Aberrant Glycosylation in Pancreatic Ductal Adenocarcinoma 3D Organoids Is Mediated by KRAS Mutations

Aberrant glycosylation in tumor cells is a hallmark during carcinogenesis. KRAS gene mutations are the most well-known oncogenic abnormalities but their association with glycan alterations in pancreatic ductal adenocarcinoma (PDAC) is largely unknown. We employed patient-derived 3D organoids to culture pure live PDAC cells, excluding contamination by fibroblasts and immune cells, to gasp the comprehensive cancer cell surface glycan expression profile using lectin microarray and transcriptomic analyses. Surgical specimens from 24 PDAC patients were digested and embedded into a 3D culture system. Surface-bound glycans of 3D organoids were analyzed by high-density, 96-lectin microarrays. KRAS mutation status and expression of various glycosyltransferases were analyzed by RNA-seq. We successfully established 16 3D organoids: 14 PDAC, 1 intraductal papillary mucinous neoplasm (IPMN), and 1 normal pancreatic duct. KRAS was mutated in 13 (7 G12V, 5 G12D, 1 Q61L) and wild in 3 organoids (1 normal duct, 1 IPMN, 1 PDAC). Lectin reactivity of AAL (Aleuria aurantia) and AOL (Aspergillus oryzae) with binding activity to α1-3 fucose was higher in organoids with KRAS mutants than those with KRAS wild-type. FUT6 (α1-3fucosyltransferase 6) and FUT3 (α1-3/4 fucosyltransferase 3) expression was also higher in KRAS mutants than wild-type. Meanwhile, mannose-binding lectin (rRSL [Ralstonia solanacearum] and rBC2LA [Burkholderia cenocepacia]) signals were higher while those of galactose-binding lectins (rGal3C and rCGL2) were lower in the KRAS mutants. We demonstrated here that PDAC 3D-cultured organoids with KRAS mutations were dominantly covered in increased fucosylated glycans, pointing towards novel treatment targets and/or tumor markers.


Introduction
To discover new frontiers in the fght against relentless pancreatic ductal adenocarcinoma (PDAC), it is necessary to explore targets beyond DNA, RNA, and proteins, which have been well studied to date [1].For this reason, glycans, the third life chain after nucleic acids and proteins, are an important post-translational modifcation and present new opportunities for targeting [2].Moreover, as glycans constitute the outermost layer of cells, diverse alterations during carcinogenesis afect the multiple processes of invasion and metastasis, making glycosylation a keystone mechanism in cancer progression [1,3].While several glycosylation changes have been reported in PDAC [4], the mechanism of glycosylation change in PDAC remains largely unknown.
In the multistep genesis and progression of PDAC, four major driver genes (KRAS, TP53, SMAD4, and CDKN2A) have been well highlighted [5].Among these, the KRAS mutation is the most frequent genetic mutation and is present in approximately 90% of PDAC cases while also occurring in the earliest stages of low-grade PanIN IA lesions [6][7][8].KRAS activation has been reported to function as a master regulator, afecting the properties of multiple tumor microenvironment components and promoting cancer progression by mobilizing cells such as fbroblasts, macrophages, neutrophils, and lymphocytes [7,9,10].We considered that profling of the aberrant glycosylation associated with KRAS mutations could be useful for elucidating the mechanisms of PDAC progression as well as for developing early diagnostic markers and novel anticancer therapeutic targets.However, any associations between KRAS mutations and glycans in PDAC remain unknown.
A signature challenge in glycan analysis of clinical samples is the difculty of eliminating the contaminating infuence of noncancer components such as stroma and immune cells [11][12][13].Obtaining isolated epithelial samples of PDAC for glycan analysis, without high ratios of these noncancer contaminants, is particularly difcult.Furthermore, simultaneous analysis of glycans, genes, and transcriptomes is limited due to the small number of cells available from clinical specimens.To solve these problems, we focused on three-dimensional (3D) organoid technology to generate PDAC samples that preserve only epithelial cancerous cells [14].Tese 3D organoids allow for specifc epithelial amplifcation by serving as a biologically relevant structure with polarized cells that recapitulate tumor morphology and biology [15,16].
In this study, we established patient-derived organoids from PDAC surgical specimens and analyzed the glycan status in these organoid-derived tumor cells.We analyzed the relationship between altered glycans and glycan-related genes, with a particular focus on the status of KRAS mutations.

Patient Information.
Te establishment of organoid libraries was approved by the Institutional Review Board of the University of Tsukuba Hospital, Japan (IRV code: H28-90) and informed consent was obtained from all PDAC patients.Surgical specimens were obtained from a total of 24 patients who underwent surgery for PDAC from April 2020 to January 2021.Te obtained samples included 24 PDACs, 4 intraductal papillary mucinous neoplasms (IPMNs), and 4 nocancerous pancreatic ducts (Figure 1(b)).

