Abstract

   Liver metastasis is the primary factor contributing to unfavorable
   prognosis in colorectal cancer (CRC). Although N‐glycosylation is
   implicated in metastasis, there is a notable paucity of comprehensive
   studies addressing the N‐glycosylation proteomics associated with liver
   metastasis in CRC. In this study, N‐glycosylated proteins and
   N‐glycosylation sites of differential expression between primary
   lesions and paired liver metastatic lesions are identified. Cathepsin D
   (CTSD) is further screened as a potentially pivotal N‐glycosylated
   protein in CRC liver metastasis. Glycosyltransferases complex DDOST and
   STT3B can regulate N‐glycosylation modification at residue 263 of CTSD
   (a protease), thereby affecting CTSD protease to lyse ACADM. ACADM can
   regulate ferroptosis‐related proteins (ACSL4, SLC7A11, and GPX4) to
   further influence the invasion and metastasis of CRC cells. This newly
   discovered mechanism provides potential therapeutic targets for CRC
   treatment and insights for controlling CRC progression and metastasis.

   Keywords: cathepsin D, colorectal cancer, liver metastasis, N‐
   glycosylation modification
     __________________________________________________________________

   The glycosyltransferase complex, comprising DDOST and STT3B, modulates
   the N‐glycosylation modification at residue 263 of cathepsin D (CTSD).
   The extent of N‐glycosylation at this specific residue influences
   CTSD's enzymatic capacity to hydrolyze acyl‐CoA dehydrogenase, medium
   chain (ACADM), thereby facilitating the liver metastasis of colorectal
   cancer.

   graphic file with name ADVS-12-2411740-g004.jpg

1. Introduction

   Colorectal cancer (CRC) ranks among the top 5 deadliest cancers
   globally.^[ [52]^1 ^] In China, CRC has emerged as the second most
   common malignant cancer, surpassing rates in many developed countries
   and presenting a concerning trend.^[ [53]^2 ^] Despite advancements in
   the understanding, diagnosis, and treatment of CRC in recent years, the
   prognosis of patients with CRC remains unfavorable. Metastasis,
   particularly hepatic metastasis, is the primary factor contributing to
   the poor prognosis in these patients.^[ [54]^3 ^] 25% of patients with
   CRC have liver metastases at initial diagnosis and 15%–25% experience
   recurrence resulting from liver metastases after surgery.^[ [55]^2 ^]
   Furthermore, 90% of those with liver metastases cannot access effective
   treatment, resulting in a low 5‐year survival rate.^[ [56]^2 ^] For
   patients in IV stage (TNM), the 5‐year survival rate is just 24%.^[
   [57]^4 ^] Therefore, elucidation of the mechanisms underlying liver
   metastasis in CRC is crucial for improving patient prognosis.

   Glycosylation occurs in over 50% of natural proteins^[ [58]^5 ^] and
   plays key roles in protein folding, quality control, stability,
   transport, and function of proteins.^[ [59]^6 ^] Recent research has
   demonstrated that N‐glycosylation of glycoproteins plays a significant
   role in the metastasis of malignant tumors, promoting the disruption of
   cell adhesion and neovascularization as well as the activation of
   signal transduction pathways.^[ [60]^7 ^] The stability of the
   interferon‐gamma receptor protein was found to be reduced following
   N‐glycosylation modification, resulting in inhibition of the
   interferon‐gamma R signaling, thereby promoting immune evasion of CRC
   cells.^[ [61]^8 ^] A study demonstrated that the absence of
   N‐glycosylation modification at the Asn150 or Asn261 sites of the
   neuropilin‐1 protein can facilitate the invasion and metastasis of CRC
   cells.^[ [62]^9 ^]

   However, the aforementioned studies were scattered and single studies,
   which could not elucidate the key N‐glycosylated proteins involved in
   liver metastasis of CRC. Studies examining N‐glycosylated proteomics in
   CRC are limited. Research in this area remains in its nascent stages,
   particularly with regard to the differences in N‐glycosylated proteins
   between primary lesions and paired liver metastatic lesions.
   Elucidation of the N‐glycosylated proteomic profiles of primary and
   paired liver metastatic lesions will enhance our understanding of the
   molecular mechanisms underlying liver metastasis in CRC. This research
   is expected to provide substantial theoretical and practical value in
   identifying novel therapeutic targets for the treatment of liver
   metastases.

   In this study, we aimed to analyze N‐glycosylated proteomic profiles in
   primary and paired liver metastatic lesions from CRC patients and to
   identify key factors involved in regulating metastases through
   N‐glycosylation modifications.

2. Results

2.1. N‐glycosylated Protein Profiling of Primary and Paired Liver Metastases
Lesions from Patients with CRC

   To elucidate the molecular alterations of N‐glycosylated proteins
   involved in CRC metastasis, we compared the differences in
   N‐glycosylation modification between primary (n  = 14) and paired liver
   metastatic (n  =  14) lesions using mass spectrometry. We detected 139
   N‐glycosylated proteins, 185 N‐glycosylation modification sites, 490
   N‐glycopeptides with entire structures, and 71 glycosyls (Table [63]S1,
   Supplementary Table). Among these sites, 46 N‐glycosylation
   modification sites had not been reported by the Uniprot platform (Table
   [64]S2, Supplementary Table). Compared with primary lesions, there were
   13 sites in paired liver metastatic lesions that exhibited changes in
   N‐glycosylation modification levels exceeding 1.5‐fold (Table [65]S3,
   Supplementary Table). Figure [66] 1A illustrates conserved amino acid
   sequences surrounding all N‐glycosylation sites. A total of 44.94% of
   N‐glycosylated sites exhibited modification by a singular, fixed
   glycosyl (Figure [67]1B), and 20.86% of N‐glycoproteins contained
   multiple N‐glycosylation modification sites (Figure [68]1C).

Figure 1.

   Figure 1
   [69]Open in a new tab

   N‐glycosylated protein profiles of primary lesions and paired liver
   metastatic lesions. A) Conserved amino acid sequences surrounding all
   N‐glycosylation sites (N‐X‐S/T, X≠P). Specifically, the Asparagine (N)
   motif exhibits a relatively conserved nature; the terminal amino acid
   of Asparagine (N) can be any amino acid (X) with the exception of
   Proline (P); the penultimate residues are predominantly Serine (S) and
   Threonine (T). B) The ratio of the different numbers of glycosyls for
   each site. C) The proportion of N‐glycoproteins with the different
   numbers of N‐glycosylated modification sites. D−F) Gene Ontology
   enrichment analysis of cellular component, biological process analysis,
   and molecular function for these N‐glycosylated proteins (Fisher's
   exact test, p < .05). G) Kyoto Encyclopedia of Genes and Genomes
   enrichment analysis of potential pathways for these N‐glycosylated
   modification proteins (Fisher's exact test, p < .05). H) Volcano plot
   shows N‐glycosylation sites of differential expression. 2488.06545
   (Precursor MH) exhibits the highest |fold change| of modification in
   liver metastatic lesions compared with primary lesions. I) The relative
   levels of various glycosyls in primary lesions and paired liver
   metastatic lesions.

   Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes analyses
   (KEGG) were used to investigate the potential functions of these
   N‐glycosylated proteins. Cellular Component analysis indicated that
   these N‐glycosylated proteins may belong to secretory and membrane
   proteins (Figure [70]1D) and were predominantly engaged in biological
   processes such as intercellular adhesion, proteolysis, and immune
   responses (Figure [71]1E), and primary molecular functions associated
   with binding activities (Figure [72]1F). KEGG analysis identified
   possible pathways related to N‐glycosylated proteins, such as lysosome,
   extracellular matrix‐receptor interaction, and drug‐receptor
   (Figure [73]1G).

