PLVAP protein expression correlated with microbial composition, clinicopathological features, and prognosis of patients with stomach adenocarcinoma

Abstract

Purpose

Plasmalemma vesicle-associated protein (PLVAP) is involved in many immune‑related signals; however, its role in stomach adenocarcinoma (STAD) remains to be elucidated. This study investigated PLVAP expression in tumor tissues and defined the value in STAD patients.

Methods

A total of 96 patient paraffin-embedded STAD specimens and 30 paraffin-embedded adjacent non-tumor specimens from the Ninth Hospital of Xi’an were consecutively recruited in analyses. All RNA‑sequence data were available from the Cancer Genome Atlas database (TCGA). PLVAP protein expression was detected using immunohistochemistry. Microbial community analysis was performed by 16S rRNA gene sequencing using Illumina MiSeq. PLVAP mRNA expression was explored with the Tumor Immune Estimation Resource (TIMER), GEPIA, and UALCAN databases. The effect of PLVAP mRNA on prognosis was analyzed via GEPIA, and Kaplan–Meier plotter database. GeneMANIA and STRING databases were used to predict gene/protein interactions and functions. The relationships between PLVAP mRNA expression and tumor-infiltrated immune cells were analyzed via the TIMER and GEPIA databases.

Results

Significantly elevated transcriptional and proteomic PLVAP expressions were found in STAD samples. Increased PLVAP protein and mRNA expression were significantly associated with advanced clinicopathological parameters and correlated with shorter disease-free survival (DFS) and overall survival (OS) in TCGA (P < 0.001). The microbiota in the PLVAP-rich (3+) group was significantly different from that in the PLVAP-poor (1+) group (P < 0.05). The results from TIMER showed that high PLVAP mRNA expression had significant positive correlations with CD4 + T cell (r = 0.42, P < 0.001).

Conclusion

PLVAP is a potential biomarker to predict the prognosis of patients with STAD, and the high level of PLVAP protein expression was closely related to bacteria. The relative abundance of Fusobacteriia was positvely associated with the level of PLVAP. In conclusion, positive staining for PLVAP was useful for predicting the poor prognosis of STAD with Fusobacteriia infection.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00432-023-04607-3.

Introduction

Stomach adenocarcinoma (STAD) is the fourth most common cancer and the second leading cause of cancer-related mortality in the world (Wee et al. 2012). STAD is characterized by high recurrence rates and a dismal prognosis, given the lack of early detection and effective treatment methods for this cancer type (Van Cutsem et al. 2016). Thus, the development of novel biological markers and molecular-targeting drugs that could predict STAD and improve clinical management is essential (Hu et al. 2019).

Plasmalemma vesicle-associated protein (PLVAP) is an endothelial-associated protein that is involved in endothelial diaphragm formation and maintenance of basal vascular permeability (Stan et al. 2012). A previous study indicated that PLVAP is an independent marker for colorectal cancer recurrence and its increased detection is associated with bacteria (Bertocchi et al. 2021). Notably, bacteria are crucial to the formation of the tumor microenvironment (TME). Entry into microbes and microbial products into the evolving TME potentiates tumor growth by eliciting tumor-promoting immune cell responses (Kostic et al. 2013). At the same time, PLVAP has roles in immunity by facilitating leukocyte diapedesis at inflammatory sites and controlling peripheral lymph node morphogenesis and the entry into soluble ags into lymph node conduits (Elgueta et al. 2016).

Here, we first used multiple public databases to detect whether PLVAP mRNA is differentially expressed in the stomach between patients with STAD and healthy controls. Second, the PLVAP protein expression level was validated by immunohistochemistry (IHC) staining. Third, microbes demonstrated significant abundance differences between the PLVAP-rich and PLVAP-poor groups by 16 S ribosomal RNA (16 S rRNA). Finally, we investigated the relationship between PLVAP and immune cells based on two different public databases.

Materials and methods

Patients’ cohort

A total of 96 patient paraffin-embedded STAD specimens and 30 paraffin-embedded adjacent non-tumor specimens were included in this study. All cases were obtained from the Ninth Hospital of Xi’an from January 2018 to May 2021. Inclusion criteria: (1) adjuvant radiotherapy and chemotherapy were not performed preoperatively; (2) the clinical and pathological data were complete; (3) willingness to provide informed consent. Patients combined with other tumors were excluded from this study. STAD classification, grade, and pathological staging were determined according to WHO classification 2016. There were 36 women and 60 men enrolled in our present study. The study was approved by the hospital and all patients signed the informed consent.