Membrane Protein Isolation and Quantifcation.
Hydrophobic fractions containing membrane proteins were obtained using the Sigma-Aldrich CelLytic MEM Protein Extraction Kit according to the manufacturer's instructions.Protein concentrations were measured using a micro-BCA assay kit (cat#23235, Termo Fisher Scientifc).Proteins were then fuorescently labeled with the monoreactive dye Cy3 (cat#16968983, GE Healthcare, Boston, MA, USA).

RNA Extraction.
Organoid RNA was extracted using the RNeasy Plus Mini Kit (Qiagen, cat# 74134).Based on the RNA Integrity Number (RIN) value obtained from TapeStation 4150, RNA quality was assessed.In this investigation, only specimens with an RIN >8.0 were employed.

Establishment of a Pancreatic Ductal Adenocarcinoma (PDAC) Organoid
Library.Among a total of 24 attempted PDAC 3D organoids (see Table S1 for detailed characteristics of patients), 19 were recognized as established organoids since they could successfully passage more than 5 times.Likewise, 1 of 4 IPMN and 1 of 4 normal main pancreatic duct organoids were established.Among those 21 established organoids, sufcient amounts of protein and RNAs for analysis were extracted from 14 PDACs, 1 IPMN, and 1 normal PDAC.Pairings of a PDAC organoid with a non-PDAC counterpart were available from 2 patients; one was a PDAC-normal MPD pair and the other was PDAC-IPMN pair, eliminating the infuence of individual variations in glycan expression status for those pairs (Figures 1(a) and 1(b)).
To confrm that the KRAS mutations existed in our PDAC organoids, EGF selection pressure was applied.EGF is an important niche factor of the KRAS pathway, as KRAS mutants do not require EGF to survive [17].As expected, KRAS wild-type organoids were incapable of long-term culture without EGF in the medium while all KRAS mutant organoids survived without EGF (Figure S1).
Although it is widely known that sialic acid is involved in carcinogenesis [1,24], there were few signifcant changes in the sialic acid-binding lectins in this study (Figure 3(a)).

Comparison of Glycosylation-Related Gene Expression Based on RNA-Seq between KRAS Mutant and Wild-Type
Organoids.To address which enzymes are responsible for glycan changes in KRAS mutant organoids, we performed RNA-seq and focused on glycan-related genes.Such genes, including glycosyltransferases and hydrolases, were extracted from the GlyCosmos Portal database (https:// glycosmos.org/)and 212 gene expression instances were analyzed.PCA analysis showed that the KRAS wild-type organoids tended to have similar expression profles (Figure 4(a)) while several glycan-related genes were signifcantly altered (p < 0.05) between KRAS mutant and wild-type organoids (Figure 4(b)).Among them, the fucosyltransferase genes FUT6 and FUT3 were signifcantly elevated in the KRAS mutant group (Figures 4(c), 4(d)).
We also compared the gene expression of glycosyltransferases involved in the synthesis of mannosylated, galactosylated, and sialylated glycans (Figure S3), fnding that, in terms of site-specifc mutation diferences, FUT6 in G12V and G12D was signifcantly upregulated versus the KRAS wild-type (Figure S2(b)).

Comparison of KRAS Mutant
Wild-Type Organoids from the Same Patients.Tese lectins and glycosyltransferases changes were also similarly compared in organoids derived from the same patients (Figure 5).In addition, lectin microarray and RNA-seq data were also compared between normal pancreatic ducts/cancer (Case 1) and IPMN/cancer (Case 2).In both cases, the reactivity of AAL and rAAL was signifcantly elevated in the mutants, with FUT6 and FUT3 elevated in both cases.Te binding intensity of the mannosebinding lectins (rRSL, rBC2LA, and rPAIIL) and galactosebinding lectins (rGal3C, rCGL2), plus expression of their respective responsive genes, were also compared but only inconsistent results were obtained (data not shown).

