Abstract

Background

   The like-Smith (LSM) family is a group of RNA-binding proteins involved
   in RNA metabolism. However, their involvement in tumors, particularly
   skin cutaneous melanoma (SKCM), is not fully understood. In this study,
   we focused on the expression profiles and prognostic values of the LSM
   family in SKCM.

Methods

   Raw data were downloaded from The Cancer Genome Atlas. The expression
   profile and prognostic value of LSM genes in SKCM were explored using
   the GEPIA, cBioPortal, and HPA databases. Protein–protein and gene–gene
   interaction analyses were performed using STRING and GeneMANIA.
   Enrichment and Cox regression analysis were conducted using R software.
   The TISIDB database was used to explore the relationship between LSMs
   and immunomodulators. Receiver operating characteristic curves and
   nomogram models were constructed to validate prognostic values.

Results

   mRNA and protein expression levels of LSM2, LSM4, and LSM12 were
   significantly elevated in SKCM. The upregulated mRNA expression of LSM2
   (p = 0.0013) and LSM4 (p = 0.0043) was significantly correlated with
   poor overall survival in patients with SKCM, whereas only LSM2
   (p = 0.049) overexpression was markedly associated with worse
   disease-free survival. LSM2 overexpression was an independent risk
   factor (p = 0.013) and was confirmed to have a high prognostic value in
   SKCM using the receiver operating characteristic curve (AUC = 0.942)
   and nomogram models. All LSM genes were identified as genomic
   mutations, whereas alteration of LSM2 (p = 0.0153) significantly
   affected the overall survival in patients with SKCM. Significant
   correlations were observed between LSM family expression, immune cell
   infiltration, and immunomodulator. Furthermore, function and pathway
   enrichment analysis showed that the LSM family was mainly RNA binding
   proteins and involved in RNA splicing and degradation.

Conclusion

   Expression profiles and prognostic values of LSM in SKCM were
   inconsistent. Among the LSM family, only LSM2 may serve as a potential
   poor prognosticator and immunotherapeutic target of SKCM.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12920-022-01395-6.

   Keywords: Biomarker, Skin cutaneous melanoma (SKCM), LSM family, LSM2,
   RNA splicing

Background

   Although skin cutaneous melanoma (SKCM) constitutes less than 5% of all
   skin cancers, it accounts for more than 75% of skin cancer related
   deaths [[31]1]. The incidence and morbidity of SKCM have been steadily
   increasing worldwide over the past few decades, with 350,000 new cases
   and 60,000 deaths worldwide annually [[32]2, [33]3]. The incidence of
   SKCM varies among ethnic groups, with the White population having the
   highest incidence (approximately 21.6/100,000 people), followed by the
   Black population (approximately 1.0/100,000 people), and the incidence
   rate in the Chinese population is slightly lower than that in the Black
   population [[34]4, [35]5]. In recent years, the incidence of SKCM in
   China has increased significantly, with approximately 20,000 new cases
   occurring every year [[36]6]. Therefore, early diagnosis and treatment
   are imperative to decrease SKCM-related mortality. Despite the
   development of surgical and systemic therapies [[37]7], the survival
   rate of patients with SKCM remains relatively low. In terms of
   precision medicine, various studies have revealed that differentially
   expressed genes (DEGs) could be helpful in clarifying the tumor
   molecular biology, prognosis, and molecular targeted drugs in SKCM
   [[38]8, [39]9]. To date, a large number of biomarkers have been
   discovered [[40]10–[41]12], but most are in the basic research stage,
   and there are only a few biomarkers that can be used to guide clinical
   practice. Therefore, identifying highly specific and selective
   biomarkers associated with SKCM is particularly important, which may
   facilitate the diagnosis, treatment, and prognosis of SKCM in the
   future.