Immunohistochemical staining (IHC)

The EnVision method was used for IHC. STAD specimens obtained from primary tumors were fixed in 10% formalin, embedded in paraffin, and cut into 4 µm sections. Formalin-fixed paraffin-embedded (FFPE) received routine baking and were dewaxed with xylene and rehydrated with a series of graded ethanol. 3% hydrogen peroxide was adopted to clear endogenous hydrogen peroxidase. Antigen retrieval was performed in sodium citrate buffer. The sections were incubated with the anti‑LPS (Cloud-Clone Corp, Wu Han, China, 1:100, MAB526Ge22), anti-PLVAP rabbit monoclonal antibody (Cloud-Clone Corp, Wu Han, China, 1:100, PAH363Hu01), and anti-CD34 (Maixin Biotech Co., Ltd., Fuzhou, China, Kit-0004) primary antibody overnight at 4 °C. Subsequently, followed by incubation with the horseradish peroxidase‑conjugated anti‑rabbit secondary antibody [HRP] (Maixin Biotech Co., Ltd., Fuzhou, China, Kit-5020) for 20 min at room temperature. Antibodies were visualized with diaminobenzidine and counterstained with hematoxylin before prior microscopic observation.

Evaluation of immunohistochemical staining (IHC)

IHC was assessed by two independent pathologists who were blinded to the clinical data. The percentage of positively stained cells was graded as (i) 0, No; (ii) 1, light; (iii) 2, brown; and (iv) 3, dark brown. In addition, the percentage of positive cells were rated as follows: (i) 0, < 10%; (ii) 1, 10–25%; (iii) 2, 25–50%; (iv) 3, 50–75%; and (v) 4, > 75%. LPS/PLVAP staining positivity was calculated as score = percentage score × intensity score. Thus, division of score into four groups according to the percentage and intensity scores: (i) score ≤ 3, negative; (ii) score > 3 and ≤ 6, weak; (iii) score > 6 and ≤ 9, moderate; and iii) score > 9, strong (Jiang et al. 2018). Human Protein Atlas (HPA) (https://www.proteinatlas.org) was used to validate the expression of PLVAP protein and the IHC images were obtained from the HPA database.

Microbial community analysis

Microbial DNA was extracted from Human STAD specimen samples using the TIANamp FFPE DNA Kit (Tiangen Biotech, Beijing, China) according to manufacturer’s protocols. The final DNA concentration and purification were determined by NanoDrop 2000 UV–Vis spectrophotometer (Thermo Scientific, Wilmington, USA), and DNA quality was checked by 1% agarose gel electrophoresis. Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 300) on an Illumina MiSeq platform (Illumina) based on the standard protocols by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China).

Analysis of the stomach microbial community was carried out using the free online platform of Majorbio Cloud Platform (www.majorbio.com). Raw fastq files were demultiplexed, quality filtered by Trimmomatic, and merged by FLASH with the following criteria: (i) the reads were truncated at any site receiving an average quality score < 20 over a 50 bp sliding window. (ii) Primers were exactly matched allowing two nucleotide mismatching, and reads containing ambiguous bases were removed. (iii) Sequences whose overlap was longer than 10 bp were merged according to their overlap sequence.

Operational taxonomic units (OTUs) were clustered with 97% similarity cut-off using UPARSE (version 7.0.1090, http://drive5.com/uparse/), and chimeric sequences were identified and removed using UCHIME.

Based on 16 S rRNA gene sequence data, Tax4Fun, an open-source R package, was used to predict functional capacities of microbial communities mapping with Kyoto Encyclopedia of Genes and Genomes (KEGG) reference database.

GEPIA analysis

The GEPIA database (http://gepia.cancer-pku.cn, 17 January 2022) is an online website that can be used to analyze the RNA expression data, based on TCGA and the Genotype-Tissue Expression (GTEx) Projects (Tang et al. 2017). In addition, GEPIA was used to further verify the prognostic value of the genes. The correlations between immune cell markers and PLVAP mRNA expression were identified using correlation modules in GEPIA2.

UALCAN analysis

We analyzed comprehensive TCGA cancer genome expression data using UALCAN platform (Chandrashekar et al. 2017) (http://ualcan.path.uab.edu/index.html, 17 January 2022). UALCAN was used to analyze PLVAP mRNA relative transcriptional expression levels of genes in STAD and adjacent normal tissues.

