Identification of SRSF2 as a potential immunodiagnostic biomarker for osteosarcoma: A study based on focused protein microarray

Abstract

BACKGROUND

Osteosarcoma (OS) is a primary solid tumor of bone. Due to the insidious onset and malignancy of OS, it is necessary to diagnose OS earlier.

AIM

To search for an optimal diagnostic biomarker, such as autoantibody against tumor associated antigen (TAA), for early diagnosis of OS.

METHODS

Differential expression of biomarkers in sera from 28 patients with OS and 49 normal human sera were screened by a focused protein microarray with 154 human recombinant proteins. Enzyme-linked immunosorbent assay (ELISA) was applied to validate the three potential biomarkers from 49 OS patients matched with 49 normal controls and 39 osteochondroma. The diagnostic value of each TAA for OS was analyzed by receiver operating characteristic curve. Western blotting and immunohistochemistry were used to verify the results of ELISA.

RESULTS

Based on the protein microarray, three differentially expressed TAAs [G-protein subunit alpha 11, serine/arginine-rich splicing factor 2 (SRSF2), and phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha] were screened out, whose content was higher than in normal human sera. According to the results of ELISA, anti-SRSF2 autoantibody showed higher expression in OS than in normal controls, with sensitivity, specificity, and area under the curve of 16.33%, 95.92%, and 0.648, respectively. The ELISA results were confirmed by western blotting and immunohistochemistry.

CONCLUSION

Autoantibody is potential serological biomarker in detecting OS. Expression of anti-SRSF2 autoantibody has the potential to distinguish OS from the normal individuals.

Key Words: Osteosarcoma; Tumor associated antigen; Protein microarray; Detection; SRSF2; Biomarker

Core Tip: Osteosarcoma (OS) is, a common malignant tumor, with no specific serological biomarker for early diagnosis. In this study, we focused on identifying novel autoantibodies against tumor associated antigens using the focused protein microarray based on cancer-driving genes for detecting OS. Three tumor associated antigens were identified through the focused protein microarray. Expression of these three autoantibodies was further detected by enzyme-linked immunosorbent assay, which was further validated by western blotting and immunohistochemistry. Serine/arginine-rich splicing factor 2 was identified as a potential serological biomarker for OS, and a possible supplementary screening approach.

INTRODUCTION

Osteosarcoma (OS) is the most frequent primary solid malignancy of bone. OS is prone to occur in the growth spurting period of adolescence, and it is more common in the parts with the most vigorous growth[1]. The American Cancer Society has shown that, although the incidence of bone tumors is lower than that of other common malignant tumors, its incidence ranked eighth among adolescents aged 15-19 years, with mortality ranked second, and a 5-year survival rate of approximately 67%[2]. Incidence of OS ranks first among primary bone tumors[3]. OS progresses rapidly. About 20% of the patients have lung metastasis at the time of diagnosis[4-6]. However, due to the insidious onset of the disease, OS is easily confused with trauma and other diseases, and its high degree of malignancy can lead to metastasis at an early stage. Widely used detection methods include imaging examination (such as X-ray, computed tomography, and magnetic resonance imaging) and invasive pathological biopsy. However, imaging examination dose expose the body to radiation. Biopsy is invasive and traumatic. Although the current clinical application of neoadjuvant chemotherapy combined with extensive surgical resection has significantly improved the survival rate, chemotherapy has severe side effects, drug resistance and long treatment duration[3]. Therefore, it is urgent to find an optimal clinical biological marker for early diagnosis of OS.

