Clinical powers of Aminoacyl tRNA Synthetase Complex Interacting Multifunctional Protein 1 (AIMP1) for head-neck squamous cell carcinoma

Abstract

BACKGROUND:

Head and neck squamous cell carcinoma (HNSC) is one of the most common cancer types in the world. The study in molecular markers for HNSC prognosis is of great significance. We hypothesized that Aminoacyl tRNA Synthetase Complex Interacting Multifunctional Protein 1 (AIMP1), a gene that encodes a cytokine, is a critical biomarker for HNSC.

METHODS:

We acquired clinical data, mRNA expression data, protein staining data, and single-cell expression data of HNSC from open databases and evaluated the clinical prognostic value of AIMP1, and explored the potential roles of AIMP1 in HNSC biology and tumor immune microenvironment.

RESULTS:

AIMP1 was overexpressed in HNSC compared to normal tissues. Higher AIMP1 expression was associated with a worse survival rate. A survival nomogram was constructed for HNSC patients. One thousand two hundred and eighty-one genes were identified as positively associated with AIMP1 and enriched in proliferation-related terminologies, while 303 genes were identified as negatively associated with AIMP1 and enriched in terminologies related to skin development and immune cell regulation. AIMP1 was positively correlated with stemness, cell cycle, and DNA repair, and negatively correlated with angiogenesis, quiescence, metastasis, hypoxia, inflammation, EMT, DNA damage, and invasion in single cells. AIMP1 was expressed higher in malignant cells than immune cells and there was no difference in AIMP1 expression among immune cell types. AIMP1 high group had a lower immune score, stroma score, and microenvironment score.

CONCLUSION:

AIMP1 is a potential diagnostic and prognostic biomarker for HNSC patients and can potentially affect the proliferation and tumor immune microenvironment of HNSC cells. This study provided a novel molecular marker for the improvement of clinical HNSC treatment.

Keywords

AIMP1 HNSC gene expression survival immune cells

1. Introduction

Head and neck squamous cell carcinoma (HNSC) is one of the most common cancer types in the world. Each year, more than five hundred thousand new HNSC cases occur, which lead to about 3 hundred thousand deaths worldwide [1]. Traditional treatment methods for HNSC include tumor surgical treatment, radiotherapy, and chemotherapy, but the therapeutic effect of these therapies is not ideal [2]. Although in recent years, researchers have developed several novel methods to improve the therapy of HNSC, these efforts failed to improve the overall survival rate of HNSC patients over the past several years [2]. One of the most challenging issues in HNSC treatment is the late diagnosis of the disease, in which cases most patients have occurred lymph node metastasis and more recurrences that result in a high mortality rate of patients [3]. Therefore, the study in molecular markers that can improve the reliability and effectiveness of HNSC diagnosis and prognosis is of great significance. Understanding the role of these genes in the biological process of HNSC can also facilitate the development of HNSC therapy.

Aminoacyl tRNA Synthetase Complex Interacting Multifunctional Protein 1 (AIMP1) is a gene that encodes a cytokine that is specifically induced by apoptosis [4]. AIMP1 protein is a member of the small inducible cytokine subfamily E and can regulate wound healing, inflammation, and angiogenesis [5, 6, 7]. The precursor protein of AIMP1 is identical to the p43 protein, which is involved in the multi-tRNA synthetase complex and the modulation of aminoacylation activity of tRNA synthetase [8, 9, 10]. These indicated that AIMP1 was involved in the expression of proteins.

Recent studies revealed additional roles of AIMP1 in cancer cells. Studies showed that the release of the AIMP1 cytokine renders the tumor-associated vasculature sensitive to tumor necrosis factor [5]. AIMP1 has also been reported to regulate multiple immune cells in cancers, including NK cells, Macrophages, T-cells, etc. [11, 12, 13]. A study has suggested that AIMP1 is an exogenous anti-tumor agent in a lung cancer xenograft model [14]. As for HNSC, a mass spectrometric analysis revealed that AIMP1 can bind to fascin actin-bundling protein 1 (FSCN1), a critical protein for squamous cell carcinoma which enhances cell proliferation, migration, and invasion [15]. Although the roles of AIMP1 in cancers have not been clearly identified, the expression and the presence of AIMP1 in cancer can be potentially valuable for clinical cancer treatment. AIMP1 has been previously identified as one of the prognostic signatures that might be associated with prognosis and might promote cell proliferation and tumor progression [16]. However, so far, the clinical application of AIMP1 in cancers remains lacking. The exact role of AIMP1 in HNSC required further studies.