Tumor immune estimation resource (TIMER) 2.0 analysis

The TIMER 2.0 database (http://cistrome.org/timer, 17 January 2022) was used to determine the abundances of tumor infiltrates based on gene expression analyses. The DiffExp module with default parameters was used to obtain the different expression levels of PLVAP mRNA in STAD and normal tissues. B cells, the clusters of differentiation 8-positive (CD8 +) T cells, CD4 + T cells, neutrophils, lll; macrophages, and dendritic cells (DCs) were selected as the test types (Li et al. 2017).

Kaplan–Meier (KM) plotter analysis

The KM plotter database (http://kmplot.com/analysis/, 17 January 2022) interprets information regarding gene expression and survival analysis of different types of cancer including breast, liver, ovarian, lung, and stomach cancer (Lánczky et al. 2016). KM plotter was used to analyze the prognostic values of PLVAP in patients with STAD.

GeneMANIA and STRING analysis

GeneMANIA (http://genemania.org, 17 January 2022) (Mostafavi et al. 2008) and STRING (https://string-db.org/cgi/, 17 January 2022) (Szklarczyk et al. 2017) are web tools to identify the interactions between genes and proteins, respectively. The interactions between PLVAP mRNA and other gene/proteins are utilized by GeneMANIA and STRING at the gene or protein level.

Statistical analysis

All statistical analyses were performed using IBM SPSS software (version 26; IBM Corp., Armonk, NY, USA). The data are presented as the means ± SDs. Differences in the clinicopathological characteristics were determined using the Chi-square test. The Kaplan–Meier method and log rank test were used to calculate the rate of OS and to compare differences between the survival curves. Logistic regression was performed to investigate the association between protein expression status and clinicopathological factors using a backward regression procedure that included variables where P < 0.1 in the univariate analysis. Correlation was assessed using a Pearson’s correlation test. Univariate and multivariate models were established using a Cox proportional hazards regression analysis to evaluate the prognostic significance of various factors. All statistical tests were two sided. P < 0.05 was considered significant.

Results

PLVAP mRNA expression level correlates with poor prognosis in STAD patients from multiple datasets

An analysis using the UALCAN online bioinformatics tool with default settings showed that the mRNA levels of PLVAP were higher in STAD tissues, compared to normal tissues (Fig. 1A). A similar result was found in the TIMER database (Fig. 1B). However, the GEPIA database showed that no obvious difference could be observed for PLVAP mRNA expression levels between STAD and adjacent tissues to the carcinoma (Fig. 1C).

Fig. 1.

The expression of PLVAP mRNA in stomach adenocarcinoma (STAD) and normal tissues. A UALCAN database; B TIMER database; C GEPIA database. (scatter diagram; box plot). *P < 0.05; **P < 0.01; ***P < 0.001 vs. normal tissue. ****P < 0.0001

To further investigate the correlations between PLVAP mRNA signature expressions and clinical prognoses, we predicted the overall survival percentage using multiple datasets. GEPIA (Supplementary Figure S1) showed that PLVAP mRNA was not associated with overall survival (OS), disease-free survival (DFS). Furthermore, results from the KM plotter revealed that patients with the high PLVAP mRNA level exhibited a shorter overall survival (OS) compared with patients with low levels (Fig. 2A). Next, we performed KM survival curves to evaluate the OS of these patient subgroups (Fig. 2B, C). The results revealed that the high PLVAP mRNA level was associated with a shortened OS in STAD patients with HER-2 positive (HR = 1.94; P < 0.0001) and negative (HR = 2.66; P < 0.0001). High PLVAP mRNA level was significantly associated with poor OS time in STAD patients that had received surgery (HR = 1.84; P < 0.0001). However, this result was not observed in STAD patients who received 5-FU-based adjuvant chemotherapy (HR = 0.69; P = 0.07).

Fig. 2.

Prognostic values of PLVAP in stomach adenocarcinoma. Analysis is demonstrated for A all patients (P < 0.0001); B Her-2 positive (P < 0.0001); C Her-2 negative (P < 0.0001); D only surgery (P = 0.0001); E 5-FU (P = 0.07); HER‑2, human epidermal growth factor receptor 2; 5-FU: fluorouracil 5