Many studies have shown that a series of genetic information changes or abnormally expressed gene products can drive normal cells to develop into life-threatening tumor cells[7]. The expression of abnormal proteins is often accompanied by the occurrence and development of tumor. Some of these proteins, which are known as tumor associated antigens (TAAs)[8,9]. When TAAs have changes such as aberrant expression, overexpression, or mutation, and the immune system of tumor patients can recognize these abnormally expressed proteins and produce autoantibodies against TAA in the early stages of tumor development[9,10]. Autoantibodies against TAAs have a high serum titer, persist for a long time, and can be detected several months or even years before clinical diagnosis, but do not exist in the sera of non-tumor patients and healthy individuals. These autoantibodies exhibit high titers and persist for extended periods in the serum of tumor patients, detectable months or even years prior to clinical diagnosis. And they are absent or undetectable in the serum of non-tumor patients and healthy individuals due to either non-existence or extremely low concentrations. Consequently, these anti-TAA autoantibodies hold potential as biomarkers in tumor immunology. Therefore, these autoantibodies have the potential as tumor immunological markers[9-12]. Autoantibodies used for diagnosis vary across cancer types. Therefore, it is important to search for and identify potential autoantibodies with high sensitivity and specificity for early diagnosis of OS, which could be a target for immunotherapy in the future.

Carcinogenic factors can cause gene mutations in normal cells, especially in specific genes known as cancer-driver genes, which would promote the growth and proliferation of cancer cells. Cancer cells need such driver genes to continue to function, while normal cells do not. This study was mainly based on 138 cancer driver genes published by Vogelstein et al[13] on Science. Many methods have been used to search for new TAAs, such as serological analysis of recombinant cDNA expression libraries, phage display techniques, serological proteome analysis, and protein microarray. Protein microarrays, which can detect multiple TAAs or autoantibodies on a single chip with only a small number of samples and reagents, has the advantages of automation, speed and high sensitivity[14,15], and has been widely used in previous studies[16-20]. Osteochondroma (OC) has similar histological characteristics, site of occurrence and affected population as OS, therefore, it was selected as a benign control.

In this study, 154 corresponding proteins were used to customize a special protein microarray to screen and identify OS-related TAAs. Enzyme-linked immunosorbent assay (ELISA) was conducted to validate the results from the first stage, and western blotting (WB) and immunohistochemistry (IHC) confirmed the diagnostic value of the protein microarray for early detection of OS.

MATERIALS AND METHODS

Study subjects

Serum samples and tissues of OS and OC were collected intermittently from surgical inpatients of Luoyang Orthopedic Hospital of Henan Province (Orthopedic Hospital of Henan Province) during 2013 to 2022. The patients were histopathological diagnosed with OS or OC before any treatment, without other malignancies. The cohort used for focused protein microarray included 28 OS and 49 normal human sera (NHS; without other cancers or autoimmune diseases) and was provided by Tumor Epidemiology Laboratory of Henan Province during the same period. The cohort used for further validation included 49 OS, 39 OC (as benign bone tumor controls), and 49 normal controls (NC; without other cancers or autoimmune diseases) during the same period collected from physical examination department of Luoyang Orthopedic Hospital of Henan Province (Orthopedic Hospital of Henan Province). The OS and NC were matched according to age (± 3 years) and gender. The samples allocated for WB and IHC were chosen from the validation cohort for ELISA. After the blood samples were centrifuged at 1500 × g for 10 minutes, the pellet was removed. The sera were collected and stored at -80 °C. The demographic and clinical data were collected retrospectively. The detailed information of the subjects is shown in Table 1. This study was approved by the Ethics Committee of Luoyang Orthopedic Hospital of Henan Province (Orthopedic Hospital of Henan Province).

Table 1 Characteristics of study subjects, mean ± SD/n (%).