   2488.06545 (PrecursorMH) was found to be a structural modification of
   the H(6)N(2) glycosyl at residue 263 of Cathepsin D (CTSD) and it
   exhibited the highest |fold change| of modification in paired liver
   metastatic lesions relative to primary lesions (Figure [74]1H).
   N‐glycosylation modification levels of H(5)N(2) and H(7)N(2) were also
   significantly increased in paired liver metastatic lesions relative to
   primary lesions (Figure [75]1I). Consequently, paired liver metastatic
   lesions had higher N‐glycosylation modification (at residue 263 of
   CTSD) levels than primary lesions.

   N‐glycosylation modification levels were quantified by determining the
   molecular weights of CTSD. To confirm the presence of N‐glycosylation
   modifications on CTSD, we initially evaluated the glycosylation levels
   of CTSD in 5 patients with CRC. Paired liver metastatic lesions from
   patients with CRC had higher levels of N‐glycosylation modification on
   CTSD than primary lesions (Figure [76] 2A). In addition, we also
   evaluated the glycosylation levels of CTSD in CRC cell lines
   (Figure [77]2B). PNGase F is an amidase enzyme that cleaves N‐linked
   oligosaccharide chains from glycoproteins. Tunicamycin is a mixture of
   homologous nucleoside antibiotics that inhibit N‐linked glycosylation
   and block GlcNAc phosphotransferase (GPT). Following treatment with
   PNGase F (Yi Sheng, China) and Tunicamycin (MCE, USA), molecular
   weights of CTSD were found to decrease, indicating that CTSD had
   transitioned from a glycosylated form to a non‐glycosylated form
   (Figure [78]2C,D). These results indicated that CTSD exhibited
   N‐glycosylation modification, and paired liver metastatic lesions from
   patients with CRC demonstrated higher N‐glycosylation levels of CTSD
   than primary lesions.

Figure 2.

   Figure 2
   [79]Open in a new tab

   Verification of N‐glycosylated modification of CTSD. A) N‐glycosylation
   levels of CTSD in primary lesions and paired liver metastatic lesions
   from 5 patients with CRC. B) N‐glycosylation levels of CTSD in CRC cell
   lines. C, D) N‐glycosylation levels of CTSD in CRC cell lines following
   treatment with PNGase F and tunicamycin. E) Conserved sequences at the
   Asparagine residue 263 (N263) of CTSD for 5 species (Homo sapiens, Mus
   musculus, Eublepharis macularius, Bombina bombina, Anguilla anguilla).
   F) Asparagine was substituted with Glutamine (Asn → Gln) at residues
   134 (N134Q) and 263 (N263Q) of CTSD. (G) CTSD was knocked down in SW480
   and SW620 cell lines. H) Comparison of molecular weight of CTSD pre‐
   and post‐PNGase F treatment.

2.2. Construction of Mutants for N‐glycosylation Modification Sites of CTSD

   We downloaded CTSD RNA sequences for 5 species (Homo sapiens, Mus
   musculus, Eublepharis macularius, Bombina bombina, and Anguilla
   anguilla) from the NCBI database. The 5 species exhibited conserved
   evolution at the Asparagine residues 263 (N263) of CTSD
   (Figure [80]2E). The UniProt protein database predicted 2 potential
   N‐glycosylation sites, specifically at Asp residues 134 and 263, for
   CTSD. Given that Glutamine (Q) is the most analogous to Asparagine (N)
   in terms of physical and chemical properties, Asparagine was
   substituted with Glutamine (Asn → Gln, N263Q, N134Q) to minimize
   potential disruptions to other functions of CTSD (Figure [81]2F). Owing
   to the absence of low molecular weight (non‐glycosylated) bands and
   obvious N‐glycosylation modifications of CTSD in SW480 and SW620 cell
   lines (Figure [82]2B), the 2 cell lines were chosen for subsequent
   experimental investigations.

   We silenced endogenous CTSD expression in SW480 and SW620 cell lines
   (Figure [83]2G). Then, these CTSD‐deficient cell lines were transduced
   with exogenous plasmids encoding wild‐type CTSD (CTSD WT) or mutant
   forms of CTSD (CTSD N263Q, CTSD N134Q/N263Q) (Figure [84]2H). This
   process yielded stable cell models that expressed CTSD WT, CTSD N263Q,
   and CTSD N134Q/N263Q. Figure [85]2H shows the presence of
   N‐glycosylation at residues 134 and 263 of CTSD. Furthermore, we
   observed a reduction in the molecular weight (owing to the transition
   from glycosylated to non‐glycosylated forms) of CTSD following
   mutations at the 2 sites (Figure [86]2H). CTSD N134Q/N263Q cells
   treated with PNGase F did not show alteration of molecular weight in
   CTSD; therefore, CTSD had only 2 N‐glycosylation modification sites at
   residues 134 and 263 (Figure [87]2H).

2.3. N‐glycosylation Modification of CTSD Affected Biological Processes in
CRC Cells

   In vitro experiments, invasion, proliferation, and anti‐apoptotic
   capacities of SW480 and SW620 cells were found to be significantly
   inhibited following a mutation at residue 134 and 263 of CTSD (Figure
   [88]S1A− F, Supporting Information).

   We constructed xenograft tumor models and liver metastasis models in
   nude mice by injecting CTSD WT and CTSD N263Q SW480 cells. In the
   xenograft tumor models, a mutation at residue 263 of CTSD (CTSD N263Q)
   significantly inhibited subcutaneous tumor growth in nude mice
   (Figure [89] 3A−C). In the liver metastasis model, no liver metastases
   were observed in CTSD N263Q nude mice, although metastases were present
   in the CTSD WT nude mic (Figure [90]3D−G). Specifically, the liver
   metastasis ability of SW480 cells was inhibited after N‐glycosylation
   modification levels of CTSD were down‐regulated (Figure [91]3E−G).
   These findings suggest that reducing N‐glycosylation modification
   levels at residue 263 of CTSD may inhibit CRC growth and metastasis.

Figure 3.

   Figure 3
   [92]Open in a new tab

   N‐glycosylation of CTSD affects biological processes in CRC cells in
   vivo. A) CTSD N263Q and CTSD WT cells were implanted into nude mice by
   subcutaneous injection. Tumor volume was measured at three‐day
   intervals. After a period of 4 weeks, nude mice were euthanized, and
   tumors were excised for photographic comparison. B) Macroscopic
   examination of tumors. CTSD N263Q cells exhibited reduced tumorigenic
   potential, forming significantly smaller tumors in comparison with CTSD
   WT cells. C) Subcutaneous tumor volume was quantified. The volume of
   tumors formed by CTSD N263Q cells was lower than that of tumors formed
   by CTSD WT cells (n = 4; t‐test; *p < .05; mean ± SEM; each group
   underwent 3 independent experiments). D) CTSD N263Q and CTSD WT cells
   were injected beneath the splenic capsule, and nude mice were
   euthanized after a period of 4 weeks. The presence of liver metastases
   in nude mice was initially assessed through gross examination, and
   metastases were subsequently confirmed via hematoxylin and eosin
   staining of tissue sections. E) Macroscopic examination of tumors. No
   liver metastasis was observed in the livers of CTSD N263Q nude mice,
   whereas metastasis was detected in the livers of CTSD WT nude mice. F)
   The number of metastasis tumors in livers. (n = 6; t‐test; *p < .05;
   mean ± SEM; each group underwent 3 independent experiments). G)
   Hematoxylin and eosin staining corroborated the presence of liver
   metastasis in CTSD WT nude mice, while no liver metastasis was observed
   in CTSD N263Q nude mice.