   Sm protein was first studied in research on precursor RNA. It is a
   highly conserved homologous protein family containing Sm motifs;
   therefore, it is called the Like-Smith (LSM) protein [[42]13]. The
   conserved folding of Sm and LSM subunits consist of a 5-strand
   antiparallel β sheet, and in some cases it can be enhanced by the
   addition of secondary structural elements and/or unstructured
   C-terminal extensions of varying length [[43]14]. There are 13 members
   of the LSM protein family (LSM1-LSM14B). The LSM protein family is an
   RNA binding protein (RBP) that is mainly involved in the shearing and
   processing of precursor mRNA in the nucleus [[44]13, [45]15]. In
   eukaryotic cells, a hetero-heptameric complex of seven LSM proteins
   (LSM1-7) promotes mRNA decapping and decay in the cytoplasm, whereas a
   different hetero-heptameric complex of LSM proteins (LSM2-8) affects
   the processing of pre-mRNA and small stable RNAs in the nucleus
   [[46]16, [47]17]. The LSM1-7 complex physically interacts with other
   decay factors, such as Pat1p, Dhh1p, and Xrn1p [[48]18] and
   preferentially associates in vivo with a pool of oligoadenylated mRNPs
   targeted for decapping enzyme rather than the translation factors
   [[49]19, [50]20]. The LSM2-8 complex shares six out of seven subunits
   with the LSM1-7 complex, and binds the 3′ ends of U6 and U6^atac
   snRNAs. The U6 snRNA is then processed by the 3′-5′ exoribonuclease
   Usb1, resulting in a 2′, 3′ cyclic phosphate in most organisms
   [[51]21]. LSM12 was identified as an essential participant in
   NAADP-evoked TPC activation and Ca^2+ mobilization from acidic stores
   [[52]22]. The oncologic roles of the LSM family have been explored in
   several cancers. LSM1 participates in the proliferation of viruses in
   cells and regulates their replication [[53]23, [54]24]; it also acts as
   an oncogene in breast cancer (BC) after 8p11-12 amplification [[55]25].
   Moreover, LSM1 promotes chemoresistance, transformation, and metastasis
   of pancreatic cancer cells [[56]26]. Meanwhile, downregulation of LSM1
   is associated with prostate cancer progression [[57]27]. LSM2
   homologous U6 small nuclear RNA-associated form, namely the LSM2-8
   complex, participates in the splicing of mRNA precursors and regulates
   the expression of multiple genes by controlling the spliceosomal cycle.
   RNA processing and stabilization are also maintained in the nucleus
   [[58]17]. In ovarian cancer, the decrease in the degree of LSM2
   methylation may lead to an increase in its expression, which results in
   a disorder in the processing and circulation of most RNA, affecting the
   expression of various proteins and accelerating tumor development
   [[59]28]. LSM3 has been confirmed to be not only major pathogenic gene
   of Alzheimer’s disease (AD), but also bridge gene [[60]29] and to
   significantly affect the proliferation of non-small cell lung cancer
   cell lines [[61]30]. In addition, LSM4 overexpression was significantly
   correlated with poor survival in BC [[62]31], pancreatic ductal
   adenocarcinoma (PDAC) [[63]32], and hepatocellular carcinoma (HCC)
   [[64]33]. Overall, these studies indicate that the LSM family may be a
   promising therapeutic target and prognostic biomarker for multiple
   tumors. However, the functions of LSM family members have not been
   fully elucidated. In particular, there are no reports about LSM family
   members in SKCM.

   With the continuous progress of bioinformatics and bioinformatic
   approaches, comprehensive analysis of gene expression, epigenetic
   regulation, and biological functions in numerous malignancies is
   becoming increasingly accessible. Hence, we assessed the expression
   patterns, prognostic values, and biological functions of LSM family
   genes in SKCM using multiple public databases. We found that the
   expression profiles and prognostic values of LSM family members in SKCM
   were inconsistent. Only LSM2, in the LSM family, might serve as a
   potential biomarker for SKCM.

Methods

The cancer genome atlas

   The Cancer Genome Atlas (TCGA) is a large tumor gene chip database that
   can be used to identify outliers, analyze differences in gene
   expression, and predict co-expressed genes. The clinical data in TCGA
   includes tumor stage, tissue type, grade, and overall survival. It can
   be used to identify DEGs, therapeutic targets, and diagnostic
   biomarkers. The RNA sequencing data and clinical data of SKCM were
   downloaded from the Genomic Data Commons (GDC,
   [65]https://portal.gdc.cancer.gov/).