PLVAP expression and clinicopathological features in STAD tissues using IHC

To evaluate the clinical significance of PLVAP in STAD progression, we detected the expression level of PLVAP protein in STAD tissues via IHC. Positive PLVAP IHC was observed in 84.4% (81/96) of STAD tissues and only 43.3% (13/30) of the normal tissues (χ² = 37.33; P < 0.001). The 96 STAD cases were classified into 3 groups according to the PLVAP expression level in cancer tissue. High PLVAP expression was observed in 50% (48/96) of samples, and low expression was observed in 16% (15/96) of cases. Figure 3 and Table 1 show the correlation between PLVAP expression and various clinicopathological parameters. The expression of PLVAP was not related to the age, gender, morphology, differentiation, tumor size, diabetes, cyst, Lauren type, H. pylori, mucus, necrosis, P53 levels, or Her-2 levels of STAD patients, but were related to TNM stage (P = 0.02), vascular invasion (P = 0.04), nerve invasion (P = 0.01), and Ki-67 (P = 0.01).

Fig. 3.

PLVAP protein expression in STAD (EnVision 100). A PLVAP: 3 + ; B PLVAP:1 + 

Table 1.

Correlation between PLVAP expression and clinicopathological characteristics of STAD

Clinicopathological characteristics	PLVAP			P	χ2
	−/1 +	2 +	3 +
Age(year)	68.7 ± 9.2	65.9 ± 8.7	69.3 ± 12.4	0.61	–
Gender
Male	9	25	26	0.13	4.07
Female	6	8	22
Morphology
Eminence	5	15	12	0.16	3.66
Ulcer	10	18	36
Differentiation
Middle/poor	8	22	30	0.68	0.77
High	7	11	18
Tumor size	5.4 ± 1.8	4.0 ± 1.6	4.6 ± 2.0	0.23	–
Vascular invasion
Yes	7	15	34	0.04	6.25
No	8	18	14
Nerve invasion
Yes	4	10	28	0.01	8.50
No	11	23	20
Diabetes
Yes	9	16	25	0.76	0.55
No	6	17	23
Cyst
Yes	8	13	19	0.61	0.98
No	7	20	29
Lauren type
Intestinal	5	11	10	0.08	8.22
Diffuse	6	17	17
Mixed	4	5	21
Hp
Yes	6	9	18	0.56	1.18
No	9	24	30
Mucus
Yes	5	12	17	0.98	0.04
No	10	21	31
Necrosis
Yes	6	14	21	0.97	0.07
No	9	19	27
TNM
I/II	9	7	11	0.02	8.10
III/IV	6	26	37
P53
0–1	2	10	10	0.31	4.76
2	5	5	16
3	8	18	22
Her-2
0 ~ 1	10	28	41	0.27	2.61
2 ~ 3	5	5	7
MVD	20.7 ± 2.1	27.0 ± 3.6	30.7 ± 1.5	0.01	–
Ki-67
0 ~ 50%	9	14	10	0.01	9.25
50% ~	6	19	38

Correlation between PLVAP protein expression and MVD (microvessel density)

To investigate whether PLVAP exerts a stimulative effect on tumor-associated blood vessel formation and growth, we estimated the intratumoral microvessel density (MVD) by CD34-positive staining, which was mainly detected in the vascular endothelial cell membrane. MVD was significantly higher in the PLVAP-rich group (3+) than in the PLVAP-poor group (−/1 +) (P = 0.01; Table 1; Fig. 4), indicating that PLVAP may pose a facilitator effect on tumor-associated blood vessel formation and growth, at least partially.

Fig. 4 .

CD34 protein expression in STAD (EnVision 100). A CD34 protein expression in STAD with PLVAP 3 + ; B CD34 protein expression in STAD with PLVAP:1 + 

Correlation between PLVAP protein expression and LPS

Previous studies have confirmed the presence of bacteria in human tumors (Nejman et al. 2020). To evaluate the bacterial distribution in STAD initially, we examined LPS protein expression by IHC. The relationship between the LPS expression and the clinicopathological features was analyzed in 96 STAD patients (Fig. 5, Supplementary Table S1). The upregulation of LPS tended to be associated with the TNM stage (P = 0.01), Ki-67 (P = 0.01). Significant correlation between LPS expression and age, gender, morphology, differentiation, tumor size, diabetes, cyst, Lauren type, H. pylori, mucus, necrosis, P53 levels, and Her-2 levels was not observed.

Fig. 5.

LPS protein expression in STAD (EnVision 100). A LPS: 3 + ; B LPS:1 + 

To assess the correlation between LPS  and PLVAP protein expression in STAD tissues, Pearson’s bivariate correlation test was applied to analyze the association between PLVAP- and LPS-related proteins (Table 2). There were moderately significant associations of PLVAP and LPS (r = 0.36; P = 0.01). This result suggested that bacteria may be able to promote the expression of PLVAP protein.