2.4. N‐glycosylation Modification Alters the Folding Location, Stability, and
Functional Activity of CTSD

   The GeneCards dataset^[ [93]^10 ^] showed that CTSD was mainly located
   in the extracellular matrix, lysosomes, and endosomes (Figure [94] 4A).
   Different organelles were labeled with specific markers (EEA1 for the
   early endosomal marker, Rab7 for the late endosomal marker, LAMP1 for
   the lysosomal marker). CTSD was unable to co‐localize with endosomal
   and lysosomal markers in CTSD N263Q and CTSD N263Q/N134Q cells
   (Figure [95]4B). In other words, down‐regulating N‐glycosylation
   modification levels at residue 263 of CTSD may inhibit its transport to
   endosomes and lysosomes in SW480 cells.

Figure 4.

   Figure 4
   [96]Open in a new tab

   N‐glycosylation at residues 134 and 263 of CTSD regulated its folding
   location, stability, and functional activity. A)The intracellular and
   extracellular distribution of CTSD. CTSD was mainly localized in the
   extracellular matrix, lysosomes, and endosomes. B) Co‐localization of
   CTSD, lysosomes, and endosomes. CTSD failed to co‐localize with
   endosomal and lysosomal markers in both CTSD N263Q and CTSD N263Q/N134Q
   cell lines. (EEA1 for early endosomal marker, Rab7 for late endosomal
   marker, LAMP1 for lysosomal marker). C) In comparison with that in CTSD
   WT cells, secretion dynamics of CTSD N263Q cells decreased, while
   secretion dynamics of CTSD N263Q/N134Q cells increased (n = 3; t‐test;
   *p < .05, **p < .01; mean ± SEM; each group underwent 3 independent
   experiments). D) Stability analysis of CTSD. CTSD from CTSD N263Q cells
   exhibited significant degradation after 3 h of cycloheximide treatment,
   whereas CTSD WT did not show a significant decrease even after 5 h of
   cycloheximide treatment. E) CTSD enzyme activity analysis. The
   enzymatic activity of CTSD N263Q and CTSD N263Q/N134Q was observed to
   be lower than that of CTSD WT (n = 3; t‐test; **p < .01, ***p < .001
   and ****p < .0001; mean ± SEM; each group underwent 3 independent
   experiments).

   We also investigated the potential impact of N‐glycosylation
   modification at residue 263 of CTSD on its secretion dynamics. In
   comparison with CTSD WT cells, CTSD secretion of CTSD N263Q cells with
   the partial loss of N‐glycosylation modification decreased, while CTSD
   N263Q/N134Q cells with the absence of N‐glycosylation modification
   showed increased CTSD secretion (Figure [97]4C). Therefore, CRC cells
   could not secrete CTSD owing to the loss of N‐glycosylation
   modification of CTSD at residue 263.

   Additionally, we used cycloheximide to study whether N‐glycosylation
   modification of CTSD at residue 263 affected its stability. Compared
   with that in CTSD WT cells, the quantity of CTSD was significantly
   reduced in CTSD N263Q cells after 3 h of cycloheximide treatment
   (Figure [98]4D), indicating that N‐glycosylation modification of CTSD
   at residue 263 affects its stability.

   CTSD is an acid protease. We used the CTSD protease activity kit to
   detect CTSD N263Q and CTSD N263Q/N134Q enzyme activity. Fluorescence
   values were used to quantify the strength of enzyme activity. CTSD from
   CTSD N263Q cells and CTSD N263Q/N134Q cells had lower fluorescence
   values than that from CTSD WT cells (Figure [99]4E). Specifically,
   reducing N‐glycosylation modification levels at residues 134 and 263 of
   CTSD may decrease its protease activity (Figure [100]4E).

   In conclusion, N‐glycosylation modification of CTSD at residues 134 and
   263 may alter its folding location, stability, and functional activity.

2.5. STT3B and DDOST are N‐glycosyltransferases of CTSD

   We used co‐immunoprecipitation to enrich proteins binding with CTSD,
   and those proteins were subsequently identified by mass spectrometry
   analysis (Figure [101] 5A). The analysis results revealed that 235
   proteins interacted with CTSD WT and 191 proteins combined with CTSD
   N263Q. Among the 45 proteins that exclusively interacted with CTSD WT,
   DDOST (also known as OST48) was the only N‐glycosyltransferase
   (Figure [102]5B) and may be involved in N‐glycosylation modification at
   residue 263. The initial N‐glycosyltransferase, also known as
   oligosaccharide transferase complex (OST),^[ [103]^11 ^] consisted of
   multiple proteins in a complex. OST existed in 2 forms, OST‐A and
   OST‐B, with STT3A and STT3B as their respective catalytic subunits,
   each responsible for initial N‐glycosylation. Catalytic subunits were
   capable of independently completing the N‐glycosylation addition
   process. However, in certain instances, various non‐catalytic subunits,
   such as DDOST, were required to form stable complexes with the
   substrate during the process.^[ [104]^12 ^] We used CPTAC, an online
   proteomic database, to investigate the expression levels of STT3A,
   STT3B, and DDOST between colon cancer tissues and paired normal
   tissues, and found that the 3 proteins were upregulated in colon cancer
   tissues (Figure [105]5C).

Figure 5.

   Figure 5
   [106]Open in a new tab

   Identification of N‐glycosyltransferase for CTSD. A) CTSD with a FLAG
   label was enriched using FLAG magnetic beads and silver‐stained gel
   showed that proteins, such as DDOST, were enriched by CTSD. B) 191
   proteins were enriched by CTSD N263Q and 235 proteins were enriched by
   CTSD WT, while 45 proteins exclusively interacted with CTSD WT. C)
   Expression levels of STT3A, STT3B, and DDOST were upregulated in colon
   cancer relative to normal tissues. Boxplots show median (central line),
   upper and lower quartiles (box limits), and 1.5 × interquartile range
   (whiskers). D) N‐glycosylated modification at residue 263 of CTSD was
   regulated by STT3B and DDOST. When STT3B and DDOST were knocked out,
   molecular weights of CTSD decreased significantly. E)
   Co‐immunoprecipitation and western blot analysis demonstrated that CTSD
   formed complexes with STT3B and DDOST. F) Immunofluorescence analysis
   illustrated the co‐localization of DDOST and STT3B with CTSD, whereas
   no such co‐localization was observed with STT3A.

   To investigate the effects of DDOST on N‐glycosylation modification at
   residue 263 and to identify the corresponding catalytic subunit, we
   constructed stable SW480 cell lines with targeted knockouts of STT3A,
   STT3B, and DDOST (Figure [107]5D). When STT3B and DDOST expression were
   reduced, the molecular weights of CTSD significantly decreased relative
   to the negative control, indicating a shift from N‐glycosylation to
   non‐N‐glycosylation (Figure [108]5D). These findings suggest that
   N‐glycosylation at residue 263 of CTSD is potentially regulated by
   DDOST, and STT3B likely served as the catalytic subunit at this site.

   We used co‐immunoprecipitation and western blot to confirm CTSD
   combined with STT3B and DDOST (Figure [109]5E). We employed multiple
   immunofluorescence assays to determine the localization of CTSD, DDOST,
   STT3A, and STT3B in CRC tissues. Figure [110]5F demonstrated the
   co‐localization of DDOST and STT3B with CTSD, but not STT3A. These
   findings suggest that STT3B and DDOST are integral components of the
   glycosyltransferase enzyme complex, regulating N‐glycosylation
   modification at residue 263 of CTSD.