Variables	Discovery cohort (n = 77)		Validation cohort (n = 137)
	OS (n = 28)	NHS (n = 49)	OS (n = 49)	OC (n = 39)	NC (n = 49)
Age (year)	32.18 ± 19.31	40.09 ± 12.79	24.63 ± 14.84	18.77 ± 14.38	24.82 ± 11.47
Gender
Male	17 (60.71)	22 (44.90)	32 (65.31)	23 (58.97)	32 (65.31)
Female	11 (39.29)	27 (55.10)	17 (34.69)	16 (41.03)	17 (34.69)
Anatomic site
Left femur	7 (25.00)		14 (28.57)
Right femur	5 (17.86)		12 (24.49)
Left tibia	3 (10.71)		4 (8.16)
Right tibia	5 (17.86)		11 (22.45)
Right humerus	2 (7.14)		3 (6.12)
Others	6 (21.43)		5 (10.20)
Histologic type
Osteoblastic	5 (17.86)		8 (16.33)
Fibroblastic	3 (10.71)		4 (8.16)
Chondroblastic	4 (14.29)		8 (16.33)
Small cell OS	3 (10.71)		6 (12.24)
Others	4 (14.29)		7 (14.29)
Unknown	9 (32.17)		16 (32.65)
TNM
IA	12 (42.86)		25 (51.02)
IB	3 (10.71)		13 (26.53)
IIA	1 (3.57)		3 (6.12)
IIB	1 (3.57)		3 (6.12)
IV	11 (39.29)		1 (2.04)
Unknown	0 (0.00)		4 (8.16)
Tumor size
< 5 cm	8 (28.57)		10 (20.41)
≥ 5 cm	16 (57.14)		35 (71.43)
Unknown	4 (14.29)		4 (8.16)
Distant metastasis
No	17 (60.71)		48 (97.96)
Yes	11 (39.29)		1 (2.04)

OS: Osteosarcoma; TNM: Tumor-node-metastasis; NHS: Normal human sera; OC: Osteochondroma; NC: Normal control.

Protein microarray

The protein microarray contained 154 recombinant proteins that were expressed by 138 cancer driver genes (some genes expressed more than one protein fragment), which customized the HuProt^TM protein chip from BCBIO Biotechnology Corporation (Guangzhou, Guangdong Province, China). The focused protein microarray included 143 human recombinant proteins encoded by cancer driving genes and 11 proteins (CyclinB1, c-Myc, CIP2A/p90, RalA, IMP1, IMP2, IMP3, YWHAZ, RBM39, and two fragments of survivin) studied previously in our laboratory. The layout of the focused protein microarray and evaluation of the operational stability of different chips at different time points have been described previously[17,19,20]. The recombinant proteins were combined with specific IgG antibody in sera, and tested with anti-human IgG fluorescent secondary antibody to screen potential TAAs. Each chip could detect 14 serum samples simultaneously. The detailed protocol of the experimental was in accordance with our previous studies[19]. The level of autoantibodies to potential TAAs was measured by the signal-to-noise ratio.

Indirect ELISA

Three recombinant proteins [G-protein subunit alpha 11 (GNA11), serine/arginine-rich splicing factor 2 (SRSF2), and phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA)] purchased from Cloud-Clone Corporation (Wuhan, Hubei Province, China) were coated onto a 96-well enzyme plate to detect anti-TAA autoantibodies at a concentration of 0.25 μg/mL in phosphate buffered saline, incubated in 100 μL at 4 °C overnight. The plates were blocked with 200 μL 2% bovine serum albumin at 4 °C overnight. Plates were washed three times with 1 × phosphate buffered saline Tween (PBST). Serum samples were added 100 μL diluted at 1:100 to each plate and incubated for 1 hour at 37 °C. The plates were washed five times with 1 × PBST. 100 μL horseradish peroxidase-conjugated goat anti-human IgG (100 μL) diluted at 1:5000 was added to each plate and incubated for 1 hour at 37 °C. After the plates were washed five times with 1 × PBST, 100 μL 3,3’,5,5’-tetramethylbenzidine substrate solution and 50 μL ELISA stop solution (Solarbio Life Sciences Corporation, Beijing, China) were used to proceed and terminate the color reaction. Optical density (OD) was measured at 450 nm using an automated plate reader. Positive and negative control sera, confirmed by WB, screened from the preliminary experiments of each TAA, were set in each plate for quality control. The background signal was corrected by subtracting the average OD value of the blank wells.