Table 2.

Correlation between LPS and PLVAP protein expression of STAD

LPS	PLVAP			R	P
	− /1 +	2 +	3 +
− /1 +	3	5	9	0.36	0.01
2 +	2	20	15
3 +	10	8	24

The microbiota in the PLVAP-rich group was significantly different from that in the PLVAP-poor group

To further investigate whether there was a difference between the gut microbial communities of the PLVAP-rich group (3+) and PLVAP-poor group (−/1+), we employed 16S ribosomal RNA (16 S rRNA) gene sequencing of 15 fixed, paraffin-embedded STAD samples (10 samples with the PLVAP-rich group, and the 5 samples as the PLVAP-poor group). Supplementary Figure S2 displays all samples with V3-V4 sequencing using 16 S rRNA, with operational taxonomic units (OTUs) based on 97% sequence similarity. Alpha rarefaction plots of observed species were constructed to determine that adequate sequence coverage was obtained to reliably describe the full diversity present in these samples. The results showed that there were significant differences between PLVAP-rich and PLVAP-poor groups (P < 0.05). The core salivary microbiome of both PLVAP-rich patients and poor controls comprised 919 species (462 species shared by the 2 groups, while 457 differed). At the phylum level (Fig. 6A), Fusobacteriota (PLVAP-rich group: 11.48%; PLVAP-poor group: 0.16%; P = 0.02) and Spirochaetota (PLVAP-rich group: 0.96%; PLVAP-poor group: 0.08%; P = 0.01). At the class level (Fig. 6B), Fusobacteriia (PLVAP-rich group:11.48%; PLVAP-poor group:0.16%; P = 0.02), Clostridia (PLVAP-rich group: 4.72%; PLVAP-poor group: 0.38%; P = 0.01), Negativicutes (PLVAP-rich group: 1.70%; PLVAP-poor group: 0.62%; P = 0.04), Spirochaetia (PLVAP-rich group: 0.96%; PLVAP-poor group: 0.08%; P = 0.01), Coriobacteriia (PLVAP-rich group: 0.73%; PLVAP-poor group: 0.04%; P = 0.02). At the genus level (Fig. 6C), Prevotella (PLVAP-rich group: 6.08%; PLVAP-poor group: 0.22%; P = 0.01), Acinetobacter (PLVAP-rich group: 1.65%; PLVAP-poor group: 3.68%; P = 0.04), Lactobacillus (PLVAP-rich group: 3.30%; PLVAP-poor group: 0.15%; P = 0.02), Lachnoanaerobaculum (PLVAP-rich group: 1.73%; PLVAP-poor group: 0.02%; P = 0.03), Treponema (PLVAP-rich group: 0.91%; PLVAP-poor group: 0.08%; P = 0.01).

Fig. 6.

Differences in microbial community composition between the PLVAP-rich and PVLAP-poor groups at phylum, class, and genus level. A at phylum level; B at class level; C at the genus level. Rich PLVAP-rich group, Poor PLVAP-poor group

Predicted functional composition of gut microbiome with PICRUSt

To determine whether changes in the composition of the microbiome in different groups (PLVAP-rich group; PLVAP-poor group) affect microbiome function, we inferred microbiome function using PICRUSt. COG analysis indicated the participation of identified proteins in a diverse number of biological processes, including amino acid transport and metabolism, energy production and conversion, carbohydrate transport and metabolism, transcription, translation, ribosomal structure, and biogenesis. Multiple microbial functions were disturbed in the PLVAP-rich group, including metabolism, cellular processes, and environmental information processing at KEGG level 1 (Fig. 7A–C); amino acid metabolism, lipid metabolism, carbohydrate metabolism at KEGG level 2 (Fig. 7D–F); glycolysis, apoptosis at KEGG level 3 (Fig. 7G, H).

Fig. 7.

Functional alterations of the stomach adenocareinoma microbiome between the PLVAP-rich group（n=10）and PLVAP-poor group（n=5）using PICRUSt2. A cellular processes, B environmental information processing, C metabolism; D metabolism: amino acid metabolism, E metabolism: lipid metabolism, F metabolism: carbohydrate metabolism; G carbohydrate metabolism: glycolysis, H cell growth and death: apoptosis  PICRUSt2 ***Phylogenetic Investigation of Communities by Reconstruction of Unobserved States 2

Co‑expression and neighboring genes of PLVAP gene.