Gene expression profiling interactive analysis

   Gene Expression Profiling Interactive Analysis (GEPIA;
   [66]http://gepia.cancer-pku.cn/) is an online database. It is a network
   assistant in exploring the RNA sequencing expression of 9736 tumors and
   8587 regular cases from TCGA and Genotype Tissue Expression (GTEx) by
   applying a qualified processing method [[67]34]. In this study, GEPIA
   was used to explore LSM family gene expression profiles and survival
   analysis.

Human protein atlas

   The Human Protein Atlas (HPA; [68]https://www.proteinatlas.org/)
   database was used to determine the protein levels of the LSM family in
   SKCM and normal skin tissues.

cBiPortal

   The cBio Cancer Genomics Portal (cBioPortal,
   [69]https://www.cbioportal.org/) is a platform based on TCGA that
   provides multidimensional visual data. A total of 2549 cutaneous
   melanoma samples were used for further analysis. The type and frequency
   of LSM family gene mutations in cutaneous melanoma were analyzed using
   the “OncoPrint” and “Cancer Type Summary” module. The prognostic value
   of gene alterations was explored in the “survival” module.

GeneMANIA analysis

   GeneMANIA ([70]http://genemania.org/) is a user-friendly tool. It can
   predict the interactions between genes, including co-expression,
   co-locations, pathways, physical interactions, and genetic
   interactions, to generate prioritized genes for further functional
   analysis.

STRING

   STRING ([71]https://cn.string-db.org/) was used to probe known
   interactions between proteins and help speculate their interactions. In
   our study, STRING was used for protein–protein interaction (PPI)
   analysis of the LSM family and peripheral interaction proteins.
   Statistical significance was set at a confidence score > 0.4 was
   considered significant.

Function enrichment analysis

   Gene Ontology (GO), which includes three categories: biological process
   (BP), cellular component (CC), and molecular function (MF), can provide
   a framework to reveal the function of gene products and to clarify
   transcriptome data of high-throughput genes and biological properties.
   The Kyoto Encyclopedia of Genes and Genomes (KEGG) [[72]35] can
   identify specific pathways that can be used for the functional
   interpretation of genomic information. GO terms and KEGG pathways
   enrichment analysis were performed by “clusterProfiler” R package. An
   adjusted p-value < 0.05 was considered as significant enrichment.

Immune-related analysis

   First, the R package “GSVA” was used to present infiltration enrichment
   of 24 common immune cells, including T helper cells, type 2 Th (Th2)
   cells, CD8 T cells, T central memory (Tcm), Th1 cells, T gamma delta
   (Tgd), B cells, T cells, dendritic cell (DC), activated dendric cells
   (aDC), plasmacytoid DC (pDC), immature DC (iDC), natural killer (NK)
   cells, NK CD56bright cells, Cytotoxic cells, type 17 Th (Th17) cells, T
   follicular helper (TFH), NK CD56dim cells, regulatory T cell (Treg),
   macrophages, mast cells, neutrophils, Tem, and eosinophils. The
   relationship between LSM gene expression and immune cell infiltration
   levels was evaluated using Spearman’s analysis.

   TISIDB ([73]http://cis.hku.hk/TISIDB/index.php) is a user-friendly web
   portal for the analysis of interactions between tumors and
   immunomodulatory factors. It integrates multiple heterogeneous data
   types, including RNA sequencing data of patients receiving
   immunotherapy, literature mining results from PubMed, high-throughput
   screening data, and TCGA data [[74]36]. Here, to further clarify the
   immune correlation of LSMs in SKCM, we used the “Immunomodulator”
   module in the TISIDB database to explore and evaluate the relationship
   between LSM expression and the level of immune checkpoint genes.

Statistical analysis

   The receiver operating characteristic (ROC) curve analysis according to
   sensitivity and specificity was performed using the “pROC” and
   “ggplot2” packages. (R software, version 4.1.3,
   [75]https://www.r-project.org/). The area under the curve (AUC) ranging
   from 0.5 (no diagnosis) to 1.0 (perfect diagnosis) was established. The
   interactions of LSM family genes, as well as the relationship between
   LSM family gene expression and immune cell infiltration, were
   determined using Spearman’s analysis. The univariate and multivariate
   Cox regression analysis were used to analyze the individual hazard
   ratio (HR) for overall survival (OS) via the “survival” R package. HR
   with a 95% confidence interval (CI) was measured to estimate the hazard
   risk of individual factors. The “rms” R package was used to draw the
   nomogram and construct a prediction model. A p-value < 0.05 was
   considered significantly different.

