cancer (SPDO1P) and
   corresponding metastatic cancer of the lymph node (SPDO1LM) were
   established. The “S” in the name has two meanings: “Shanghai” and
   “Stomach”. PDO means patient-derived organoid. The “1” represents
   number one. The letter “P” means primary tumor, while “LM” represents
   lymphatic metastasis. The two organoid lines have been deposited in the
   China Center for Type Culture Collection (CCTCC) with collection number
   C202387 for SPDO1P and C2023149 for SPDO1LM. Although there are several
   gastric cancer cell lines, such as HGC-27 [[153]31] from lymphatic
   metastasis of poorly-differentiated adenocarcinoma and NCI-N87
   [[154]32] and MKN-45 [[155]33] from liver metastases of gastric cancer,
   there are no reports about paired cell lines from the primary cancer
   and the lymph node metastasis. Thus, this is the first time that such
   matched organoid cell lines have been established. These organoid lines
   will provide valuable experimental models for exploring metastatic
   mechanisms and therapeutic drugs for gastric cancer. More importantly,
   these organoid lines have translational value for drug screening. The
   two organoid lines showed high sensitivity to first-line therapeutic
   drugs and guided the choice of chemotherapy, i.e., the SOX regimen
   [[156]34]. Our bench results reflected the drug response at the bedside
   and provided support for the FDA Modernization Act 2.0, that is,
   organoids can be applied as an alternative to animal experiments for
   drug sensitivity studies [[157]24, [158]35, [159]36].

   Although SPDO1P and SPDO1LM are derived from the same patient, they
   showed different biological characteristics in some aspects
   (Table [160]1). The SPDO1LM line is derived from metastatic cancer of
   the lymph node and showed stronger growth, invasion, and metastatic
   abilities than the SPDO1P line from the primary cancer. The protein
   expression levels of N-cadherin and Ki67 were higher in SPDO1LM than in
   SPDO1P. Their difference in biological characteristics can be explained
   by scRNA-seq analysis. The differentially expressed genes of SPDO1LM
   were mainly enriched in EMT and angiogenesis pathways. Activation of
   EMT and angiogenesis pathways plays important roles in promoting
   metastasis [[161]37, [162]38]. In our study, the SPDO1LM organoid line
   revealed a stronger ability to induce neovascularization, migration,
   and invasion than that of SPDO1P. In the metabolic dependence analysis,
   the highly expressed gene set of the SPDO1P line was enriched in the
   oxidative phosphorylation pathway, which may be related to its
   dependence on oxygen for cell growth. In contrast, the highly expressed
   gene set of SPDO1LM was enriched in hypoxia- and glycolysis-related
   pathways, which likely explains the hypoxia-resistant cell growth.
   Lactate, a glycolytic product, was significantly higher in SPDO1LM than
   in SPDO1P. Recently, Zong and coworkers reported that lactate produced
   by cancer cells can lactylate important tumor suppressor, resulting in
   functional inactivation of proteins and promoting tumor progression
   [[163]39]. The chemical 2-deoxy-D-glucose can inhibit glycolysis by
   targeting HK2 [[164]40], which may be a new direction to inhibit
   glycolysis and prevent tumor progression.

Table 1.

   Biological characteristics of SPDO1P and SPDO1LM organoid lines
   Parameters Characteristics SPDO1P
   (Primary tumor) SPDO1LM
   (Lymph node)
   Growth Culture time 570 days 570 days
   Doubling time 120 h (3D) 72(3D) or 96 (2D) hours
   3D culture Yes Yes
   2D culture No Yes
   Culture medium

   Essential cytokine (Noggin, Y27632, SB202109, EGF, R-Spondin1)

   WNT3a-free

   Essential cytokine or 10% FBS DMEM

   WNT3a-free
   STR Unique Unique
   Karyotype 46, XX, + 8, -21 48, XX, + 8, +12, r(21)(p13q22.3)
   Xenograft 100% 100%
   Genomics CNV 8 gain, 21 loss 8 gain, 12 gain, 21 loss
   TP53 Wild-type Wild-type