Here we identified AIMP1 as a critical biomarker for clinical HNSC. In this study, we tested AIMP1 as a prognostic biomarker for HNSC patients. We also proposed that AIMP1 is critical for the cancer biology and tumor immune microenvironment of HNSC cells. Thus, we evaluated the clinical prognostic value of AIMP1 and explored the potential roles of AIMP1 in HNSC biology and tumor immune microenvironment. This study provided a novel molecular marker for the improvement of clinical HNSC treatment.

2. Methods

2.1 RNA-seq and clinical data acquisition

Clinical and genomic data of HNSC were downloaded from The Cancer Genome Atlas (TCGA, http:// cancergenome.nih.gov/) in January 2020, in which the method of acquisition and application complied with the guidelines and policies.

2.2 RNA-seq data analysis

All the analysis methods and R package were implemented by R foundation for statistical computing (2020) version 4.0.3 and ggplot2 (v3.3.2). The xCell algorithms were used to estimate the immune cell infiltration levels [17]. A $P$ -value $<$ 0.05 was considered statistically significant.

2.3 Multiple data overexpression analysis

Multiple data sets of the mRNA overexpression and DNA copy number gain of AIMP1 in HNSC were accessed and analyzed using the Oncomine [18].

2.4 AIMP1 associated genes enrichment analysis

Differentially expressed genes between AIMP1 high (75–100%) and low (0–25%) groups were identified. The cutoff values of 1.5 and 0.001 were set for fold change and $P$ -value respectively. The volcano plot and heat map were plotted for the significant genes. These genes were further enriched in GO terminologies and KEGG pathways.

2.5 Immunohistochemistry staining

Immunohistochemistry staining images of AIMP1 in normal squamous epithelial tissues, normal lymph nodes, and HNSC tissues were accessed from Human Protein Atlas (HPA, https://www.proteinatlas.org/). Antibody HPA018476 and CAB017618 were used to stain AIMP1. All images of tissues stained by immunohistochemistry were manually annotated by a specialist followed by verification by a second specialist. Annotation of each different normal and cancer tissue was performed using fixed guidelines for the classification of immunohistochemical results. Each tissue was examined for representability, and subsequently, immunoreactivity in the different cell types present in normal or cancer tissues was annotated. Basic annotation parameters include an evaluation of staining intensity (negative, weak, moderate, or strong) and the fraction of stained cells ( $<$ 25%, 25–75%, or $>$ 75%)

2.6 Single-cell sequencing data acquisition and analysis

Correlations between AIMP1 expression and functional states in HNSC single-cell datasets GSE103322 were accessed and analyzed using the CancerSEA (http://biocc.hrbmu.edu.cn/CancerSEA/). Single-cell sequencing data of immune cells and malignant cells HNSC_GSE103322 and HNSC_GSE139324 were accessed and analyzed using the Tumor Immune Single-cell Hub (TISCH, http://tisch.comp-genomics.org/). The gene expression level was displayed using UMAP (Dimensionality reduction via Uniform Manifold Approximation and Projection). Cell-type prediction and violin plot of AIMP1 expression in different cell types were conducted using TISCH. A $P$ -value $<$ 0.05 was considered statistically significant.

2.7 Immune cell infiltration analysis

The immune cell infiltration level was calculated using the TCGA cohort. We calculated the immune cell infiltration score of TCGA data using the xCell algorithms. The immune cell infiltration levels between AIMP1 high (75–100%) and low (0–25%) groups were compared.

2.8 Statistical analysis

T-test and one-way ANOVA were applied to compare gene expression differences. Chi-Squared tests were used to compare the distribution of HNSC patients with different clinicopathological variables in the AIMP1 high and low groups. Kaplan-Meier analysis, log-rank test, and multivariate cox regression test were used to conduct survival analysis. Pearson’s correlation test was conducted to evaluate the correlation of two variables. A $P<$ 0.05 was considered to be statistically significant.

3. Results

3.1 AIMP1 was overexpressed in HNSC

In this study, we investigated the mRNA expression of AIMP1 genes in HNSC. Results showed that in both the non-paired and paired comparison, the expression of AIMP1 was significantly higher in HNSC than in normal tissues (Fig. 1A and B). We also calculate the diagnostic ROC, results showed that the AUC of AIMP1 diagnostic ROC was 0.759, indicating that AIMP1 was a promising diagnostic biomarker for HNSC. To validated the overexpression of AIMP1 in HNSC, we also compared the mRNA overexpression or DNA copy number gain of AIMP1 in HNSC in 11 data sets using the Oncomine. Results showed that AIMP1 was significantly overexpressed or obtain copy number gain in HNSC compared to normal tissues among all of these data set (Fig. 1D). To investigate whether AIMP1 was overexpressed in HNSC compared to normal tissue at the protein levels, we compared immunohistochemistry staining images of AIMP1 in normal squamous epithelial tissues and HNSC tissues from Human Protein Atlas for reference (Fig. 2). To avoid false-positive resulted from inappropriate antibodies, we compared staining data of two different AIMP1 antibodies. The stainings of both of the antibodies showed that the staining intensity of AIMP1 in normal oral squamous epithelial tissues and normal esophagus squamous epithelial tissues were between weak to moderate (Fig. 2B1–B2, C1–C2), whereas the staining intensity of AIMP1 in HNSC tissues was between moderate to strong (Fig. 2A1–A2).