Results

Aberrant expression of LSM family genes in patients with SKCM

   Based on the GEPIA database, the expression levels of LSM family
   members were compared between 461 SKCM tissues and 558 normal skin
   tissues. The expression levels of LSM2, LSM4, LSM8, LSM10, and LSM12
   were remarkably elevated in SKCM specimens compared to normal cutaneous
   specimens (Fig. [76]1B, D, H, I, K). There were no statistically
   significant differences in the expression of LSM1, LSM3, LSM5, LSM6,
   LSM7, LSM11, LSM14A, and LSM14B compared with that in normal tissue
   (Fig. [77]1A, C, E, F, G, J, L, M). A summary of LSM family expression
   in SKCM is shown in Fig. [78]1N. We further assessed the LSM family
   expression at various pathological stages (Fig. [79]2). The results
   showed that the expression of LSM1, LSM2, LSM3, LSM6, LSM14A, and
   LSM14B (Fig. [80]2A–C, F, L, M) were significantly correlated with the
   pathological stages (p < 0.05). The expression of LSM4, LSM5, LSM7,
   LSM8, LSM10, LSM11 and LSM12 were not related to the pathological
   stages (Fig. [81]2 D, E, G, H, I, J, K). Subsequently, the protein
   expression patterns of LSM genes in patients with SKCM were explored
   using the HPA database. Immunohistochemical images of LSM2
   (Fig. 3A1–A4), LSM4 (Fig. 3B1–B4), LSM8 (Fig. 3C1–C4), and LSM12
   (Fig. 3D1–D4) in patients with SKCM are presented in Fig. [82]3. LSM8
   was weakly expressed in cutaneous melanoma and normal tissues. The
   protein expression of LSM2, LSM4, and LSM12 was upregulated in
   cutaneous melanoma. Most of the LSM2 cells were strongly stained in
   melanoma tissues compared to the normal control.

Fig. 1.

   [83]Fig. 1
   [84]Open in a new tab

   Differential expression of LSM family genes between the SKCM samples
   and normal skin tissues at mRNA level. LSM2, LSM4, LSM8, LSM10, and
   LSM12 were significantly overexpressed in SKCM (n = 461) compared with
   normal skin tissues (n = 558). Red box, tumor; grey box, normal
   samples. T, tumor; N, normal. *p < 0.05. N Expression profile of LSM
   family members in SKCM samples. Abbreviations: LSM, Like-Smith; SKCM,
   skin cutaneous melanoma

Fig. 2.

   [85]Fig. 2
   [86]Open in a new tab

   Associations between the expression of LSM genes and the pathological
   stage in patients with SKCM. The expression of LSM1, LSM2, LSM3, LSM6,
   LSM14A, and LSM14B was significantly correlated with the tumor stage,
   p < 0.05

Fig. 3.

   [87]Fig. 3
   [88]Open in a new tab

   The IHC images of upregulated LSM family members (LSM2, LSM4, LSM8, and
   LSM12) in SKCM patients. A[1], B[1], C[1] and D[1] represent the
   protein levels of LSM2, LSM4, LSM8, and LSM12, respectively, in normal
   skin tissue. A[2–3], B[2–3], C[2–3], and D[2–3] represent the protein
   expression of LSM2, LSM4, LSM8, and LSM12 in cutaneous melanoma
   tissues. The IHC staining intensities of LSM2 (12 patients), LSM4 (12
   patients), LSM8 (10 patients), and LSM12 (11 patients) are shown in the
   bar charts (A[4], B[4], C[4], and D[4]). All data were obtained from
   HPA. Abbreviations: IHC, immunohistochemistry; HPA, human protein
   albumin