2.6. N‐glycosylation Modification of CTSD Regulates ACADM/Ferroptosis Axis to
Promote CRC Progression

   To elucidate the potential downstream proteins influenced by
   N‐glycosylation modifications of CTSD, we conducted a comparative
   proteomic analysis between CTSD WT and the CTSD N263Q cells
   (Figure [111] 6A). We identified 391 differentially expressed proteins
   (DEPs, |log[2] fold change| >0.263) between CTSD WT and CTSD N263Q
   cells (Figure [112]6B). Given the protease activity of CTSD, we
   conducted an intersection analysis between protein sets only
   interacting with CTSD WT and DEPs. This analysis yielded a total of 8
   candidate molecules:ACADM, PRDX3, ALDH2, EHHADH, DHCR7, TGFB1I1, IDH2,
   and ALDH1A1 (Figure [113]6C).

Figure 6.

   Figure 6
   [114]Open in a new tab

   CTSD protease potentially degrades ACADM. A) Protein quality control
   staining charts. B) Differential protein abundance ranking scatter plot
   (t‐test, |log[2] fold change| >.585, p < .05). C) Venn diagram
   illustrating intersecting proteins between differentially expressed
   proteins (|log[2] fold change| > 0.263, p <0.05) and proteins enriched
   by CTSD WT. D) Survival analysis for 8 candidate molecules. Patients
   with high ACADM and PRDX3 expression demonstrated a more favorable
   prognosis than those with low ACADM and PRDX3 expression (p < .05). E)
   The CPTAC proteomics database was used to examine ACADM and PRDX3
   protein levels in various stages of colon cancer and normal tissues.
   The expression levels of ACADM exhibited a decline concomitant with the
   progression of malignancy. Boxplots show median (central line), upper
   and lower quartiles (box limits), and 1.5 × interquartile range
   (whiskers). F) When CTSD WT and CTSD N263Q are expressed at equal
   levels, SW480 cells expressing CTSD N263Q exhibit a higher level of
   ACADM. G) Co‐immunoprecipitation analysis demonstrated that ACADM
   interacted with the CTSD WT, whereas no binding was observed with the
   CTSD N263Q variant. H) Immunohistochemical analysis showed that
   expression levels of ACADM were negatively correlated with expression
   levels of CTSD. I) Immunofluorescence analysis demonstrated the
   co‐localization of ACADM and CTSD.

   In comparison to the normal group, the expression levels of ACADM and
   ALDH1A1 were significantly down‐regulated in the tumor group, whereas
   the expression level of DHCR7 was significantly up‐regulated in the
   tumor group (Figure [115]S2, Supporting Information). Kaplan–Meier
   survival analysis revealed that expression levels of ACADM and PRDX3
   were significantly associated with survival prognosis (Figure [116]6D).
   We investigated the link between candidate molecules and metastasis
   (TNM staging). Results indicated that expression levels of ACADM
   decreased as malignancy progressed (Figure [117]6E), suggesting a
   stronger association with liver metastasis (Stage 4). Therefore, ACADM
   was selected for the subsequent analysis.

   When both CSTD WT and CTSD N263Q cells exhibited equivalent levels of
   CTSD expression, we observed that CTSD N263Q cells displayed higher
   levels of ACADM than CSTD WT cells (Figure [118]6F).
   Co‐immunoprecipitation analysis revealed a significant interaction
   between ACADM and CTSD WT, whereas no such interaction was observed
   with the CTSD N263Q variant (Figure [119]6G).

   Immunohistochemical techniques were employed to assess the positive
   expression scores of CTSD and ACADM in 90 CRC tissues and corresponding
   adjacent normal tissues within the same visual field. The analysis
   revealed a negative correlation between expression levels of CTSD and
   ACADM (Spearman correlation, Figure [120]6H). Multiple
   immunofluorescence assays indicated the co‐localization of CTSD and
   ACADM (Figure [121]6I). These results further verify that
   N‐glycosylation modification at residue 263 of CTSD affects the
   proteolytic degradation of ACADM by the CTSD protease.

   ACADM, an enzyme involved in fatty acid metabolism, does not directly
   regulate cellular proliferation, invasion, or metastasis.^[ [122]^13 ^]
   Therefore, we hypothesized that the regulation of ACADM expression via
   N‐glycosylation modification of CTSD at residue 263 may influence the
   invasiveness and metastasis of CRC cells through mechanisms associated
   with fatty acid metabolism. We utilized the WikiPathways dataset^[
   [123]^14 ^] to conduct an enrichment analysis of DEPs between CTSD WT
   and CTSD N263Q cells. Our findings indicated that these DEPs were
   enriched in ferroptosis pathways related to lipid metabolism (WP4313
   FERROPTOSIS) (Figure [124]S3A, Supporting Information). Therefore, we
   hypothesized that N‐glycosylation modification of CTSD at residue 263
   may influence the biological behaviors of CRC cells through the
   ACADM/ferroptosis axis.

   We primarily analyzed expression levels of ferroptosis proteins ACSL4,
   SLC7A11, and GPX4. The expression levels of ferroptosis‐related
   proteins (SLC7A11 and GPX4) were lower in CTSD N263Q cells than in CTSD
   WT cells (Figure [125]S3B, Supporting Information). Additionally,
   expression levels of ACADM exhibited a positive correlation with those
   of ACSL4, while demonstrating a negative correlation with the
   expression levels of SLC7A11 and GPX4 (Figure [126]S3B, Supporting
   Information). Therefore, ACADM and N‐glycosylation modification of CTSD
   at residue 263 regulate the expression levels of ferroptosis‐related
   proteins, such as ACSL4, SLC7A11, and GPX4.

   We also observed that the overexpression of ACADM inhibited the
   proliferation and invasion of SW480 and SW620 cells (Figure [127]S3C−E,
   Supporting Information). Following the knockdown of the
   N‐glycosyltransferase STT3B and DDOST in CTSD, the invasive and
   proliferation capability of CTSD WT cells was significantly reduced,
   which was consistent with that observed in CTSD N263Q and CTSD
   N263Q/N134Q cells (Figure [128]S3C−E, Supporting Information). These
   findings reaffirm that ACADM, STT3B, and DDOST modulate the biological
   behaviors of CRC cells through the regulation of CTSD N‐glycosylation.
   Therefore, CTSD, modulated by the N‐glycosyltransferases complex STT3B
   and DDOST through the addition of N‐glycosylation modification at
   residue 263, mediates the ACADM/ferroptosis‐related pathway axis and
   subsequently affects the invasive capacity of CRC cells.

3. Discussion

   CRC poses a serious health threat globally.^[ [129]^1 ^] In patients
   with CRC, metastasis is the primary determinant of unfavorable
   prognosis, with hepatic metastasis being the most prevalent form.^[
   [130]^3 ^] It is imperative to elucidate the molecular mechanisms
   underlying liver metastasis in CRC, with the aim of identifying novel
   therapeutic targets. Numerous studies have demonstrated that
   N‐glycosylation of proteins plays a significant role in various
   processes associated with malignant tumor metastasis.^[ [131]^7 ^]
   However, the differences in N‐glycosylated protein profiles between
   primary lesions and paired liver metastatic lesions in CRC remain
   unclear. In our study, we identified the differential expression
   profiles of N‐glycosylated proteins between primary lesions and paired
   liver metastatic lesions in 14 cases of CRC. This research offers a
   significant theoretical foundation for advancing the understanding of
   the molecular mechanisms underlying liver metastasis in CRC and for
   identifying novel therapeutic targets.