Figure 1.

The overexpression of AIMP1 in HNSC. A. Unpaired comparison of expression of AIMP1 in HNSC and normal tissues. B. Paired comparison of expression of AIMP1 in HNSC and normal tissues. C. Diagnostic ROC curve of AIMP1 for HNSC. TCGA data were plotted. D. The mRNA overexpression or DNA copy number gain of AIMP1 in HNSC. Multiple data sets were accessed and analyzed using the Oncomine. Heatmap and overall copy number difference analysis between HNSC and normal tissues. The references of the data were listed in the legend.

Figure 2.

Representative immunohistochemistry staining images of AIMP1 from Human Protein Atlas. Antibody HPA018476 (A1B1C1) and CAB017618 (A2B2C2) were used to stain AIMP1. A. HNSC tissues, B. normal oral squamous epithelial tissues. C. normal esophagus squamous epithelial tissues. Sample information and histochemistry results were shown on the right side of each image.

3.2 The expression of AIMP1 and clinicopathological features

In order to further explore the association between AIMP1 expression and clinicopathological features in HNSC, multiple groups of analyses were performed within the tumor. In this study, we found that grade III HNSC expressed significantly higher AIMP1 than grade I HNSC (Fig. 3A1). Although there was no significant difference between grade I and II or grade II and III, we can see a general trend that AIMP1 expressed higher in higher grade HNSC (Fig. 3A1). In addition, we also compared different pTNM stages, different pT stages, different pN stages, gender, different races, whether smoke, whether drink, whether with radiotherapy and whether HPV positive, and we found no significance in all these comparisons except for a slight difference between pTNM I and TNM IV (Fig. 3A2–9). We also plotted a Sankey diagram for AIMP1 expression and clinicopathological features of HNSC patients to present the basic clinical information of the patients included in this study (Fig. 3B). Detailed distribution data of AMIP1 expression median divided groups were presented in Table 1.

Figure 3.

The association between AIMP1 expression and clinicopathological features in HNSC. TCGA data were plotted. A1–10. AIMP1 distribution in HNSC and different sub-groups (A1. tumor grade, A2. pTNM stage, A3. pT stage, A4. pN stage, A5. gender, A6. race, A7. smoke, A8. drink, A9. radiotherapy, A10. HPV status) of HNSC. B. Sanguini diagram of age, gender, race, AIMP1, and survival.

Table 1

Distribution of HNSC patients with different clinicopathological variables in AIMP1 high and low groups

Term	AIMP1 high	AIMP1 low	$P$ -value
	(50–100%)	(0–50%)
Survival
Alive	130	154
Dead	121	97	0.038
Age mean (SD)	60.7 (11.6)	61.4 (12.2)	0.524
Gender
FEMALE	60	74
MALE	191	177	0.19
Tumor staging
Tumor Lymph Node Metastasis Stage
T1	15	19
T2	67	77
T3	65	68
T4	14	11
T4a	77	75
T4b	2	1
TX	11		0.888
N0	110	131
N1	41	40
N2	9	10
N2a	10	8
N2b	35	41
N2c	28	13
N3	4	3
NX	14	5	0.086
M0	230	247
M1	3	2
MX	18	2	0.001
I	8	17
II	38	43
III	44	46
IVA	152	138
IVB	8	5
IVC	1	2	0.381
Tumor differentiation stage
G1	23	39
G2	150	150
G3	66	53
G4	2
GX	8	8	0.137
Therapy
Non-radiation	39	23
Radiation	78	43	0.964
Neoadjuvant	8	2
No neoadjuvant	243	249	0.11
Chemotherapy	74	85
Chemotherapy: Targeted Molecular therapy	6	1
Immunotherapy	1
Chemotherapy: Immunotherapy	3
Chemotherapy: Other. specify in notes	1
Chemotherapy: Targeted Molecular therapy Vaccine	1	0.1
Others
Metastasis	14	5
Primary	5	4
Recurrence	21	18	0.339
Non-smoking	52	59
Smoking	194	187	0.518