WB

After the proteins underwent sodium dodecyl sulfate-polyacrylamide gel electrophoresis, SRSF2 was transferred onto a 0.45-μm polyvinylidene fluoride (PVDF) membrane. The PVDF membrane was blocked with 5% nonfat milk diluted with 1 × Tris-buffered saline Tween (TBST) for 1 hour at room temperature. The PVDF membrane was washed three times with 1 × TBST. The PVDF membrane was cut into strips and incubated with sera from OS, OC, and NC at a dilution of 1:200 with 5% nonfat milk in 1 × TBST for 1 hour at room temperature. The PVDF membrane was washed five times with 1 × TBST. Goat anti-human IgG-horseradish peroxidase (1:5000 dilution) was used as secondary antibody with 5% nonfat milk in 1 × PBST for 1 hour at room temperature. After washing five times in 1 × TBST, the protein in membranes was detected by chemiluminescence using an imaging system (Amersham Imager 680; Cytiva, Uppsala, Sweden).

IHC

The tissues of OS, OC, and NC (from para-tumor bone tissue) were embedded in paraffin, deparaffinized, and rehydrated. Hydrogen peroxide (3%) was used to block the endogenous peroxidase for 10 minutes. After the slides were washed three times, 10% goat serum albumin was added for 1 hour at room temperature. The slides were incubated with SRSF2 rabbit monoclonal antibody diluted at 1:200 at 4 °C overnight. This was followed by incubation with the secondary antibody for 30 minutes at room temperature. SRSF2 staining was performed on Bond-III (Leica Biosystems, Nussloch, Germany) stainers: Antigen retrieval for 20 minutes at pH 9, incubation with SRSF2 for 20 minutes, and counterstaining with hematoxylin for 10 minutes. Slides were dehydrated with sequential ethanol washes. The slides were observed under a microscope (Olympus, BX53; Olympus Corporation, Tokyo, Japan). The pathological analysis was performed using Aipathwell software developed by Wuhan Servicebio Technology Co. Ltd. (Hubei Province, China)[21] to obtain the IHC score. Histochemistry score was performed to evaluate the staining intensity of SRSF2: Histochemistry score ∑(pi × i) = (percentage of weak intensity cells × 1) + (percentage of moderate intensity cells × 2) + (percentage of strong intensity cells × 3), where i represents a positive grade and pi represents the percentage positive signal pixel area.

Statistical analysis

All statistical analyses were performed by IBM SPSS Statistics (version 21.0) and GraphPad Prism (version 9.5.0). If the data conformed to a normal distribution, Student’s t-test or one-way analysis of variance (ANOVA) was applied, otherwise nonparametric Kruskal-Wallis test were used to do the analysis. The cut-off value of TAAs was set as the 95^th percentile of NCs to define a positive reaction. A χ² test or Fisher’s exact test was used for qualitative variables. To evaluate the diagnostic value of anti-SRSF2 autoantibody in sera, the area under the receiver operating characteristic curve with area under the curve (AUC) was calculated. Comparison of IHC scores for OS, OC, and NC tissues was undertaken using one-way ANOVA. P < 0.05 was considered statistically significant.

RESULTS

Characteristics of the study population

In the cohort for the focused protein microarray, 77 subjects were included (28 OS and 49 NHS). The P-values of the statistical comparison between OS and NHS in terms of age and gender were 0.060 and 0.182, respectively. For the cohort used for further validation, 137 subjects were included (49 OS, 39 OC, and 49 NC). The P-values of the statistical comparison among these three groups in terms of age and gender were 0.072 and 0.786, respectively. There was no significant difference between the age and gender in these two cohorts. The basic clinical characteristics of the included subjects from these two cohorts are shown in Table 1.