Discussion
A comprehensive analysis of glycan expression in 16 organoids derived from 14 pancreatic cancer patients showed that the reactivity of fucose-binding lectins (AAL, AOL) was higher in 13 KRAS mutant organoids compared to 3 wild-type organoids.Furthermore, we were able to analyze wild-type versus mutant characteristics in samples from the same patients, precluding arguments that difering fucosylation may be due to individual glycosylation statuses independent of KRAS [25].In these 2 unique cases, elevated fucosylation was found in only the KRAS mutants, supporting our hypothesis that the cell surface glycans of KRAS mutant PDAC organoids are hyper-fucosylated.
Hyper-fucosylation is reported to associate with diverse functions, including regulation of infammatory responses, signal transduction, cell proliferation, transcription, and adhesion [26].Such machinery may also promote higher expression of abnormal fucosylation during carcinogenesis and tumor progression [27,28].As for the diagnostic detection of altered fucosylation, the CA-19-9 (sialyl lewis a) 4 Journal of Oncology antigen, is a frequently used biomarker for PDAC and emphasizes the role of this posttranslational modifcation on cancer adaptation and progression [4,29,30].
As the responsible gene for fucosylation, fucosyltransfereases such as FUT6 mediate α1-3 fucosylation while FUT3 mediates α1-3/4 fucosylation [26], playing a role as key enzymes for sialyl-Lewis X and/or CA19-9 generation [1,[31][32][33][34].Our data demonstrated that KRAS mutants were hyper-α1-3/4 fucosylated and, at the same time, FUT6 and FUT3 RNA expression was actually elevated in our KRAS mutant organoids.Although we did not obtain direct mechanistic evidence for this hyper-fucosylation regarding KRAS, a study by Kakuma et al. indicates that EGF may be important [35].Tey reported that, when EGF was used in colon cancer cell lines, glycosyltransferase, including FUT3, FUT6, and ST3GAL1/3/4, was elevated.Since EGF is the upstream molecule of the RAS cascade, the administration of EGF is considered synonymous with RAS mutation through activation of the RAS signaling pathway.In addition, another report demonstrated that RAS oncogenes upregulate glycosyltransferases, namely, MGAT5, via Ets-1 gene expression [36], and RAS oncogenes also upregulate ST6GAL1 expression via RalGEF signaling [37].Tose studies support our idea that KRAS mutations are associated with increased expression of FUT6/3, resulting in increased α1-3/4-fucosylation in PDAC.Based on those reports, we regard it as reasonable that hyper-fucosylation of PDAC surface glycans may be mediated by KRAS mutation.
We also found that reactivity to mannose-binding lectins was increased and reactivity to galactose-binding lectins was decreased in KRAS mutant organoids.Mannose glycans that bind rRSL and rBC2LCA lectins have been shown to have anticancer and anti-infammatory efects by promoting TGF-β activation and Treg cell induction [38,39].In other types of cancer, a high abundance of mannose glycans has been reported in the liver, ovaries, breasts, and lungs, but their presence in PDAC is largely unknown, and further studies are needed [40][41][42][43].
Another fnding was that galactose-binding lectins, such as rGal3C and rCGL2, were decreased in KRAS mutant organoids.Tese galactose-binding lectins are soluble immunomodulatory proteins that have roles in intercellular adhesion, cell-matrix adhesion, and the modulation of cell surface receptor signaling efciency [44].Furthermore, as the carbohydrate-binding specifcity of each galactosebinding lectin is intricately governed by sulfation, sialylation, fucosylation, and repeating N-acetyllactosamine units [45], the increased galectin binding to KRAS wild-type organoids that we observed may rely on these complex mechanisms and therefore warrant further investigation.
We further investigated whether the site of the KRAS mutation afects fucose glycans by comparing the KRAS G12V mutation group (7 cases) with the KRASG12D mutation group (5 cases).Both AAL and rAAL binding were signifcantly increased in the KRAS G12V mutant group but there were no signifcant diferences in FUT6 expression

6
Journal of Oncology between the two groups.As diferent KRAS subtypes are known to cause various changes in signal activation [46], diferent glycans may be expressed depending on the KRAS mutation site.Further investigation is needed to understand the molecular mechanism of glycan changes caused by KRAS mutation sites.We demonstrated the importance of the reactivity of AAL and AOL lectins to PDAC.Since AAL lectin has been shown to have a higher binding afnity to the surface of pancreatic cancer cells than the CA19-9 antibody, it may be possible to use those lectins as a novel therapeutic carrier and diagnostic tool for pancreatic cancer [47][48][49].Another interesting plan derived from our results is to analyze the relationship between KRAS mutations and serum CA19-9, as it is well reported that CA19-9 does not substantially increase in proportion within a PDAC patient's serum even with positive Lewis antigen [50].If our interpretation that the KRAS mutation causes an elevation in FUT3/6, in turn raising CA-19-9, is correct, then the PDAC tumors from serum CA19-9-negative patients may also be KRAS mutation-negative.However, it should be noted that, in Lewis-negative patients, FUT3 is defective, and thus CA19-9 expression would not be observed regardless of KRAS mutation status.
Two of the 14 PDAC organoids were established from patients who received preoperative chemotherapy.As it is commonly accepted that chemotherapies tend to be more efective against proliferative subpopulation which may bias the established organoid lines toward the more quiescence subpopulation that may display a very distinct phenotype with regard to FUT3/6 expression and lectin reactivity; we repeated the analyses, segregating samples from patients receiving prior neoadjuvant therapy from those that did not.FUT3/6 mRNA levels were not signifcantly diferent between the preoperative chemotherapy and no treatment groups, and both were higher compared to the KRAS wildtype group.Te gene expression data suggested that neoadjuvant therapy did not impact the phenotype of the established organoid lines (Figure S2(c)).
However, AAL, rAAL, AOL, and rAOL lectin intensities were lower in the preoperative chemotherapy group than in the no-treatment group and were not signifcantly diferent from those in the KRAS wild-type group, suggesting that neoadjuvant therapy may preferentially deplete the population that displayed high reactivity toward AAL and AOL despite retaining higher expression of FUT3/6 mRNA.Although our bulk data cannot discern whether there is a spectrum of AAL and AOL reactivity among the KRASmutated FUT3/6 high cells and the small sample size of this study precludes a robust conclusion from being drawn, this observation warrants further studies.For one, it may lead to the understanding of the link between AAL and AOL reactivity and sensitivity to chemotherapy.Moreover, since  Journal of Oncology these cells retain higher FUT3/6 expression, it may be possible to devise a combinatorial therapy that frst pushes the KRAS mutated, high FUT3/6 expression to the AAL and AOL high phenotypic states that are more sensitive toward chemotherapy.
Reports focusing on glycan conformational changes associated with KRAS mutations are scarce in pancreatic cancer, and this is the frst report using organoids to analyze glycan changes associated with KRAS mutations.Te identifcation of lectins that bind predominantly to KRAS mutant organoids compared to KRAS wild-type organoids may contribute to the development of markers that recognize KRAS mutant pancreatic cancer cells and cancer proteins for CTC and hematological diagnosis.