3.3 Survival prognostic analysis of AIMP1 gene in HNSC.

To explore the prognostic power of AIMP1 for clinical HNSC patients, we analyzed the association of AIMP1 expression and the overall survival of HNSC patients. Results show that there was significance between the high AIMP1 group and the low AIMP1 group (Fig. 4A). To further evaluate the prognostic value of AIMP1 for HNSC patients, we calculated the time-dependent receiver operating characteristic (ROC) analysis of the AIMP1 for overall survival prediction. Results showed that the area under the curve (AUC) for 3-, 5-, 10-year overall survival prediction were 0.651, 0.653, and 0.756 respectively (Fig. 4B). Therefore, we suggested that AIMP1 was a promising predictive factor for the overall survival of HNSC, especially for long-term prognosis. To apply the AIMP1 expression to the clinical prognosis of HNSC, we constructed a survival nomogram. Variables pT_stage, pN_stage, pTNM_stage, AIMP1 expression, gender, age, race, smoker, alcohol history, radiation therapy were taken into account (Fig. 4C). Detailed Cox regression results were displayed in Table 2. The prediction results of the nomogram calibration curves of 1-, 5-, 10-year overall survival were consistent with all patients’ observation results (Fig. 4D).

Table 2
Cox regression analysis of AIMP1 in HNSC

Characteristics	Total (N)	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	$P$ value	Hazard ratio (95% CI)	$P$ value
AIMP1	501	2.092 (1.474–2.969)	$<$ 0.001	2.346 (1.540–3.574)	$<$ 0.001
T stage	486
T1	33	Reference
T2	143	1.086 (0.568–2.074)	0.803
T3	131	1.461 (0.769–2.773)	0.247
T4	179	1.249 (0.665–2.344)	0.490
N stage	479
N0	238	Reference
N1	80	1.058 (0.728–1.539)	0.768	1.285 (0.826–2.000)	0.267
N2&N3	161	1.404 (1.038–1.900)	0.028	1.750 (1.200–2.550)	0.004
M stage	476
M0	471	Reference
M1	5	4.745 (1.748–12.883)	0.002	3.606 (1.023–12.704)	0.046
Radiation therapy	440
Yes	287	Reference
No	153	1.631 (1.203–2.212)	0.002	1.968 (1.390–2.785)	$<$ 0.001
Age	501
$>$ 60	256	Reference
$\leqslant$ 60	245	0.799 (0.610–1.046)	0.102
Gender	501
Female	134	Reference
Male	367	0.764 (0.574–1.018)	0.066	0.775 (0.548–1.096)	0.149
Race	485
White	428	Reference
Black or African American&Asian	57	1.470 (0.973–2.220)	0.067	1.172 (0.702–1.956)	0.544
Smoker	491
Yes	380	Reference
No	111	0.918 (0.656–1.285)	0.618
Alcohol history	490
Yes	332	Reference
No	158	1.051 (0.790–1.397)	0.733

Figure 4.

Prognostic analysis of AIMP1 gene in HNSC. A. Kaplan-Meier survival analysis of High (50–100%) and low (0–50%) AIMP1 groups. log-rank test p-value was shown. B. Time-dependent receiver operating characteristic (ROC) analysis of the AIMP1 overall survival prediction. The area under the curve (AUC) and 95% confidential interval (95% CI) were shown. C. Nomogram for prediction of 1-, 5- and 10-year survival. D. Calibration plots of the nomogram.

Figure 5.

AIMP1 associated genes enrichment analysis. A. Differential gene analysis in HNSC AIMP1 high (75–100%) and low (0–25%) groups. Left panel: Volcano plots were constructed using fold-change $>$ 1.5 and $P<$ 0.01. The red spots in the plot represent the gene positively associated with AIMP1 and the blue spots represent the gene negatively associated with AIMP1. Right panel: Heat map and hierarchical clustering analysis of AIMP1 associated genes. Top 25% AIMP1 (Tumor_G1) vs low 25%AIMP1 (Tumor_G2). B. The GO and the KEGG signaling pathways enrichment analysis of AIMP1 associated genes.