Detection of potential TAAs in OS based on focused protein microarray

The stability of different chips in operation at different times was evaluated, and the samples were tested repeatedly 30 times. The overall average of the correlation between samples after linear fitting of the repeatability among different batches of protein microarrays was 0.98, indicating good overall stability. Supplementary Figure 1 shows the representative results of the response from four OS and four NHS samples. Green spots represent the positive reaction, which indicates that the protein was bound to the specific IgG antibody in the serum.

According to the results of the focused protein microarray based on analysis of 154 TAAs (Supplementary Table 1), three potential TAAs (GNA11, SRSF2 and PIK3CA) were screened out for the subsequent validation according to AUC > 0.5 and P < 0.05. The difference of signal-to-noise ratio of these three anti-TAA autoantibodies were significant between the OS and NHS. The receiver operating characteristic curves and scatter plots of these three anti-TAA autoantibodies are shown in Figure 1, and the AUC of these three anti-TAA autoantibodies are 0.768, 0.698, 0.641, respectively.

Figure 1 Receiver operator characteristic curve and scatter plots for three anti-tumor associated antigen autoantibodies in the cohort for focused protein microarray. A-C: Analysis of receiver operator characteristic curve with area under the curve and 95% confidence interval; D-F: Serological levels in signal to noise ratio of three anti-tumor associated antigen autoantibodies in osteosarcoma and normal human sera in scatter plots. The longest line means median, and the 25^th percentiles and 75^th percentiles are presented by the shorter line. GNA11: G-protein subunit alpha 11; AUC: Area under the curve; CI: Confidence interval; SRSF2: Serine/arginine-rich splicing factor 2; PIK3CA: Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; SNR: Signal to noise ratio; OS: Osteosarcoma; NHS: Normal human sera.

Diagnostic value of anti-SRSF2 autoantibody

OD values of three autoantibodies in sera of all subjects are shown in Table 2 and Figure 2. ELISA showed that among these TAAs, the concentration of anti-SRSF2 autoantibody in the sera of OS was significantly higher than that of NC. The AUC of anti-SRSF2 autoantibody was 0.648 (95% confidence interval: 0.537-0.758), P = 0.012.

Figure 2 Receiver operator characteristic curve and scatter plots for three anti-tumor associated antigen autoantibodies in the validation cohort. A-C: Analysis of receiver operator characteristic curve with area under the curve and 95% confidence interval; D-F: Expression levels of optical density about three anti-tumor associated antigen autoantibodies in osteosarcoma, osteochondroma, and normal controls in scatter plots. The longest line means median, and the 25^th percentiles and 75^th percentiles are presented by the shorter line. ^aP < 0.05. GNA11: G-protein subunit alpha 11; AUC: Area under the curve; CI: Confidence interval; SRSF2: Serine/arginine-rich splicing factor 2; PIK3CA: Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; OS: Osteosarcoma; NC: Normal control; OC: Osteochondroma; OD: Optical density.

Table 2 Optical density of autoantibodies of osteosarcoma, osteochondroma, and normal controls, median (25^th percentile-75^th percentile).

	OS	OC	NC	H	P value
GNA11	0.424 (0.295-0.705)	0.544 (0.324-0.962)	0.426 (0.357-0.506)	3.025	0.220
SRSF2	0.673 (0.524-0.845)	0.683 (0.315-1.051)	0.498 (0.311-0.793)	6.354	0.042
PIK3CA	0.727 (0.516-0.971)	0.708 (0.380-1.051)	0.827 (0.561-1.072)	3.846	0.146

GNA11: G-protein subunit alpha 11; SRSF2: Serine/arginine-rich splicing factor 2; PIK3CA: Phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; OS: Osteosarcoma; OC: Osteochondroma; NC: Normal control.

When the cut off value was set as the 95^th percentile of NC to define a positive reaction, higher than or equal to the cut-off value was judged as OS, while others were defined as normal. The diagnostic value of anti-SRSF2 autoantibody for sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, positive predictive value, negative predictive value, accuracy, and Youden index were 16.33%, 95.92%, 4, 0.87, 80.00%, 53.41%, 56.12%, and 0.1224, respectively.