Prognostic values of LSM family in SKCM

   The GEPIA database was used to investigate the prognostic value of LSM
   family members in SKCM. The results suggest that LSM2, LSM4, and LSM6
   were significantly correlated with OS. Higher expression of LSM2
   (p = 0.0013) and LSM4 (p = 0.0043) and lower expression of LSM6
   (p = 0.018) were prognostic factors for OS (Fig. [89]4B, D, F). The
   other LSMs had no significant influence on OS (Fig. [90]4 A, C, E, G,
   H, I, J, K, L, M). However, only LSM2 (Fig. [91]5B) was closely linked
   to disease-free survival (DFS), and patients with low LSM2 expression
   had better DFS (p = 0.049). LSM1, LSM3, LSM4, LSM5, LSM6, LSM7, LSM8,
   LSM10, LSM11, LSM12, LSM14A, and LSM14B expressions were not associated
   with DFS (Fig. [92]5A, C, D, E, F, H, I, J, K, L, M). In univariate
   analysis, the HR of LSM2, LSM4, and LSM6 were 1.354, 1.354, and 0.593,
   respectively, which were significantly related to patient survival
   (Additional file [93]1: Figure S1). Some members of the LSM family were
   significantly upregulated, and their overexpression was associated with
   several clinical characteristics and adverse clinical outcomes in
   patients with SKCM. We further evaluated the prognostic value of LSM
   genes in patients with SKCM. ROC curve analysis showed that LSM2
   (AUC = 0.942), LSM4 (AUC = 0.894), LSM7 (AUC = 0.842), LSM8
   (AUC = 0.969), and LSM10 (AUC = 0.877) had high accuracy, whereas LSM1
   (AUC = 0.655), LSM3 (AUC = 0.688), LSM5 (AUC = 0.774), LSM6
   (AUC = 0.550), LSM9 (AUC = 0.942), LSM11 (AUC = 0.525), LSM12
   (AUC = 0.741), LSM14A (AUC = 0.536), and LSM14B (AUC = 0.604) had low
   accuracy in predicting the prognosis of SKCM (Fig. [94]6).

Fig. 4.

   [95]Fig. 4
   [96]Open in a new tab

   The relationships between OS and the expression of LSM family members
   in SKCM patients. Overexpression of LSM2 (p = 0.0013) and LSM4
   (p = 0.0043) was associated with poor OS in SKCM. Upregulated
   expression of LSM6 (p = 0.018) was associated with longer OS in
   patients with SKCM. Abbreviations: OS, overall survival

Fig. 5.

   [97]Fig. 5
   [98]Open in a new tab

   The correlations between the DFS and the expression of LSM genes in
   patients with SKCM. In particular, LSM2 overexpression significantly
   affected the DFS of SKCM patients (p = 0.049). Abbreviations: DFS,
   disease-free survival

Fig. 6.

   [99]Fig. 6
   [100]Open in a new tab

   Assessment of prognostic values of LSM genes by ROC curves. A The ROC
   curve of LSM1-LSM7. B ROC curve of LSM8-LSM14. The results indicated
   that LSM2, LSM4, LSM7, LSM8, and LSM10 had high accuracy (AUC > 0.8),
   whereas LSM1, LSM3, LSM5, LSM6, LSM9, LSM11, LSM12, LSM14A, and LSM14B
   had low accuracy in predicting SKCM. Abbreviations: ROC, receiver
   operating characteristic; AUC, area under the curve

Gene mutation of LSM family genes

   cBioPortal was used to explore the genetic alterations differentially
   expressed in LSM family genes. Our findings showed that LSM1 (1.2%),
   LSM2 (3%), LSM3 (1%), LSM4 (0.9%), LSM5 (1.3%), LSM6 (1.1%), LSM7
   (0.6%), LSM8 (1.2%), LSM10 (1.3%), LSM11 (1.3%), LSM12 (0.6%), LSM14A
   (2.2%), and LSM14B (2.4%) were altered in SKCM samples (Fig. [101]7A).
   Of the 368 cases, 18.48% exhibited alterations in LSM family genes,
   including mutation (4.08%; 15 cases), amplification (11.14%; 41 cases),
   deep deletion (2.17%; 8 cases), multiple alterations (1.09%; 4 cases)
   (Fig. [102]7B). The mutation type of LSM2 was missense (Fig. [103]7C).
   The altered LSM2 (p = 0.0153) significantly affected the OS of patients
   with SKCM (Fig. [104]7D).

Fig. 7.