   Mutation

   (shared)
   CIC, SETD1B, LAMA2, IL21R, FAT4, KAT6B, ERBB3, CEP89, LRP1B, DLG1,
   MUC16, FGFR2, MAGI2, SVEP1, SPEG, EPHA1, PAX3, PREX2, PTPRT, ZNF132,
   CACNA1D, PRPF8, SYNE1, LRRK2, CTCF CIC, SETD1B, LAMA2, IL21R, FAT4,
   KAT6B, ERBB3, CEP89, LRP1B, DLG1, MUC16, FGFR2, MAGI2, SVEP1, SPEG,
   EPHA1, PAX3, PREX2, PTPRT, ZNF132, CACNA1D, PRPF8, SYNE1, LRRK2, CTCF

   Mutation

   (special)
   KRAS, PIK3C2B, BRAF, AKT1, FAT1, CLASP2 EPHA7, FAT3, EPPK1, FLNA,
   CDKN1B, KMT2B
   ScRNA-Seq Up-regulated pathway Oxidative phosphorylation, MYC targets
   V1 Angiogenesis, epithelial-mesenchymal transition, glycolysis, hypoxia
   Behaviors Migration ability Weak Strong
   Invasion ability Weak Strong
   Angiogenesis Weak Strong
   Drug sensitivity 5-Fu (IC50) 0.4125 μm 0.1267 μm
   OXA (IC50) 2.7940 μm 5.6873 μm
   PTX (IC50) 1.7040 nm 1.8463 nm
   [165]Open in a new tab

   Histologically, both SPDO1P and SPDO1LM retained the morphology of
   poorly-differentiated adenocarcinoma and were consistent with those
   described for the adenocarcinoma organoids established by Seidlitz
   [[166]41]. Cancer organoids often recapitulate the genetic profiles of
   the original tissues [[167]23, [168]35, [169]41]. To date, given the
   high cost of genetic analysis, organoid-related studies have not
   profiled the genetic mutations for each generation of organoids.
   Therefore, the data on genetic changes between organoids and original
   tissues are mainly based on the detection of a specific generation. For
   example, Weeber et al. have examined genetic mutations in metastatic
   colorectal cancer biopsies and early passage organoids (within 2–3
   months) and found that 90% of mutations were shared between the
   organoids and the original tissues [[170]42]. Yan et al. analyzed
   differences in the mutation spectrum between early and late organoids
   (more than 6 months) from colorectal cancer and the original tissues
   and found that the drift of genetic mutation of short-term culture of
   organoids was less than 20%, while the drift of genetic mutation in
   long-term cultured organoids could reach 50% [[171]43]. Our team has
   examined the differences in gene variants between short-term culture
   organoids (less than 3 months) and their original tissues and found
   that the shared mutation rate was over 90% (data unpublished). In the
   current study, we analyzed the mutations in organoids of the eighteenth
   passage and found that the shared mutations between organoids and
   original tissues account for 50–60%. Additional somatic mutations may
   happen during repeated cell passaging in vitro. For example, we
   identified an ERBB2 mutation in the SPDO1P line but did not identify it
   in the original cancer tissue. Although we detected 25 shared mutations
   in both organoid lines and the original cancer tissue, we also found
   some mutations that were in the organoids and the matched cancer
   tissues. The biological significance of the somatic mutations and
   whether they are related to driving cancer metastasis will be studied
   in the future. In addition, the newly established SPDO1P and SPDO1LM
   lines are wild-type for the TP53 gene, and thus they provide an
   important model for molecular biology and drug development for TP53
   wild-type gastric tumors in the future.

   Regarding the requirements for culture medium, both SPDO1P and SPDO1LM
   lines were independent of WNT signaling activators, such as WNT3a,
   indicating that cancer cells are auto-activated for the WNT pathway.
   SPDO1P is highly dependent on growth factors and chemicals, especially
   Y27632, while SPDO1LM can grow well in DMEM medium containing 10% FBS
   only. This may be attributed to cell dissociation during the passaging
   procedure, which disrupts the cell-cell adhesion and induces cell
   apoptosis. SPDO1P is dependent on cell-cell adhesion, while SPDO1LM is
   independent of cell-cell adhesion. The chemical Y27632 inhibits
   apoptosis caused by cell dissociation and maintains the continued
   growth of SPDO1P.