Based on the above results, we applied the GeneMANIA and STRING databases to predict PPIs for PLVAP (Fig. 8A, B). As shown in Fig. 8A, 20 proteins/genes might interact with PLVAP, and 3 genes including fucosyltransferase 8(FUT8), sphingomyelinase(SMPD2), and neuropilin-1(NRP1) have been studied to be involved in tumor progression (Tu et al. 2017; Wei et al. 2021). The PPI network by using STRING database contained 11 nodes and 25 edges and the average node degree was 5 (P = 0.01) (Fig. 8B). Subsequently, the present study constructed a network analysis of the neighboring genes of PLVAP using the GEPIA2 database. The results are tabulated in Supplementary Table S2.

Fig. 8.

The network of PLVAP and its co‑expression genes was set up visually. A Gene–gene interaction of PLVAP with other genes proposed by GeneMANIA. B Network map of ten genes induced by the stomach adenocarcinoma using STRING database

High expression of PLVAP was associated with high infiltration levels of immune cells

On the basis of the above results, we investigated the relationship between PLVAP and immune cells. The results from TIMER showed that high PLVAP mRNA expression had significant positive correlations with CD4 + T cell (r = 0.42, P < 0.001), CD8 + T cell (r = 0.21, P < 0.001), macrophage (r = 0.29, P < 0.001), neutrophil (r = 0.21, P < 0.001), B cell (r = 0.03, P < 0.001), and dendritic (r = 0.24, P < 0.001) levels in STAD (Fig. 9). Overall, the PLVAP was moderately correlated with CD4 + T cell.

Fig. 9.

High expression of PLVAP was associated with high infiltration levels of immune cells

Furthermore, the correlations between immune cell markers and PLVAP mRNA expression were identified using correlation modules in GEPIA2. In TIMER2, PLVAP mRNA expression was significantly associated with 19 of 32 immune cell markers in STAD (Table 3). Notably, the correlations between PLVAP and immune cells were similar to those found using TIMER2. PLVAP mRNA expression in STAD was moderately correlated with Th9 cells, Th17 cells, Treg, and dendritic. Treg and Th17 cells belonged to CD4 + T cells. Therefore, these results further confirm the findings that PLVAP is specifically correlated with immune infiltrating cells in STAD, demonstrating that PLVAP has a vital role in immune escape in STAD.

Table 3.

Correlation between PLVAP and related genes and markers of immune cells in TIMER2

Cell type	Gene marker	Cor	P
B cell	CD19	0.12	**
	CD38	0.15	****
CD8 + T cell	CD8A	0.21	****
	CD8B	0.03	0.43
Tfh	CXCR5	0.23	****
	ICOS	0.22	****
Th1	IL12RB2	0.11	0.01
	T-BET	0.22	****
Th2	CCR3	0.09	0.02
	STAT6	0.18	****
Th9	TGFBR2	0.49	*
	IRF4	0.22	****
	PU.1	0.31	****
Th17	STAT3	0.32	****
Th22	CCR10	0.17	****
	AHR	0.21	****
Treg	FOXP3	0.32	****
	CCR8	0.26	****
T cell	CTLA4	0.02	0.69
Macrophage	CD68	0.18	****
	CD11b	0.28	****
M1	NOS2	− 0.02	0.65
	ROS	0.03	0.47
M2	ARG1	0.05	0.22
	MRC1	0.21	****
TAM	HLA-G	0.04	0.35
	CD80	0.17	****
	CD86	0.18	****
Neutrophil	MPO	− 0.01	0.93
NK	KIR2DL1	0.10	0.08
	KIR2DL3	0.19	0.07
	KIR2DL4	− 0.03	0.51
	KIR3DL1	− 0.03	0.51
	KIR3DL2	− 0.03	0.51
	KIR3DL3	− 0.01	0.78
	KIR2DS4	0.02	0.71
DC	CD1C	0.31	****
	CD141	0.38	*

Tfh follicular helper T cell, Th T helper cell, Treg regulatory T cell, TAM tumor-associated macrophage, NK natural killer cell, DC dendritic cell, None correlation without adjustment, Purity correlation adjusted for tumor purity, Cor R value of Spearman’s correlation