   CTSD was initially linked to cancer in breast cancer,^[ [132]^15 ^] and
   overexpression of CTSD was shown to alter its processing, cell
   localization, and secretion.^[ [133]^16 ^] CTSD overexpression is now
   recognized as a marker of poor prognosis associated with clinical
   metastasis.^[ [134]^17 ^] The distinct N‐glycosylation structure of
   CTSD in tumor tissues and serum in patients with breast cancer relative
   to normal tissues^[ [135]^18 ^] suggests that varying N‐glycosylation
   patterns of CTSD may be closely linked to tumorigenesis. Our study
   found that increased N‐glycosylation modification of CTSD at residue
   263 was significantly associated with liver metastasis in CRC and
   influenced the growth, proliferation, invasion, and metastasis of CRC
   cells. Therefore, this modification may be crucial for liver metastasis
   in CRC.

   N‐glycosylation modification of CTSD appears to play distinct roles
   depending on the specific N‐glycosylation structures, species, or
   tissues involved.^[ [136]^19 ^] Previous studies have demonstrated that
   the N‐glycosylation modification of CTSD is essential for its proper
   localization to lysosomal compartments.^[ [137]^19 , [138]^20 ^] This
   finding is also corroborated by our observation that CTSD missing the
   N‐glycosylation modification at residue 263 or both 134/263 cannot
   co‐locate with endosomal/lysosomal markers in CRC. Previous studies
   have debated whether N‐glycosylation promotes CTSD secretion.^[
   [139]^21 ^] Our findings in CRC revealed that losing N‐glycan at
   residue 263 hindered CTSD secretion, whereas losing N‐glycans at 2
   sites increased its secretion. The underlying mechanism requires
   further investigation. N‐glycosylation modification has been reported
   to influence the extracellular stability of CTSD but not its
   intracellular stability.^[ [140]^19c ^] However, in our studies,
   N‐glycosylation modification also affected the intracellular stability
   of CTSD. Previous studies have indicated that N‐glycosylation
   modification of CTSD had no effect on its enzymatic activity. However,
   utilizing the substrate sequence GKPILFFRLK(Dnp)‐D‐R‐NH2null, we found
   that N‐glycosylation modification did indeed affect the enzymatic
   activity of CTSD. This discrepancy in results may be attributed to the
   different enzyme substrates used, and warrants further investigation.

   We also attempted to identify the N‐glycosylated oligosaccharide
   transferase (OST) of CTSD. N‐glycosylation is initiated by the
   oligosaccharide transferase complex in the endoplasmic reticulum. Our
   study found that N‐glycosylation modification at residue 263 of CTSD
   may be regulated by the OST complex protein DDOST, with the OST complex
   protein STT3B potentially serving as the catalytic subunit at the
   specific site. Furthermore, by modulating the expression levels of
   STT3B and DDOST, the impact of N‐glycosylation modification at residue
   263 of CTSD on the biological behaviors of CRC cells can be replicated
   in vitro.

   ACADM, a medium‐chain acyl‐CoA dehydrogenase, initiated β‐oxidation, an
   aerobic process by which fatty acids are degraded into acetyl‐CoA for
   energy production.^[ [141]^22 ^] ACADM has been identified as a tumor
   suppressor that inhibits tumorigenesis and tumor progression in liver
   and pancreatic cancers.^[ [142]^23 ^] Our study demonstrated that
   N‐glycosylation modification at the residue 263 of CTSD modulated the
   expression levels of ACADM and ferroptosis‐related proteins, including
   ACSL4, SLC7A11, and GPX4, thereby altering the biological behaviors of
   CRC cells.

   Our study has several limitations. First, this was a single‐center
   study, and the CRC cohort exclusively comprised Chinese patients, which
   may introduce potential selection bias. Therefore, further prospective
   studies involving multi‐center cohorts and diverse ethnic populations
   are necessary to validate our findings. Second, we did not explore
   potential causes for discrepancies observed in comparison with prior
   studies. Lastly, a detailed investigation of the downstream regulatory
   factors influenced by N‐glycosylation modification at residue 263 of
   CTSD was not conducted. Future research should incorporate additional
   analyses and experiments to elucidate the precise roles of the
   N‐glycosylation modification of CSTD, and the underlying mechanisms, in
   liver metastasis in CRC.

   In conclusion, N‐glycosylation modification at residue 263 of CTSD has
   the potential to alter liver metastasis in CRC.

4. Conclusion

   In this study, we analyzed the N‐glycosylation modification protein
   expression profiles in primary lesions and paired liver metastatic
   lesions of CRC by Liquid Chromatography ‐Mass Spectroscopy (LC‐MS)
   analysis. N‐glycosylation modification at residue 263 of CTSD affects
   liver metastasis in CRC. This modification, regulated by the
   glycosyltransferases complex DDOST and STT3B, affects the cellular
   localization, stability, and enzymatic activity of CTSD. Furthermore,
   it influences the invasion and metastasis of CRC cells by modulating
   the ferroptosis pathway mediated by ACADM. These findings may provide
   potential therapeutic targets for CRC treatment as well as strategies
   for controlling CRC progression and metastasis.

5. Experimental Section

Clinical Sample Acquisition

   This study was approved by the Research Ethics Committee of Peking
   University People's Hospital (approval number: 2021PHB267‐001). Prior
   to sample collection, written informed consent was obtained from all
   participants. The clinicopathological characteristics of included
   patients are presented in Table [143]S4 (Supplementary Table).

   Patients who received surgical treatment at the Department of
   Gastrointestinal Surgery, Peking University People's Hospital, between
   November 2019 and July 2022, and were subsequently pathologically
   diagnosed with liver metastasis secondary to synchronous CRC, were
   selected. Samples from primary lesions and paired liver metastatic
   lesions were collected. Cases with incomplete clinicopathological data
   or those that failed quality inspection following total protein
   extraction were excluded. Additionally, cases presenting with
   intestinal obstruction, tumor‐induced intestinal perforation, or cancer
   thrombus formation during treatment were excluded.

Cell Lines

   293T, NCM460, SW480, SW620, RKO, DLD1, LoVo, HCT8, and HCT116 cell
   lines were obtained from the Cell Resource Center of the Chinese
   Academy of Sciences. Cells were cultured in 10% fetal bovine serum
   (Gibco, USA) and 1% penicillin‐streptomycin (Gibco, USA), and incubated
   at 37 °C and 5% CO2 in an incubator.

LC‐MS Analysis for Glycopeptide—Glycopeptide Enrichment and Proteolysis

   Primary lesions and paired liver metastatic lesions from patients were
   cut into pieces, and ground in liquid nitrogen using a mortar and
   pestle until the tissues became mucilaginous. Radio Immunoprecipitation
   Assay lysis buffer was added to the tissue samples to form a
   homogeneous solution, and samples were incubated for 30 min. The
   samples were then centrifuged at 12 000 g for 20 min, and the
   supernatant was transferred to ice. The concentration of extracted
   proteins was measured using the Pierce BCA Protein Assay which is an
   approach with high‐precision and detergent‐compatible.

   The extracted proteins were incubated with 10 mM DTT at 37 °C, for 1 h,
   to reduce the disulfide bonds in the glycoproteins. To avoid
   regeneration of disulfide bonds, carboxymethylation of sulfhydryl
   groups was subsequently performed by incubating the samples in 20 mM
   IAM at room temperature in the dark for 2 h. Redundant chemicals were
   eliminated before tryptic digestion using 3K ultrafiltration tubes.
   Then, the glycoproteins were digested with trypsin at a ratio of 50:1
   (w/w). Subsequently, the glycopeptides were purified and enriched using
   a hydrophilic HILIC solid phase extraction (SPE) column. Briefly, the
   tryptic peptides were dissolved in 80% ACN and 1% TFA. Before
   sample‐loading, the HILIC SPE columns were washed twice with 0.1% TFA
   and 80% ACN and 1% TFA. After samples were loaded, the columns were
   washed twice with 0.5 mL 80% ACN and 1% TFA. The glycopeptides bound to
   the columns were eluted twice with 0.5 mL 0.15%TFA solution. The eluted
   glycopeptides were dried via vacuum centrifugation, and stored at
   −80 °C for later use.