3.4 AIMP1 associated genes enrichment analysis

To explore the potential mechanisms involved in the effect of the AIMP1 gene on HNSC, we identified differentially expressed genes between AIMP1 high and low groups. Results showed that 1281 (up) and 303 (down) genes were identified as differentially expressed genes positively and negatively associated with AIMP1 in HNSC respectively. The heat map showed the clusters of samples and significant genes (Fig. 5). These genes were further enriched in GO terminologies and KEGG pathways. Results of KEGG enrichment showed that genes positively associated with AIMP1 were enriched in “Spliceosome”, “RNA transport”, “Cell cycle”, and “DNA replication”, while genes negatively associated with AIMP1 were enriched in “Ras signaling pathway”, “Cytokine-cytokine receptor interaction”, “IL $-$ 17 signaling pathway”, “Ether lipid metabolism”, and “Arachidonic acid metabolism”. In terms of GO enrichment, genes positively associated with AIMP1 were enriched in “ribonucleoprotein complex biogenesis”, “organelle fission”, “nuclear division”, “DNA replication”, “regulation of cell cycle phase transition”, “ncRNA metabolic process”, and “chromosome segregation”, while genes negatively associated with AIMP1 were enriched in “epidermis development”, “skin development”, “epidermal cell differentiation”, and “keratinocyte differentiation” (Fig. 5B).

3.5 AIMP1 associated functional states in HNSC

To further explore the potential role of AIMP1 in HNSC, we investigated a single-cell data set of HNSC. We analyzed the correlation of AIMP1 expression and cancer functional state scores of 2150 single HNSC cells and identified 11 functional states that significantly correlated with AIMP1 ( $P<$ 0.05). Results showed that AIMP1 was positively correlated with stemness, cell cycle, and DNA repair with coefficients of 0.2, 0.09, and 0.07 respectively. On the other hand, AIMP1 was negatively correlated with angiogenesis, quiescence, metastasis, hypoxia, inflammation, EMT, DNA damage, and invasion, with coefficients of $-$ 0.15, $-$ 0.13, $-$ 0.13, $-$ 0.12, $-$ 0.11, $-$ 0.09, $-$ 0.08, and $-$ 0.06 respectively (Fig. 6).

Figure 6.

Correlations between AIMP1 expression and functional states in HNSC single-cell datasets. The GSE103322 data ( $n=$ 2105) were accessed and analyzed using the CancerSEA. Scatter plot of the significant correlations between AIMP1 expression and functional states. The X-axis represented the expression of AIMP1. The Y-axis represented gene signatures of corresponding functional states. Grey points were not considered to compute the correlations. The correlation factors and significance were presented. ${}^{***}P<$ 0.001; ${}^{**}P<$ 0.01; ${}^{*}P<$ 0.05.

Figure 7.

The distribution of AIMP1 expression in different types of cells in HNSC samples. The data were accessed and analyzed using TISCH (http://tisch.comp-genomics.org/). A1-3. HNSC_GSE103322. B1-3. HNSC_GSE139324. The gene expression level was displayed using UMAP (Dimensionality reduction via Uniform Manifold Approximation and Projection). A1B1. Cell-type prediction. A2B2. AIMP1 expression in single cells. A3B3. Violin plot of AIMP1 expression in different cell types. A full list of abbreviations can be found from TISCH.

Figure 8.

Immune cell infiltration levels of high AIMP1 (75–100%, G1) vs. low AIMP1 (0–25%, G2) groups. A. Immune cell score (xCell) heat map for AIMP1 high and low groups. The significance was analyzed using the Wilcox test. ${}^{*}P<$ 0.05, ${}^{**}P<$ 0.01, ${}^{***}P<$ 0.001. B. The percentage abundance of tumor-infiltrating immune cells in each sample.

3.6 Effect of AIMP1 on immune cell infiltration level in HNSC

To study the impact of AIMP1 expression on immune cells in HNSC, we first compared AIMP1 expression in different types of cells in HNSC using single-cell seq data. We analyzed two data sets that cover most of the immune cell types in HNSC. Results showed that all of the cell types in HNSC expressed AIMP1. The violin plot of AIMP1 expression in different cell types revealed that malignant (tumor cells) was one of the cell types highly expressing AIMP1 and there was no remarkable difference among immune cell types (Fig. 7). Therefore, we suggested that the expression level of AIMP1 in HNSC samples was mainly dependent on the expression of AIMP1 in tumor cells. Subsequently, we calculated the immune cell infiltration score of TCGA data using the xCell algorithms. We compared immune cell infiltration levels between AIMP1 high (75–100%) and low (0–25%) groups. Results showed that AIMP1 high group had a significantly lower total immune score, stroma score, and microenvironment score. Specifically, compared to the AIMP1 low group, the AIMP1 high group was only significantly higher in the level of Common lymphoid progenitor and T cell CD4 $+$ Th2. However, AIMP1 high group was significantly lower at levels of mast cell, Common myeloid progenitor, Granulocyte-monocyte progenitor, Monocyte, Neutrophil, T cell CD4 $+$ effector memory, cell plasma, B cell naïve, B cell, B cell memory, T cell CD4 $+$ naïve, Class-switched memory B cell, T cell CD4 $+$ memory, Endothelial cell, Hematopoietic stem cell, Plasmacytoid dendritic cell, T cell CD8 $+$ , Myeloid dendritic cell activated, Myeloid dendritic cell, and T cell CD4 $+$ central memory (Fig. 8). These results suggested that AIMP1 was negatively associated with infiltration levels of most immune cells and might negatively regulate immune and stroma in HNSC.