*P < 0.01; **P < 0.001; ***P < 0.0001, ****P < 0.0001

Discussion

PLVAP is type II integral membrane glycoprotein with a molecular weight of ~ 60 kDa (Guo et al. 2016). PLVAP was shown to be a preferable therapeutic target for cancer therapy since the administration of PLVAP antibodies effectively suppressed tumor growth and had minimal systemic toxicity (Wang et al. 2014). Previous reports demonstrated that PLVAP expression was upregulated in the endothelial cell of the tumor (Strickland et al. 2005). Recently, it is reported that PLVAP is highly expressed on cholangiocarcinoma cells (Wang et al. 2021). In our study, we found that the high PLVAP protein level correlated with III/IV stage, vascular invasion, nerve invasion, and high Ki-67 level in STAD patients. Moreover, a high PLVAP protein level was significantly expressed in STAD compared to normal tissues. In addition, a significant high expression level of PLVAP mRNA was detected in STAD tissue compared with matched para-carcinoma tissue in the TIMER and UALCAN database. These results suggest that PLVAP may be involved in tumorigenesis and progression. Using the KM plotter database, we found that high mRNA expression of PLVAP was significantly correlated with poor OS in patients with STAD. However, GEPIA database analysis results showed there was no significant difference between the two groups. These findings suggest that PLVAP may act as a novel prognosis marker for STAD patients.

In this study, 20 proteins/genes might interact with PLVAP, and 3 genes including FUT8, SMPD2, and NRP1 have been studied to be involved in tumor progression. A study has revealed a positive feedback mechanism of FUT8-mediated receptor core fucosylation that promotes TGF-β signaling and EMT, thus stimulating breast cancer cell finvasion and metastasis (Tu et al 2017). Downregulation of NRP1 contributes to Bladder urothelial carcinoma progression, which is associated with activation of MAPK signaling and molecular mechanisms involved in cancer pathways (Dong et al 2021). STRING network analysis for PLVAP demonstrated a higher confidence interaction with CLDN5. CLDN5 regulates the permeability of the blood-brain barrier by regulating the proliferation, migration, and permeability of human brain vascular endothelial cells, which was involved in reducing the formation of lung cancer brain metastasis (Ma et al. 2017).

Further analyses suggested that the MVD value was significantly higher in the PLVAP high group when compared with the PLVAP-poor patients. Spearman correlation coefficient analysis demonstrated a positive correlation between the PLVAP IHC score and MVD value in STAD samples. This point is consistent with the results of previously published studies (Wang et al. 2021). Notably, high levels of PLVAP protein have been linked to vascular invasion of cancer cells. This might have been due to the following reasons: the more the number of vessels in a tumor, the more active the tumor growth is, the higher the malignancy of the tumor. PLVAP promotes the new blood vessel formation and increases vascular permeability resulting in the growth and metastasis of tumor cells; on the other hand, the greater the number of tumor vessels, the greater the probability that cancer cells enter blood circulation, the greater the probability of tumor metastasis.

Metastasis is facilitated by the formation of a “premetastatic niche,” which is fostered by primary tumor-derived factors. A previous study showed that the increased levels of PLVAP are associated with liver bacteria dissemination and metachronous distant metastases (Bertocchi et al. 2021). The investigators (Bertocchi et al. 2021) observed a statistically more bacterial abundance in PLVAP-rich than in PLVAP-poor CRC patients. To observe the difference between PLVAP expression level and bacterial abundance in STAD tissues, we detected paraffin samples of STAD via IHC, and the results showed that the high level of PLVAP protein expression was closely related to bacteria.

Subsequently, paraffin samples from 15 STAD patients were selected for bacterial DNA extraction and it was found that the increased bacterial abundance was higher in the PLVAP-rich group than in the PLVAP-poor group. Although H. pylori infection has been identified as the strongest risk factor for STAD, H. pylori often become undetectable (Wang et al 2016). This is confirmed by the results of the present study. Moreover, we also observed a higher abundance of Fusobacteriia in STAD tissues with high levels of PLVAP protein. Unexpectedly, Fusobacteriia may be involved in STAD progression (Hsieh et al. 2018). In recent years, some molecular mechanisms by which F. nucleatum contributes to colorectal cancer (CRC) progression have been reported. F. nucleatum promotes CRC growth through its FadA adhesin and the autotransporter protein Fap2. FadA allows bacteria attachment and induction of human CRC cell proliferation in a FadA-dependent β-catenin signaling activation and proinflammatory response associated with NF-κβ upregulation (Rubinstein et al. 2013). Another FadA function is vascular endothelial VE-cadherin removal from cell–cell junctions, increasing endothelial cell permeability, thus allowing bacteria to cross junctions (Fardini et al. 2011).