LC‐MS Analysis of Intact N‐glycopeptides

   Intact N‐linked glycopeptide samples were analyzed using the Orbitrap
   Eclipse Tribrid mass spectrometer (Thermo Fisher, USA). Glycopeptides
   were dissolved in 0.1% formic acid (buffer A). Then, samples were
   separated using the C18 column (homemade), and the mobile phase flow
   rate was 300 nL mi^−1n which consisted of 0.1% FA in water (A) and 80%
   ACN/0.1% FA (B). Samples were separated in a single LC‐MS run using a
   60‐min gradient, and the gradient profile for LC separation was set as
   follows: 5%–8% buffer B, 0–10 min; 8%–22% buffer B, 10–34 min; 22%–32%
   buffer B, 34–48 min; 32%–90% buffer B, 48–49 min; 90% buffer B,
   49–60 min.

   The Orbitrap MS^1 spectra were collected from 350–2000 m z^−1 at a
   resolution of 60 000, and the spray voltage was set at 2000 V in the
   positive mode. A dynamic exclusion time of 20 s was used to
   discriminate against previously selected ions. Data‐dependent HCD MS^n
   (n ≥ 2) scanning was performed at the mass range of 120–2000 m z^−1
   using an isolation window of 1.6 m z^−1 at a resolution of 30 000. The
   collision energy was set at 20%‐40%, and charge state screening was
   enabled to reject unassigned and singly charged ions. Tune application
   (Thermo Fisher Scientific) and Xcalibur software (Thermo Fisher
   Scientific) were used for instrument control and mass spectrometry data
   collection.

Identification and Quantification of N‐Glycopeptides

   Identification of intact N‐glycopeptides was performed using pGlyco 3.0
   and our previously developed intelligent glycan identification
   strategy, GIPS.^[ [144]^24 ^] Peptide sequences, N‐glycosylation sites,
   and glycan components were analyzed using pGlyco 3.0. Briefly, raw data
   were directly fed into pGlyco 3.0, and the database of human proteins
   extracted from UniProt was searched. The flow type was set as N‐linked,
   and trypsin was set as enzymatic digestion with up to 2 missed
   cleavages. The carbamidomethylation was set as dynamic modification,
   and the mass tolerance for the precursor and fragment ions were defined
   as ±5 and ±20 ppm, respectively. The N‐glycosylation sequence motifs
   (N‐X‐S/T, X ≠ P) were modified by switching “N” to “J” (both amino
   acids with the same mass). A quality control method for the
   identification of N‐linked glycopeptides was established, and the false
   discovery rate (FDR) for glycopeptide spectrum matching was set at 1%.
   The peptide sequence identified using pGlyco 3.0 was considered as
   fixed modifications of glycans, and the identification of glycan
   structures was carried out using the intelligent GIPS approach.
   Briefly, the GIPS approach was used to simulate fragments of all the
   isomeric glycan structures and to calculate the probability of isomeric
   glycan structures being the actual glycan structure using the
   integrated algorithm. The resulting files from pGlyco 3.0 were
   subsequently imported into pGlycoQuant for quantification of identified
   N‐glycopeptides.^[ [145]^25 ^]

Identification of Differential Expression N‐glycopeptides

   Differentially expressed glycopeptides were identified by the Clinical
   Medical Statistical Analysis Platform
   ([146]http://msap.cqxyy.org.cn/login?redirect = %2Findex). Briefly, we
   first performed a normal distribution test and homogeneity test of
   variance of each N‐glycopeptide, and then performed a t‐test on
   glycopeptides satisfying these 2 tests. Otherwise, a nonparametric test
   was accordingly conducted. Mann‐Whitney test was performed for a
   2‐group test. The results were subsequently visualized through an
   online platform ([147]https://www.bioinformatics.com.cn/) based on an
   “EnhancedVolcano” R package. Glycopeptides exhibiting |log[2] fold
   change| > .585 and p < .05 were classified as differentially expressed
   glycopeptides.

Enrichment Analysis of N‐Glycosylated Proteins

   UniProt ([148]http://www.uniprot.org/) was utilized to identify the
   gene name corresponding to N‐glycosylated proteins. These genes were
   subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathway enrichment analysis in DAVID
   ([149]https://david.ncifcrf.gov/) dataset. Fisher's exact test was
   applied to identify the significant enrichment GO categories and KEGG
   pathways of these genes, and FDR was used to correct p values
   (p < .05). The results were subsequently visualized utilizing an online
   platform ([150]https://www.bioinformatics.com.cn/).

Western Blot Assay

   The methodology was consistent with previous procedures.^[ [151]^26 ^]
   Anti‐rabbit CTSD (AB‐75811, Abcam), Anti‐mouse CTSD (AB‐302649, Abcam),
   Anti‐rabbit CTSD (AB‐75852, Abcam), Anti‐mouse GAPDH (60004‐1‐Ig,
   Proteintech), Anti‐rabbit IgG, HRP‐linked Antibody (Cell Signaling
   Technology), Anti‐mouse IgG, HRP‐linked Antibody (Cell Signaling
   Technology), Anti‐mouse ACADM (67742‐1‐Ig, Proteintech), Anti‐rabbit
   DDOST (14916‐1‐AP, Proteintech), Anti‐mouse STT3A (66581‐1‐Ig,
   Proteintech), Anti‐mouse STT3B (15323‐1‐AP, Proteintech), Anti‐rabbit
   ACSL4 (AB‐155282, Abcam), Anti‐rabbit SLC7A11 (AB‐307601, Abcam) and
   Anti‐rabbit GPX4 (AB‐125066, Abcam) were used.

PNGase F Processing

   A total of 20 µg of extracted cellular proteins was combined with 1 µL
   of 10× Glycoprotein Denaturing Buffer, and deionized water was added to
   achieve a final volume of 10 µL. The mixture was then subjected to
   centrifugation and incubated at 100 °C for 10 min. Following
   incubation, the denatured sample was promptly placed on ice and rapidly
   centrifuged. Subsequently, 2 µL of 10× Glyco Buffer 2 (Yi Sheng,
   China), 2 µL of 10% NP‐40 (Yi Sheng, China), 1 µL of PNGase F (Yi
   Sheng, China), and 5 µL of deionized water were added to the sample.
   The resulting mixture was centrifuged and incubated at 37 °C for 1 h.

Tunicamycin Processing

   Tunicamycin powder (MCE, USA) was dissolved in DMSO to prepare a drug
   solution with a concentration of 2 µg µL^−1. This solution was
   subsequently added to the experimental medium to achieve a final
   concentration of 2 µg mL^−1. An equivalent volume of DMSO was added to
   the control group. Total proteins were extracted from the lysed cells
   24 h post‐treatment for subsequent analysis.

Site‐Directed Mutagenesis

   The coding DNA sequence for CTSD was retrieved from the NCBI database
   ([152]http://www.ncbi.nlm.nih.gov/) and the codon sequence
   corresponding to the amino acid site to be mutated was determined. The
   Asparagine (N) residue was substituted with Glutamine (Q), chosen for
   its closely related physicochemical properties; accordingly, the codon
   was altered from AAT to CAG. Upstream primers were designed by
   selecting the sequence comprising 11–13 bases upstream and downstream
   of the mutation site, and the mutation site itself. The reverse
   complementary sequence of the upstream primers was utilized as the
   downstream primers. Plasmid DNA was isolated by PCR and agarose gel
   electrophoresis. The CTSD mutant constructs were subsequently verified
   through sequencing and used in further experimental procedures.