Expression of SRSF2 by WB

To confirm the results of ELISA, twelve OS, six OC sera, and six NC sera were randomly selected for WB. The molecular weight of SRSF2 protein is approximately 54 kDa. WB showed that nine of twelve OS serum samples had positive bands, four of six OC serum samples had positive bands, and one of six NC serum samples had positive bands near 54 kDa. The results of WB were consistent with those of ELISA. The full-length blots of all strips are shown in Supplementary Figure 2.

Expression of SRSF2 in OS tissues by IHC

Five OS, five OC, and five para-tumor bone tissues were collected. Unfortunately, one of the para-tumor bone tissue was lost due to tissue detachment during processing. The result of IHC staining showed strong expression of SRSF2 in the OS tissues, which was different from the NC tissues. The normal bone tissues were stained with a weak reaction signal against purified SRSF2. The results also indicated that SRSF2 was significantly upregulated in cancer tissues compared to OC or NC (P < 0.001; Figure 3).

Figure 3 Serine/arginine-rich splicing factor 2 is overexpressed in osteosarcoma tissue compared with the osteochondroma and normal bone tissue by immunohistochemistry. A and B: Positive staining of serine/arginine-rich splicing factor 2 (SRSF2) expressed in representative osteosarcoma tissue at 100 × and 400 × magnification, respectively; C and D: Weak positive staining of SRSF2 expressed in osteochondroma tissue at 100 × and 400 × magnification, respectively; E and F: Negative staining of SRSF2 expressed in normal bone tissue at 100 × and 400 × magnification, respectively; G: Immunohistochemistry. Scores of SRSF2 between osteosarcoma, osteochondroma, and normal controls. IHC: Immunohistochemistry; SRSF2: Serine/arginine-rich splicing factor 2; OS: Osteosarcoma; OC: Osteochondroma; NC: Normal control.

DISCUSSION

OS is a malignant tumor that seriously endangers the growth of children and adolescents. Due to the insidious onset of OS and the lack of effective early screening methods, it is important to find non-invasive methods to diagnose OS early in clinical practice for early treatment and improved prognosis. Anti-TAA autoantibody is an effective means for early diagnosis, which is present in the blood of patients[12,22,23]. The level of anti-TAA autoantibodies can increase earlier than the occurrence of tumor symptoms[24]. TAAs in the patient's serum may be cleared, while their autoantibodies can exist continuously and stably for a long time[25]. Therefore, searching for potential tumor markers to diagnose OS early by high-throughput protein microarray technology is effective and reliable. Protein microarray is a new technique to study protein interaction, which can detect multiple tumor-related antigens or autoantibodies simultaneously on a fixed chip[26]. It has the advantages of high throughput, speed, automation and high sensitivity[27], which only consumes a small number of samples and reagents. This study mainly aimed to screen potential TAAs based on a focused protein microarray encoded by 138 cancer driver genes[13], which were validated by ELISA, and verified by WB and IHC.