4. Discussion

This study, for the first time, revealed that AIMP1 was overexpressed in HNSC compared to normal squamous epithelial tissues at both mRNA level and protein levels. The expression of AIMP1 is also associated with higher grade HNSC. Hence, AIMP1 was identified to have not only potential diagnostic but also prognostic values for clinical HNSC treatment. However, a previous study reported that AIMP1 was downregulated in gastric and colorectal cancer [19]. This did not agree with the overexpression of AIMP1 in HNSC. Thus, we suggested the expression pattern and role of AIMP1 might vary from cancer types to cancer types. So far, the expression pattern of AIMP1 in other types of cancers has not been studied, and more effort to study the pan-cancer expression pattern of AIMP1 is required in the future. Nevertheless, the overexpression of AIMP1 in HNSC suggested that it can be a potential diagnostic biomarker for HNSC. A ROC is to evaluate the sensitivity, specificity, and accuracy of a factor. In general, for a diagnostic ROC, an AUC of 0.5 suggests no discrimination (i.e., ability to diagnose patients with and without the disease or condition based on the test), 0.7 to 0.8 is considered acceptable, 0.8 to 0.9 is considered excellent, and more than 0.9 is considered outstanding. The AUC of AIMP1 diagnostic ROC was 0.7 to 0.8, which is acceptable for diagnosis.

Another clinical valuable finding of this study was that AIMP1 expression is significantly associated with HNSC patient’s overall survival. This was demonstrated in both Kaplan-Meier survival analysis and Cox regression analysis. The Cox regression analysis also indicated that AIMP1 expression was independent of age, gender, grade, race, and other clinical factors. In this study, ROC analysis revealed that the AUC of AIMP for HNSC overall survival was over 0.65, and up to 0.75. Generally speaking, for a single prognostic biomarker, an AUC of over 0.65 is acceptable. To increased the accuracy of the biomarker, multiple clinical factors should be also considered. Hence, we not only identified AIMP1 association with survival but also constructed a nomogram with multiple clinical factors to demonstrate the feasibility of using AIMP1 for survival prediction. The nomogram showed considerable confidence, but more variables, such as other reliable biomarkers, should be included to optimize the accuracy of the prediction.

AIMP1 has been suggested to be an exogenous anti-tumor agent for lung cancer [14]. A later study claimed that AIMP1 might suppress gastric and colorectal cancer [19]. Although our results showed that AIMP1 was associated with worse survival of HNSC patients and AIMP1 expression might increase during the development of HNSC, it was not clear what role AIMP1 played in HNSC. Nevertheless, the enrichment study provided some hints for the potential roles of AIMP1 in HNSC. The increased AIMP1 was found associated with multiple proliferation-related terminologies and pathways, such as cell cycle and DNA replication. These results suggested that AIMP1 might be involved in the proliferation of HNSC cells. In addition, the decreased AIMP1 was negatively associated with several terminologies related to skin development and immune cells, which suggested that AIMP1 might deteriorate normal skin development and inhibited immunity. This is rational as AIMP1 is associated with more severe cancer, where cancer might inhibit normal cells.

Single-cell data also revealed some potential roles of AIMP1 in HNSC. The AIMP1 expression was significantly correlated with stemness, cell cycle, and DNA repair which was consistent with the enrichment analysis, further confirming that AIMP1 was associated with cancer proliferation. At the same time, the single-cell data also showed that AIMP1 was negatively correlated with angiogenesis, quiescence, metastasis, hypoxia, inflammation, EMT, DNA damage, and invasion, we suggested some of these states were correspond to the weakened immunity. However, to be admitted, the coefficients of most of these correlations were mostly lower than 0.2, indicating that the correlation of AIMP1 was not very strong. In addition, whether the changes in these states were the reasons or the results of AIMP1 was not clear. To date, the exact role of AIMP1 in HNSC has not been investigated, we urged more functional experiments in the future to validate our conclusion.