   [105]Fig. 7
   [106]Open in a new tab

   Alterations of LSM genes and the results obtained from survival
   analysis in SKCM. A OncoPrint visual summary of LSM gene alterations. B
   Summary of alterations in LSM family genes in cutaneous melanoma. C
   Mutation of LSM2 in SKCM. D. Kaplan–Meier plot comparing the OS in
   patients with and without LSM2 alterations in SKCM (p = 0.0153)

Correlation and functional enrichment analysis of LSM family genes in SKCM
patients

   The correlation coefficients of the LSM family in SKCM were assessed
   and are shown in Fig. [107]8 A (The details of Spearman analysis and
   p-value are shown in Additional file [108]4: Tables S1, Additional file
   [109]5: Table S2). LSM5 expression was closely correlated with that of
   LSM10 (R = 0.57, p < 0.05). LSM8 expression closely correlated with
   LSM6 expression (R = 0.57, p < 0.05). LSM11 expression closely
   correlated with LSM14B expression (R = 0.53, p < 0.05). The PPI network
   revealed associations between the LSM family genes (Fig. [110]8B).
   Furthermore, a network of LSM family genes and functionally related
   genes was constructed using GeneMANIA. The top 50 most related genes of
   the LSM family is shown in Fig. [111]8C. SNRPD2, SNRPD3, SNRPF, and
   SNRPE were identified as the genes most related to the LSM family.
   Functional and pathway enrichment of the above genes was performed and
   displayed in bubble charts using the R software (adjusted p < 0.05,
   Fig. [112]9). The main BP (Fig. [113]9A) of these genes was RNA
   splicing. The CC (Fig. [114]9B) of these genes corresponds to the
   Sm-like protein family complex. The MF (Fig. [115]9C) of these genes
   were most closely related to small nuclear RNA (snRNA) binding and U6
   snRNA binding. KEGG (Fig. [116]9D) pathway analysis revealed that these
   genes were mainly involved in spliceosome and RNA degradation.

Fig. 8.

   Fig. 8
   [117]Open in a new tab

   The correlations between LSM family and related genes. A The heatmap of
   LSM family gene correlations was analyzed using the Spearman test. B
   Protein–protein interaction network of LSM family genes. C Gene–gene
   interaction association networks of LSM family genes and the top 50
   co-expression genes

Fig. 9.

   [118]Fig. 9
   [119]Open in a new tab

   GO and KEGG analysis of the LSM family and top 50 involved genes in
   SKCM. A–C display the top five BP, CC, and MF, respectively. D KEGG
   pathways of the genes involved. Functional and pathway enrichment was
   determined using R and displayed in bubble charts. Abbreviations: GO,
   Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; BP,
   biological process; CC, cellular component; MF, molecular function

Associations of LSM family and immune cell infiltration in SKCM

   Tcm (R = −0.278, p < 0.001), Th1 cells (R = −0.24, p < 0.001), and
   macrophages (R = −0.227, p < 0.001) negatively correlated with LSM2
   expression (Fig. [120]10A). Tcm (R = −0.36, p < 0.001) and T helper
   cells (R = −0.265, p < 0.001) were negatively correlated with LSM4,
   while NK CD56bright cells (R = 0.187, p < 0.001) were positively
   correlated (Fig. [121]10B). NK cells (R = -0.339, p < 0.001) and
   eosinophils (R = −0.229, p < 0.001) were negatively correlated with
   LSM8, while T helper cells (R = 0.193, p < 0.001) and Th2 cells
   (R = 0.146, p < 0.001) were positively correlated (Fig. [122]10C). NK
   cells (R = −0.21, p < 0.001) and pDC (R = −0.302, p < 0.001) were
   negatively correlated with LSM12, while Th2 cells (R = 0.326,
   p < 0.001) and T helper cells (R = 0.308, p < 0.001) were positively
   correlated (Fig. [123]10D).

Fig. 10.

   [124]Fig. 10
   [125]Open in a new tab

   Correlations between immune infiltrating cells and the overexpression
   of LSM genes. A LSM2, B LSM4, C LSM8, and D LSM12, respectively