Bacteria modulates the host biology through direct cell interaction, as well as microbial-derived metabolites. PICRUSt2 analysis revealed that the microbial functions involved in cellular processes, environmental information processing, and metabolism were increased in PLVAP-rich patients. These microbial functional alterations at various levels were partly consistent with microbial changes in PLVAP-rich patients. Some specific metabolites can reflect the metabolic characteristics of the individual and the disease (Beger et al. 2016). In our study, the metabolites of different amino acids in the PLVAP-rich group and poor group mainly included the biosynthesis of valine, leucine, and isoleucine, phenylalanine, tyrosine, histidine, glycine, threonine, and serine; lipid differential metabolites mainly include unsaturated fatty acid synthesis, linoleic acid, and fatty acid biosynthesis; carbohydrate metabolism differential metabolites mainly include glyoxylate, dicarboxylic acid, pyruvate, glycolysis, and pentose and glucuronate interconversions. Lactobacillus produces lactate, which may serve as a fuel for the tumor cells, accelerating their growth (Vinasco et al. 2019). In a need of rapid growth, tumor cells rely primarily on anaerobic glycolysis rather than oxidative phosphorylation, which provides more lactic acid (Ward and Thompson 2012). It has been assumed that increased production of lactic acid by Lactobacillus promotes the growth of tumor cells. Prevotella is a Gram-negative anaerobic bacterium that helps break down proteins and carbohydrates (Fei et al. 2020). In our study, the relative abundance of Prevotella and Lactobacillus increased in the PLVAP-rich group compared with the PLVAP-poor group. Based on the studies above, we supposed that intratumoral bacteria might be involved in increased PLVAP detection. A recent study demonstrated that Salmonella typhimurium is capable of controlling PLVAP expression (Spadoni et al. 2016).

The low-biomass ecosystem may be composed of intratumor bacteria, thus affecting the tumor microenvironment including inflammatory mediators, such as tissue-resident and peripherally recruited immune cells (bone marrow cells, T cells, B cells, and NK cells), tumor-associated macrophage, fibroblast, endothelial cell. Our research suggested that PLVAP was significantly positively correlated with tumor CD4 + T cell, CD8 + T cell, neutrophil, dendritic, and macrophage infiltration. CD8 + T cells play an important role in immune-related tolerance and immunosuppression within TME (Iwahori 2020). The imbalance of the TME may partly contribute to the poor prognosis associated with PLVAP in cancers (Borst et al. 2018).

In conclusion, PLVAP is a potential biomarker to predict the prognosis of patients with STAD, and the high level of PLVAP protein expression was closely related to bacteria. The relative abundance of Fusobacteriia was positively associated with the level of PLVAP. Therefore, positive staining for PLVAP was useful for predicting the prognosis of STAD with  Fusobacteriia infection. To our knowledge, PLVAP has not been reported in association with STAD with Fusobacteriia infection.This study has the least two limitations. First, the number of specimens for microbial testing is very small, which affects the results may not be generalizable.. Second, future research is required to explore the detailed mechanism between distinct PLVAP in STAD and reveal the mechanism of PLVAP in other carcinomas.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

PLVAP

Plasmalemma vesicle-associated protein

STAD

Stomach adenocarcinoma

TCGA

The Cancer Genome Atlas

IHC

Immunohistochemistry

OUT

Operational taxonomic units

KEGG

Kyoto Encyclopedia of Genes and Genomes

DFS

Disease-free survival

OS

Overall survival

TME

Tumor microenvironment

16S rRNA

16S ribosomal RNA

KM

Kaplan–Meier

TIMER

Tumor immune estimation resource

MVD

Microvessel density

Author contributions

Yu.W. collected and analyzed all data, performed the statistics, drafted the initial manuscript, and reviewed the final version. Z.L. and C.H. were involved in data collection and reviewed the manuscript. Y.W. and Y.H. conceptualized and designed the study. Yu.W., Y.W., and Y.H. supervised the data collection, reviewed, and revised the manuscript.

Funding

This research is supported by Ninth Hospital of Xi’an (Grant Reference Number 2022qn02); The Foundation of Xi’an science and technology project (No. 21YXYJ0069); the science research project of Xi’an traditional Chinese medicine (No. SZY202203).

Data availability

The data underlying this study will be shared on request to the corresponding author.

Declarations

Conflict of interest

The authors declare that they have no competing interests.

Ethical approval and consent to participate

The ethics approval and consent to participate of the current study were approved and consented by the ethics committee of Ninth Hospital of Xi’an.