Another interesting finding of this study was that AIMP1 might affect the tumor immune microenvironment. Tumor immune single-cell data showed that HNSC tumor cells tended to express a higher level of AIMP1 than immune cells, and there was no difference in AIMP1 expression among different types of immune cells. These results indicated that a higher level of AIMP1 in HNSC samples of TCGA should result from higher levels of AIMP1 in tumor cells, but not any certain types of immune cells. This can partly rule out the bias in the comparison of immune infiltrations because the immune cells will not affect the division of the high and low AIMP1 groups. The immune infiltration analysis revealed that AIMP1 was negatively associated with the immune cell infiltration level, which was consistent with the enrichment results. The only two positive associated immune cell types were lymphoid progenitor and T cell CD4 $+$ Th2 (T helper cell), which often presented when the immunity was not activated [20]. The regulatory effect of AIMP1 on immune cells has been reported in both cancer and non-cancer cells. For example, a study has identified AIMP1 as an important regulator in glioma immunity [21]. It has also been reported that the absence of AIMP1 in bone marrow-derived dendritic cells reduced Th1 polarization of T cells in melanoma [11]. Yet, our analysis failed to find a difference of T cell CD4 $+$ Th1 between high and low AIMP1 groups. This indicated again that the expression pattern and role of AIMP1 might vary from cancer type to cancer type. In addition, AIMP1 was reported to activate natural killer (NK) cells through macrophages, thereby inhibited lung metastasis of melanoma cells [13]. This was consistent with the result in the single-cell statement that AIMP1 was negatively correlated with HNSC metastasis. The notion of NK cell activation by AIMP1 does not fit the observed poor survival of HNSC patients with high AIMP1 expression, however, as we can see in the heatmap, generally, the expression of NK cells in all HNSC samples were very low based on xCell calculation. The low level of NK cells prevents the significance of the correlation. Therefore, we suggested that the NK cells might not be critical in HNSC.

5. Conclusion

Availability of data and materials

The source of the raw data was provided in the paper and the raw analysis data of this study are provided by the corresponding author with a reasonable request.

Consent for publication

All authors had given final approval of the version to be published.

Funding

This study received funding from Shenzhen Traditional Chinese Medicine Hospital.

Ethical approval

Not applicable.

Authors’ contributions

Hengrui Liu contributed to the design of the study. Yixue Li and Hengrui Liu contributed to the data acquisition, data analysis, and the drafting of the manuscript. Hengrui Liu supervised the project.

Author information

The corresponding author of this study, Hengrui Liu, is a Principal Investigator in Biocomma Limited.

Footnotes

Acknowledgments

The authors thank the support of Biocomma Limited. We also thank Weifen Chen, Zongxiong Liu, Yaqi Yang, and Zirui Tian for their help.

Conflict of interest

The authors claimed that there is no conflict of interest.

References

Siegel

R.L.

Miller

K.D.

and Jemal

, Cancer statistics, 2019, CA Cancer J Clin 69 (2019), 7–34.

Miller

K.D.

Nogueira

Mariotto

A.B.

Rowland

J.H.

Yabroff

K.R.

Alfano

C.M.

Jemal

Kramer

J.L.

and Siegel

R.L.

, Cancer treatment and survivorship statistics, 2019, CA Cancer J Clin 69 (2019), 363–385.

Duprez

Berwouts

De Neve

Bonte

Boterberg

Deron

Huvenne

Rottey

and Mareel

, Distant metastases in head and neck cancer, Head Neck 39 (2017), 1733–1743.

Zou

Zhao

Zhu

Wang

Yao

Chen

Shi

Qian

Qiu

and Zheng

, The role of the AIMP1 pathway in diabetic retinopathy: AIMP1-targeted intervention study in diabetic retinopathy, Ophthalmic Res 63 (2020), 122–132.

Zhou

Sun

Huang

and Zhang

, Roles of aminoacyl-tRNA synthetase-interacting multi-functional proteins in physiology and cancer, Cell Death Dis 11 (2020), 579.

Han

J.M.

Park

S.G.

Lee

and Kim

, Structural separation of different extracellular activities in aminoacyl-tRNA synthetase-interacting multi-functional protein, p43/AIMP1, Biochem Biophys Res Commun 342 (2006), 113–118.

Park

S.G.

Kang

Y.S.

Ahn

Y.H.

Lee

S.H.

Kim

K.R.

Kim

K.W.

Koh

G.Y.

Y.G.

and Kim

, Dose-dependent biphasic activity of tRNA synthetase-associating factor, p43, in angiogenesis, J Biol Chem 277 (2002), 45243–45248.

Y.G.

Kang

Y.S.

Kim

E.K.

Park

S.G.

and Kim

, Nucleolar localization of human methionyl-tRNA synthetase and its role in ribosomal RNA synthesis, J Cell Biol 149 (2000), 567–574.