Plasmid Construction

   The CTSD shRNA was specifically engineered to target the 3′
   untranslated region (UTR) of CTSD mRNA, ensuring that it did not
   interfere with the expression of the exogenous CTSD plasmid. Puromycin
   resistance cassettes were used for the lentiviral plasmid of CTSD
   shRNA, and Blasticidin resistance cassettes were used for the
   lentiviral plasmid of CTSD WT, CTSDN263Q, and CTSDN263Q∖N134Q. To meet
   the requirements of subsequent experiments, all plasmids were designed
   to be non‐fluorescent.

Cell Transfection/Lentiviral Infection

   Lipofectamine 3000 (Invitrogen, USA) was used for transient
   transfection. The CTSD shRNA lentiviral vector was utilized to silence
   CTSD expression in CRC cells. Subsequently, CTSD WT, CTSD N263Q, and
   CTSD N263Q/N134Q lentiviruses were transfected into these CTSD‐silenced
   CRC cell lines to generate new cell lines that stably expressed CTSD
   WT, CTSD N263Q, and CTSD N263Q/N134Q proteins, respectively.

Cell Proliferation, Migration, and Invasion Assays

   These assays were performed in accordance with previous procedures.^[
   [153]^26 ^]

Flow Cytometry Assay

   A 2× EdU working solution and Click Additive Solution were prepared
   according to the manufacturer's instructions (Biyuntian Biology,
   China). Cells in the logarithmic growth phase were selected. An equal
   volume of preheated 2× EdU working solution (20 µM) at 37 °C was added
   to each well of a 6‐well plate. The cells were incubated for 45 min.
   Subsequently, cell proliferation was assessed by flow cytometry
   following a series of steps including fixation, washing,
   permeabilization, and the Click reaction with paraformaldehyde.

   Then, 1× Annexin binding buffer was prepared in accordance with the
   manufacturer's instructions (Union Biology, China). The cells were
   resuspended in 100 µL of the formulated 1× Annexin binding buffer.
   Subsequently, 5 µL of Alexa Fluor 488 Annexin V and 1 µL of
   100 µg mL^−1 propidium iodide (PI) were added to each tube. The samples
   were incubated at room temperature for 15 min, protected from light.
   Following incubation, 400 µL of the 1× Annexin binding buffer was added
   to each tube. Apoptosis was then assessed by flow cytometry.

   The cell cycle was analyzed using a cell cycle staining kit
   (Biobiology, China). Cells in the logarithmic growth phase were treated
   with 1 mL of DNA Staining Solution and 10 µL of Permeabilization
   Solution. The mixture was subjected to vortex oscillation for 5–10 s.
   Subsequently, the cells were incubated at room temperature for 30 min
   in the absence of light. Flow cytometry was employed to measure the
   cell cycle.

Nude Mouse Subcutaneous Tumor Formation and Liver Metastasis Model

   For subcutaneous tumor formation in nude mice, female Balb/c nude mice,
   aged 4–6 weeks, were randomly assigned to one of 2 groups—CTSD WT group
   and the CTSD N263Q group—based on the types of implanted cells. Then,
   100 µL of the cell suspension, containing 5 × 10^5 cells, was injected
   into the right dorsal subcutaneous region of nude mice.
   Post‐inoculation, tumor formation was monitored every 3 days. Upon
   macroscopic detection of the tumor, the longest and perpendicular
   diameters were measured using a vernier caliper. Tumor volume was
   calculated using the formula V = 0.5 × L (length) × W^2 (width). A
   tumor growth curve was then plotted based on these measurements.

   Next, a liver metastasis model was established with the following
   protocol. The cell concentration was adjusted to 2×10^7cells mL^−1
   using the aforementioned method. Cell suspensions of ≈50 µL, containing
   1 × 10^6 cells, were injected into the spleen using a 1 mL syringe.
   Four weeks post‐injection, the nude mice were euthanized to assess
   tumor metastasis in the spleen, liver, and lungs. The number of
   metastatic sites was quantified and represented as a histogram. Liver
   tissues were subsequently fixed with paraformaldehyde, followed by
   paraffin embedding and sectioning for further analysis.

Immunofluorescence

   The cell culture dish was pre‐coated with climbing tablets, and the
   cells were cultured overnight. Following cell adhesion, the cells were
   fixed using paraformaldehyde, permeabilized with Triton X‐100, blocked
   with bovine serum albumin (BSA), incubated with primary and secondary
   antibodies, counterstained with DAPI tablets, and subsequently observed
   under a fluorescence microscope. The following antibodies were used:
   Anti‐rabbit LAMP1 (Catalog Number 9091, Cell Signaling Technology),
   Anti‐rabbit EEA1 (Catalog Number 3288, Cell Signaling Technology),
   Anti‐rabbit Rab7 (Catalog Number 9367, Cell Signaling Technology),
   555‐conjugated secondary antibody (Abcam, UK) and 647‐conjugated
   secondary antibody (Abcam, UK).

Enzyme Functional Activity Detection

   The Cathepsin D Activity Assay Kit (Abcam) was used with the preferred
   CTSD substrate sequence, GKPILFFRLK(Dnp)‐D‐R‐NH2, labeled with MCA.
   Upon incubation with a cell lysate containing CTSD, the substrate was
   cleaved, resulting in the release of fluorescence. This fluorescence
   was subsequently quantified using an enzymometer, with excitation and
   emission wavelengths set at 328 and 460 nm, respectively.

LC‐MS Analysis for Protein‐CTSD Enrichment

   Initially, total proteins were extracted from both the CTSD WT stable
   strain and the CTSD N263Q stable strain of SW480 cells. Subsequently,
   the CTSD protein, tagged with a FLAG label, was enriched using FLAG
   magnetic beads. Following the treatment of equal amounts of protein
   samples, the samples were subjected to electrophoresis, fixation,
   sensitization, and silver staining. Following the dyeing of the
   samples, images of the silver‐stained gels were captured.

Collecting Total Proteins

   Similarly, total proteins were extracted from both the CTSD WT stable
   strain and the CTSD N263Q stable strain of SW480 cells. Equal amounts
   of protein samples were subjected to electrophoresis, followed by
   Coomassie Brilliant Blue staining, decolorization, and additional
   procedures. Images of the stained gels were used to document the
   results.

LC‐MS Analysis of Proteins

   The peptides were solubilized using the mobile phase A of liquid
   chromatography, which consisted of an aqueous solution containing 0.1%
   formic acid and 2% acetonitrile. Separation was performed using the
   EASY‐nLC 1200 Ultra‐High‐Performance Liquid Chromatography system
   (Thermo Fisher, USA). The liquid phase gradient was programmed as
   follows: 0–22.5 min, 6%–22% buffer B (comprising an aqueous solution of
   0.1% formic acid and 90% acetonitrile); 22.5–26.5 min, 22%–34% buffer
   B; 26.5–28.5 min, 34%–80% buffer B; 28.5–30 min, 80% buffer B, the flow
   rate was consistently maintained at 700 nL min^−1. The peptides were
   fractionated using an ultra‐high‐performance liquid chromatography
   system and subsequently introduced into a nano‐spray ionization source
   for ionization. The ionized peptides were then subjected to analysis
   via Orbitrap Exploris 480 mass spectrometry. The ion source voltage was
   maintained at 2300 V, while the FAIMS (Field Asymmetric Ion Mobility
   Spectrometry) compensation voltage was configured to −45 V.
   High‐resolution Orbitrap mass spectrometry was performed to detect and
   analyze the peptide precursor ions along with their secondary
   fragments. The first mass spectrometry was configured with a scanning
   range of 350–1400 m z^−1 and a resolution of 60000. In contrast, the
   secondary mass spectrometry was set to a fixed scanning range of
   120 m z^−1 with a resolution of 15000. Data acquisition was conducted
   using a data‐independent acquisition (DIA) method. In this procedure,
   peptide ions were sequentially introduced into the higher‐energy
   collisional dissociation collision cell through multiple successive m/z
   windows following the first scan, where they were fragmented using 27%
   fragmentation energy, subsequently undergoing secondary mass
   spectrometry analysis in turn. To enhance the efficient utilization of
   the mass spectrum, the automatic gain control (AGC) was configured to
   1E6, and the maximum injection time was adjusted to 22 ms.