Journal of Oncology
Furthermore, the high afnity of lectins for binding specifc targets on the outermost layers of cells underscores their potential as drug carriers [51].Te AAL and AOL lectins in this study have the potential to selectively avoid KRAS wild-type binding, facilitating the development of specifc drug carriers that may suppress of-target cell damage.
We must acknowledge two main limitations of this study.First, organoids are known to resemble primary tumors in genetic and protein expression, a fact also exploited in obstetric medicine.However, since stromal and infammatory cells undergo complex interactions with glycan structures in tumors, the present results may not be completely congruent with clinical Terefore, to further validate the results of this study for clinical practice, singlecell analyses of the relationship between KRAS mutations, glycosyltransferases, and lectins will be necessary in the future.
Second, in this study, we focused only on the identifcation of glycosyltransferases and corresponding lectins afected by KRAS mutations.Further functional evaluation of glycosyltransferases in wild-type and KRAS mutant organoids and cell lines will require overexpression/ knock-down cloning of glycosyltransferases and their functional evaluation.

Conclusion
Taken together, our data assert that KRAS mutations in PDAC increase FUT6/3 expression and may contribute to the increased reactivity of the fucose-binding lectins AAL and AOL.Tese results support the potential utility of AAL lectin as a diagnostic marker for PDAC and as a potential target for the development of novel anticancer drugs, suggesting an association between KRAS mutations and fucosylation.

Figure 3 :
Figure 3: Comparison of glycan expression based on lectin microarray results between KRAS mutant and KRAS wild-type organoids.(a) List with t values of lectins signifcantly changed between KRAS mutation positive organoids and KRAS wild organoids.Labels and bars at the top of the fgure indicate the type of glycan structure to which each lectin binds.Fucose (Fuc); red bar, Mannose (Man); green bar, Galactose (Gal); orange bar, sialic acid; purple bar, O-glycan; gray bar.Statistically signifcant diferences are calculated by unpaired student's t test and the lectins with p < 0.05 are selected.Lectins are categorized based on their binding specifcities.(b) Te volcano plot of diferential intensity of lectins in KRAS mutation organoids and KRAS wild organoids represents fold change and p values of the selected lectins on x-and y-axis, respectively.(Lectins listed in Figure 3(a)) (c) Lists of representative lectins that are listed by type in order of p value.

Figure 4 :
Figure 4: Comparison of glycosylation-related genes expression based on the result of RNA-seq between KRAS mutant and KRAS wild-type organoids.(a) Te principal component analysis of gene expression data (glycosyltransferase n � 212) for the 16 organoids.Red and blue dots and circles are indicated to KRAS mutation positive organoids and KRAS wild organoids, respectively.(b) Te volcano plot of mRNA diferential expressions in KRAS mutation organoids and KRAS wild organoids represents fold change and p values on x-and y-axis, respectively.Te vertical red and blue lines represent a fold-change cutof of ≥2.0.(c) List with t values of fucosyltransferase between KRAS mutation positive organoids and KRAS wild organoids.Statistically signifcant diferences are calculated by unpaired student's t test and p < 0.05 are selected.(d) Te FUT6 mRNA expression from RNA seq, and the list of details of results.* p < 0.001.