Wolfe

C.L.

Warrington

J.A.

Davis

Green

and Norcum

M.T.

, Isolation and characterization of human nuclear and cytosolic multisynthetase complexes and the intracellular distribution of p43/EMAPII, Protein Sci 12 (2003), 2282–2290.

10.

Kaminska

Havrylenko

Decottignies

Gillet

Le Maréchal

Negrutskii

and Mirande

, Dissection of the structural organization of the aminoacyl-tRNA synthetase complex, J Biol Chem 284 (2009), 6053–6060.

11.

Liang

Tian

You

Halpert

M.M.

Konduri

Baig

Y.C.

Paust

Kim

Jia

Huang

Zhang

Kheradmand

Corry

D.B.

Gilbert

B.E.

Levitt

J.M.

and Decker

W.K.

, AIMp1 Potentiates T(H)1 Polarization and Is Critical for Effective Antitumor and Antiviral Immunity, Front Immunol 8 (2017), 1801.

12.

Liang

Halpert

M.M.

Konduri

and Decker

W.K.

, Stepping out of the cytosol: AIMp1/p43 potentiates the link between innate and adaptive immunity, Int Rev Immunol 34 (2015), 367–381.

13.

Kim

M.S.

Song

J.H.

Cohen

E.P.

Cho

and Kim

T.S.

, Aminoacyl tRNA Synthetase-Interacting Multifunctional Protein 1 Activates NK Cells via Macrophages In Vitro and In Vivo , J Immunol 198 (2017), 4140–4147.

14.

Lee

Y.S.

Han

J.M.

Kang

Park

Y.I.

Kim

H.M.

and Kim

, Antitumor activity of the novel human cytokine AIMP1 in an in vivo tumor model, Mol Cells 21 (2006), 213–217.

15.

Gao

Xue

Zheng

Niu

Zhang

Liu

Zhang

Cui

Zhao

Wen

Thorne

R.F.

Zhang

and Wang

, Mass Spectrometric Analysis Identifies AIMP1 and LTA4H as FSCN1-Binding Proteins in Laryngeal Squamous Cell Carcinoma, Proteomics 19 (2019), e1900059.

16.

Tian

Wang

Qin

Zhu

Chen

Qin

Wei

Zhang

and Cai

, Identifying 8-mRNAsi Based Signature for Predicting Survival in Patients With Head and Neck Squamous Cell Carcinoma via Machine Learning, Front Genet 11 (2020), 566159.

17.

Aran

and Butte

A.J.

, xCell: Digitally portraying the tissue cellular heterogeneity landscape, Genome Biol 18 (2017), 220.

18.

Rhodes

D.R.

Shanker

Deshpande

Varambally

Ghosh

Barrette

Pandey

and Chinnaiyan

A.M.

, ONCOMINE: A cancer microarray database and integrated data-mining platform, Neoplasia 6 (2004), 1–6.

19.

Kim

S.S.

Hur

S.Y.

Kim

Y.R.

Yoo

N.J.

and Lee

S.H.

, Expression of AIMP1, 2 and 3, the scaffolds for the multi-tRNA synthetase complex, is downregulated in gastric and colorectal cancer, Tumori 97 (2011), 380–385.

20.

Zhu

, T helper cell differentiation, heterogeneity, and plasticity, Cold Spring Harb Perspect Biol 10 (2018).

21.

Cheng

Ren

Zhang

Cai

Liu

Han

and Wu

, Bioinformatic profiling identifies an immune-related risk signature for glioblastoma, Neurology 86 (2016), 2226–2234.

Clinical powers of Aminoacyl tRNA Synthetase Complex Interacting Multifunctional Protein 1 (AIMP1) for head-neck squamous cell carcinoma

Abstract

BACKGROUND:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Methods

2.1 RNA-seq and clinical data acquisition

2.2 RNA-seq data analysis

2.3 Multiple data overexpression analysis

2.4 AIMP1 associated genes enrichment analysis

2.5 Immunohistochemistry staining

2.6 Single-cell sequencing data acquisition and analysis

2.7 Immune cell infiltration analysis

2.8 Statistical analysis

3. Results

3.1 AIMP1 was overexpressed in HNSC

Table 2 Cox regression analysis of AIMP1 in HNSC

3.5 AIMP1 associated functional states in HNSC

4. Discussion

5. Conclusion

Availability of data and materials

Consent for publication

Funding

Ethical approval

Authors’ contributions

Author information

Footnotes

Acknowledgments

Conflict of interest

References

Table 2
Cox regression analysis of AIMP1 in HNSC