In this study, expression of three significant autoantibodies (GNA11, SRSF2, and PIK3CA) was screened out by focused protein microarray. G protein coupled receptor is a heterotrimer, consisting of α, β and γ subunits. As the most widely distributed membrane surface protein in mammalian cells, its coupling mediates a variety of cell functions, whose abnormal function can lead to autoimmune diseases, hypertension, tumors and other diseases[28-30]. GNA11 and G protein subunit alpha q are closely related guanine nucleotide-binding a subunit of G-proteins, acting as driver genes in the process of oncogenesis[31]. Mutations of these two genes result in the activation of several important signaling pathways, including phospholipase C, and activation of the transcription factor yes-associated protein[32]. Somatic mutations of G protein subunit alpha q and GNA11 are common in uveal melanoma[33,34]. PIK3CA encodes the p110α catalytic subunit of phosphatidyl-inositol-3-kinase, which plays an important role in the proliferation, metabolism and protein synthesis, angiogenesis and apoptosis of cells[35]. PIK3CA is an oncogene[35,36], and one of the most commonly mutated genes in solid cancers[37]. It is associated with many types of tumors, such as breast cancer, colorectal cancer, glioblastoma and oral cancer[38-42]. The ELISA results in the validation group demonstrated statistically significant anti-SRSF2 autoantibody, and the function of SRSF2 will be discussed below. SRSF2 is rich in serine/arginine, and is a crucial regulator of constitutive and alternative pre-mRNA splicing[43]. It is required for formation of the earliest adenosine triphosphate-dependent splicing complex and interacts with spliceosomal components bound to the 5’-splice and 3’-splice sites during spliceosome assembly, and for adenosine triphosphate-dependent interactions of both U1 and U2 small nuclear ribonucleoproteins with pre-mRNA. Via the arginine/serine-rich domains, it interacts with other spliceosomal components to form a bridge between the 5’-splice and 3’-splice site binding components, U1 small nuclear ribonucleoprotein and U2 small nuclear RNA auxiliary factor. It can bind to purine-rich RNA sequences, either 5’-AGSAGAGTA-3’ (S = cytosine or guanine) or 5’-GTTCGAGTA-3’. SRSF2 also binds to -globin mRNA, which commits it to the splicing pathway[43,44]. SRSF2 is involved in multiple biological processes and tumor progression, such as mRNA processing and splicing, and is often mutated or up-regulated in cancer[45]. Abnormal variable splicing is closely related to the occurrence and development of multiple genetic diseases and tumors. In different organs and tumors, SRSF2 can promote or inhibit the occurrence and development of tumors by regulating variable splicing[46,47]. SRSF2 is a key regulator of RNA splicing dysregulation in cancer, and could act as a candidate prognostic factor[48,49]. Many studies had revealed the function of SRSF2 in blood tumors. For example, Smeets et al[50] generated mutation of SRSF2 that affected hemopoiesis and myelodysplastic syndromes or myeloproliferative neoplasms. Rahman et al[51] uncovered the critical effects of SRSF2 mutants in hematological malignancies. Such research on OS has been rare. Luo et al[52] found that bromodomain-containing protein 4-mediated pre-messenger RNA of acyl-coenzyme A thioester synthetase long-chain family member 3 splicing affected erastin-induced ferroptosis by affecting arachidonic acid synthesis in OS cells. Knock down of coactivator-associated arginine methyltransferase 1 promoted nuclear accumulation of SRSF2 in U2OS cells, independent of cell cycle phase[53].

Generally, the frequency of single biomarkers is variable, but rarely exceed about 15%-20%[54], as is the sensitivity. Anti-SRSF2 autoantibody had sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and Youden index of 16.33%, 95.92%, 80.00%, 53.41%, 56.12%, and 0.1224, respectively. The AUC of this autoantibody was 0.648 (95% confidence interval: 0.537-0.758; P = 0.012), indicating its potential diagnostic value. Serological examination has unique advantages, small trauma, low concentration, high detection efficiency and simple operation. The incidence of OS is lower than that of other common tumors and the tissue is difficult to collect. The sensitivity of the selected biomarker in this study was 16.33%. Diagnosis of the disease cannot rely on a single indicator. If combined with findings from computed tomography or other serologic tests, the accuracy of diagnosis can be significantly improved, thereby reducing the rate of missed diagnosis. For individuals, anti-SRSF2 autoantibody may serve as an indicator for disease surveillance. It is possible that its sensitivity of anti-SRSF2 autoantibody to a particular subtype of OS may have a significant effect in future studies with a large enough sample size. The other function of this autoantibody is that it could be a potential therapeutic target but this needs further study. Although this autoantibody has limited sensitivity, its complementary value in early detection, specific subtypes, longitudinal monitoring, the gain in predictive values when combined with existing approaches, and the low-cost make it potentially applicable in clinical scenarios, particularly as part of a combined diagnostic model.