   STRs are the main indicator for discriminating cell homology or
   contamination of human cell lines [[172]30]. The concordance rate of
   75% is generally used as the cut-off value [[173]30, [174]44]. In STRs
   analysis, the STRs loci of SPDO1P and SPDO1LM lines do not match with
   2455 human cell lines in public cell databases, indicating their
   uniqueness as cell lines. The concordance rate of STRs between SPDO1P
   and SPDO1LM lines reached 88.9%, implying that SPDO1P and SPDO1LM lines
   are homologous. The appearance of allele drift in some loci could be
   explained by the variation of short tandem repetitive sequences along
   with cancer progression [[175]30]. Since SPDO1LM was constructed from
   the metastatic tumor of the lymph node, the chromosomal instability may
   continue along with the evolution of cancer metastasis [[176]45]. In
   addition, the spatial heterogeneity in cancer progression might result
   in a non-uniform distribution of genetic variants of different
   subpopulations of cancer, such as primary tumors and metastases
   [[177]46, [178]47]. Therefore, we recommend establishing organoids from
   primary cancer lesion, as well as metastatic cancer lesions, if
   possible, to get the real drug sensitivity for metastatic cancer.
   Karyotypes analysis of chromosomes is another indicator for cell line
   construction. We found that both SPDO1P and SPDO1LM are aneuploid,
   which is supported by the copy number analysis. The genomic instability
   of SPDO1P and SPDO1LM is in accordance with that of intestinal-type
   gastric cancer [[179]41, [180]48].

   Regarding organoid technology as part of routine healthcare, it is
   feasible, because the time of culture is short (10–15 days). If the
   drug screening is subsequently carried out, it will take a total of 30
   days. The cost of organoid culture is cheaper than that of traditional
   PDX models [[181]49]. Clinically, chemotherapy is usually started one
   month after surgery. The results of drug sensitivity screening of
   patient-derived organoids may be very helpful. To decrease the cost of
   organoid cultivation and drug screening, artificial assistant
   evaluation could be used, as in our previous work [[182]27].

Conclusion

   Paired organoid cell lines from primary and metastatic cancer of the
   lymph node were established for the first time from gastric tumors. The
   two organoid lines were deposited in the Chinese Typical Culture
   Collection Center, and form part of the living biobank at the
   institution. The cell biology, genetics, and drug sensitivity of the
   paired organoid lines were characterized, which has the potential to
   inform clinical decision-making and guide the adjuvant postoperative
   chemotherapy. These organoid lines constitute valuable models for the
   further investigation of metastatic mechanisms and new anti-cancer
   drugs.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [183]Supplementary Material 1^ (1.4MB, pdf)

Author contributions

   R, Y, Y. Qi, and W. K: study conception and design. Y. D, R.Y, L. Z,
   X.Y, C. L, Z. Z, X. Z, H. G: data acquisition, and data analysis and
   interpretation. I. H. C, Z. K, and Y.Y: manuscript writing and final
   review.

Funding

   This work was supported by grants from the Shanghai Science and
   Technology Committee (Grant No. 20DZ2201900), the National Natural
   Science Foundation of China (Grant No. 82072602, 82372933 and
   82173222), the Innovation Foundation of Translational Medicine of
   Shanghai Jiao Tong University School of Medicine (Grant No.TM202001),
   and the Collaborative Innovation Center for Clinical and Translational
   Science by Chinese Ministry of Education & Shanghai Municipal
   Government (Grant No.CCTS-2022202 and CCTS-202302). The funders had no
   role in study design, data collection and analysis, decision to
   publish, or preparation of the manuscript. We thank LetPub
   ([184]www.letpub.com.cn) for its linguistic assistance during the
   preparation of this manuscript.

Data availability

   The raw data of this study (PRJNA1015608) was uploaded into SRA
   database

Declarations

Disclaimer

   Where authors are identified as personnel of the International Agency
   for Research on Cancer/WHO, the authors alone are responsible for the
   views expressed in this article and they do not necessarily represent
   the decisions, policy or views of the International Agency for Research
   on Cancer/WHO.

Conflict of interest

   The authors declare no competing interests.

Footnotes

   Publisher’s Note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   Ruixin Yang, Yao Qi and Wingyan Kwan contributed equally to this work.

Contributor Information

   Zisis Kozlakidis, Email: kozlakidisz@iarc.fr.

   Yingyan Yu, Email: yingan3y@sjtu.edu.cn.

References