Identification and Quantification of Proteins

   The DIA data were processed using the DIA‐NN search engine (version
   1.8) and using the software's default parameters. The database utilized
   was Homo_sapiens_9606_SP_20 230 103.fasta, comprising 20389 sequences.
   The enzymatic cleavage method employed was Trypsin/P, with the maximum
   allowable number of missed cleavages set to one. Fixed modifications
   were specified as N‐terminal methionine excision and cysteine
   carbamidomethylation. The theoretical spectral library was generated
   using a deep learning algorithm, with an inverse library incorporated
   to assess the FDR attributable to random matching. The FDR threshold
   for precursor identification was established at 1%. To ensure
   high‐quality analytical outcomes, the results from the database search
   require further filtering. The filtering criteria for the
   identification results: a precursor and protein with FDR of 1%; each
   identified protein with at least one unique peptide.

Identification and Enrichment Analysis of Differential Expression Proteins
(DEP)

   The “DEP” R package, in conjunction with Benjamini‐Hochberg‐adjusted p
   values, was employed to identify DEPs between CTSD WT and CTSD N263Q
   cells (t‐test, |log[2] fold change| > .263, p < .05), and the results
   were subsequently visualized through “ggplot2” R package. We used the
   WikiPathways dataset ([154]https://www.wikipathways.org/) to annotate
   enrichment pathways of DEPs between CTSD WT and CTSD N263Q cells.
   Fisher's exact test was applied to identify the significant
   WikiPathways pathways, and FDR was used to correct p values. The
   results were subsequently visualized utilizing an online platform
   ([155]https://www.bioinformatics.com.cn/).

Co‐Immunoprecipitation Analysis (co‐IP)

   Total protein extraction was extracted using BC100 buffer, which
   comprised 100 mmol L^−1 sodium chloride, 20% glycerol, 0.2% NP40,
   50 mmol L^−1 Tris, 1 mmol L^−1 DTT, and 1 mmol L^−1 PMSF, with the pH
   adjusted to 7.9 using HCl. The protein concentration was subsequently
   quantified. The target protein was then captured at low temperatures
   using a FLAG affinity gel (A220, Sigma Aldrich), followed by a
   purification step to remove impurities. Competitive elution of the
   target protein was achieved using the FLAG peptide (Biyuntian
   Biological, China). Finally, the purified protein was analyzed using
   mass spectrometry or western blot.

Gene and Protein Expression Level Analysis and Survival Analysis

   We investigated gene expression levels (log2(TPM+1)) of Colon
   adenocarcinoma (COAD, n = 275), Rectum adenocarcinoma (READ, n = 92)
   patients and normal control (51) utilizing an online platform ^[
   [156]^10 ^] ([157]http://gepia.cancer‐pku.cn/index.html) where it based
   on the TCGA datasets, used One–way analysis of variance (ANOVA) to
   calculating differential expression between tumor (COAD and READ) and
   normal control. Patients with CRC (n = 181) were stratified into 2
   cohorts characterized by high and low gene expression levels,
   determined according to the median gene expression value. Overall
   survival curves for these 2 groups were generated utilizing the online
   platform ([158]http://gepia.cancer‐pku.cn/index.html) based on the
   Kaplan‐Meier analysis (log‐rank test). Additionally, we utilized the
   UALCAN online platform ([159]https://ualcan.path.uab.edu/), which is
   based on the Clinical Proteomic Tumor Analysis Consortium (CPTAC)
   database ([160]https://proteomics.cancer.gov/programs/cptac), to
   analyze protein expression levels across various stages in patients
   with CRC. Z‐values represent standard deviations from the median across
   samples for CRC samples. Log2 spectral count ratio values from CPTAC
   were first normalized within each sample profile and then normalized
   across samples.

Immunohistochemistry

   The paraffin‐embedded sections were subjected to a series of
   preparatory steps, including dewaxing to water, antigen retrieval,
   blocking of endogenous enzymes, preparation of slides, sealing to
   prevent non‐specific binding, incubation with primary and secondary
   antibodies, DAB chromogenic development, nuclear counterstaining,
   sealing for preservation, and microscopic evaluation. These procedures
   ensured optimal antigen exposure and antibody binding. Subsequently,
   the images were analyzed and quantified using microscopic observation
   and ImageJ software ([161]https://imagej.net/ij/index.html). Spearman's
   rank correlation was used to analyze the correlation between CTSD and
   ACADM.

Statistical Analysis

   The flowchart figure was drawn using the Figdraw platform
   ([162]https://www.figdraw.com/). Mann‐Whitney test was performed to
   identify differential expression glycopeptide. The
   Benjamini‐Hochberg‐adjusted P values of DEPs were calculated in R
   (version 4.1.3, [163]https://www.r‐project.org/). Fisher's exact test
   was employed for the enrichment analysis of categorical data.
   Experimental statistical data were analyzed and visualized using
   GraphPad Prism software (version 9,
   [164]https://www.graphpad.com/features), and continuous variables from
   2 independent samples were assessed using the t‐test. The log‐rank test
   was employed to evaluate the statistical significance of differences
   between Kaplan‐Meier survival curves. One‐way analysis of variance
   (ANOVA) was used to evaluate the differences in gene expression between
   the 2 groups. The association between 2 groups was analyzed utilizing
   Spearman's rank correlation. Each group underwent 3 independent
   experiments.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   N.X., Y.D., C.H. contributed equally to this work. N.X. contributed
   data curation, formal analysis, methodology, resources, software,
   validation, and visualization, and wrote, reviewed, and edited the
   original draft. Y.D. contributed data curation, formal analysis,
   methodology, resources, software, validation, and visualization. C.H.
   contributed data curation, formal analysis, methodology, resources,
   validation, and visualization, and wrote, reviewed, and edited the
   original draft. L.Z., C.Y., Q.S., Z.G., C.W., J.Z., H.Z., and S.W.
   contributed methodology, software, validation, and visualization. Y.Y.
   contributed data curation, formal analysis, methodology, resources, and
   supervision. Y.L. contributed data curation, formal analysis, funding
   acquisition, methodology, resources, and supervision, and wrote,
   reviewed, and edited the original draft. Z.S. contributed data
   curation, formal analysis, funding acquisition, methodology, resources,
   supervision, validation, and visualization, and wrote, reviewed, and
   edited the original draft.

Supporting information

   Supporting Information
   [165]ADVS-12-2411740-s005.docx^ (11.1MB, docx)

   Supplementary Table
   [166]ADVS-12-2411740-s003.xlsx^ (948.8KB, xlsx)

   Supplementary Table
   [167]ADVS-12-2411740-s002.xlsx^ (14.9KB, xlsx)

   Supplementary Table
   [168]ADVS-12-2411740-s001.xlsx^ (57.7KB, xlsx)

   Supplementary Table
   [169]ADVS-12-2411740-s004.xlsx^ (9.4KB, xlsx)

Acknowledgements