Expression of anti-SRSF2 autoantibody in esophageal squamous cell carcinoma patients is higher than that in NC[20]. The AUC of SRSF2 diagnosed in lung cancer was 0.661 in the discovery cohort reported by Jiang et al[19], and the serum level of four tumor-associated autoantibodies, including SRSF2, was higher than in benign lung disease. In dogs with appendicular OS, it was suggested that elevated lactate dehydrogenase at diagnosis indicated a more advanced disease stage and poorer prognosis[55]. It was highlighted by small RNA sequencing and basic research that exosomal miR-9-5p could be a promising biomarker and therapeutic target of OS screening[56]. Li et al[57] used three optimal microbial markers (6 serum metabolites) demonstrating strong diagnostic efficiency in distinguishing OS patients from healthy individuals. One study demonstrated that serum N-terminal pro-C-type natriuretic peptide concentration was significantly different between OS and controls[58]. The expression levels of two non-coding RNAs (lncRNA OIP5-AS1 and hsa_circ_0004674) were increased in the serum of patients with OS compared with bone fractures[59]. However, the diagnostic value of the above studies was not discussed. There are several potential biomarkers for detecting OS. Sun et al[60] combined serum alkaline phosphatase, tumor-supplied growth factor group, and lactate dehydrogenase to detect pediatric OS, with an AUC of 0.886. A novel diagnostic model was established consisting of four RNA biomarkers (hsa-circ-0010220, hsa-miR-326, hsa-miR-338-3p, and FAM98A) with an AUC of 0.928[61]. The function of miR-337-3p, miR-484, miR-582, and miR-3677 for the detecting OS was verified by Luo et al[62]. In our previous study, the diagnostic value of anti-enolase-1 autoantibody was validated to be a biomarker with an AUC of 0.853 and sensitivity of 23.1% for immunodiagnostic and progression of OS[63]. More indicative biomarkers need to be combined for the diagnosis of OS in the future.

In this study, through a focused protein microarray encoded by cancer driver genes, three potential autoantibodies were screened out. ELISA was used to validate the results of protein microarray analysis. WB and IHC were used to verify the ELISA results. Finally, anti-SRSF2 autoantibody was verified as the most potential diagnostic biomarker to distinguish OS. There were many advantages of this study. First, a customized protein microarray was adopted. Then, several technologies were applied to validate the autoantibodies. And in the validation cohort, the validation of three TAAs in ELISA included benign bone group (OC). The design of the study was scientific and rational. There were also some limitations of this study. A sensitivity of 16%, accompanied by high specificity (95.92%), is insufficient for standalone diagnostic application in OS. OS lacks validated serum biomarkers, making any specific signal potentially valuable in defined clinical niches. The high specificity may reduce unnecessary biopsies in patients with indeterminate bone lesions - a clinically relevant endpoint given the invasive nature of bone biopsy and associated morbidity. The diagnostic value of a single biomarker was limited. It was important to search for a multi-modal panel with higher sensitivity combining miRNA, ctDNA, metabolic markers and other clinical tumor markers for clinical utility. The diagnostic value can be further validated in studies with larger cohort samples and compared with other cancer types in the future.

CONCLUSION

Anti-TAA autoantibodies can be considered as potential diagnostic biomarkers in the detection of OS. Anti-SRSF2 autoantibody can be used as a potential serological biomarker in the detection of patients with OS, which could be a useful tool in immunodiagnostic of OS.

ACKNOWLEDGEMENTS

The authors thank Tumor Epidemiology Laboratory of Henan Province and Henan Engineering Research Center of Prevention and Treatment of Bone Tumor with Traditional Chinese Medicine for their support in this work.