Regulation of immune molecules by LSM2 and LSM4

   Based on the above analysis, it was demonstrated that LSM2 and LSM4 are
   associated with poor prognosis and have a significant relationship with
   immune cells. To further explore the regulatory effect of LSM2 and LSM4
   on immune molecules in SKCM, the TISIDB database was used to perform an
   integrated analysis to predict the correlation between LSM expression
   and immunomodulators. Immunomodulators could be further classified into
   immunoinhibitors, immunostimulators, and major histocompatibility
   complex (MHC) molecules. The correlation between LSM2 expression and
   immunoinhibitors is shown in Fig. [126]11A, and the expression of LSM2
   has the strongest correlation with CD274 (Spearman: ρ = −0.267,
   p = 4.11e−09), PDCD1LG2 (Spearman: ρ = −0.267, p = 4.11e−09), TGFBR1
   (Spearman: ρ = −0.242, p = 1.1e−07), and CSF1R (Spearman: ρ = −0.216,
   p = 2.26e−06). We also compared the correlation between LSM2 expression
   and immunostimulators (Fig. [127]11B). The immunostimulators displaying
   the greatest correlation included TNFSF4 (Spearman: ρ = −0.315,
   p = 3.35e−12), TNFSF13B (Spearman: ρ = −0.311, p = 6.07e−12), IL2RA
   (Spearman: ρ = −0.305, p = 1.74e−11), and TNFSF14 (Spearman:
   ρ = −0.273, p = 1.94e−09). Figure [128]11C presents the correlation
   between LSM2 expression and MHC molecules, and the MHC molecules
   displaying the greatest correlation included TAPBP (Spearman:
   ρ = 0.248, p = 5.26e−08), B2M (Spearman: ρ = −0.239, p = 1.7e−07),
   HLA-A (Spearman: ρ = 0.21, p = 4.61e−06), and HLA-DOA (Spearman:
   ρ = −0.199, p = 1.33e−05).

Fig. 11.

   [129]Fig. 11
   [130]Open in a new tab

   Spearman’s correlation of LSM2 and LSM4 expression with
   immunomodulators (TISIDB). A. Spearman correlation between Expression
   (exp) of LSM2 and Immunoinhibitors across SKCM. Top 4 immunoinhibitors
   displaying the greatest Spearman’s correlation with LSM2 expression. B
   Spearman correlation between Expression (exp) of LSM2 and
   Immunostimulators across SKCM. Top 4 immunostimulators displaying the
   greatest correlation with LSM2 expression. C Spearman correlation
   between Expression (exp) of LSM2 and MHC across SKCM. Top 4 MHC
   molecules displaying the greatest relationship with LSM2 expression. D
   Spearman correlation between Expression (exp) of LSM4 and
   Immunoinhibitors across SKCM. Top 4 immunoinhibitors displaying the
   greatest Spearman’s correlation with LSM4 expression. E Spearman
   correlation between Expression (exp) of LSM4 and Immunostimulators
   across SKCM. Top 4 immunostimulators displaying the greatest
   correlation with LSM4 expression. F Spearman correlation between
   Expression (exp) of LSM4 and MHC across SKCM. Top 4 MHC molecules
   displaying the greatest relationship with LSM4 expression

   The correlation between LSM4 expression and immunoinhibitors is shown
   in Fig. [131]11D, and the expression of LSM4 had the strongest
   correlation with PDCD1LG2 (Spearman: ρ = −0.305, p = 1.61e−11), CD274
   (Spearman: ρ = −0.287, p = 2.59e−10), KDR (Spearman: ρ = −0.267,
   p = 4.31e−09), and TGFBR1 (Spearman: ρ = −0.258, p = 1.54e−08). We also
   compared the correlation between LSM4 expression and immunostimulators
   (Fig. [132]11E). LSM4 was significantly correlated with
   immunostimulators, such as TNFRSF14 (Spearman: ρ = 0.412, p < 2.2e−16),
   CD28 (Spearman: ρ = −0.311, p = 6.17e−12), IL2RA (Spearman: ρ = −0.301,
   p = 3.24e−11), and TNFSF13B (Spearman: ρ = −0.298, p = 4.85e−11). LSM4
   expression was also associated with MHC molecules (Fig. [133]11F),
   including HLA-DOA (Spearman: ρ = −0.268, p = 4.03e−09), HLA-DMB
   (Spearman: ρ = −0.226, p = 7.27e−07), B2M (Spearman: ρ = −0.2,
   p = 1.24e−05), and HLA-DRA (Spearman: ρ = −0.178, p = 0.000105). These
   results suggest that LSM2 and LSM4 affect the prognosis of patients and
   may play a role as immune regulators in SKCM.

Comprehensive results of LSM2

Expression pattern and prognostic value of LSM2

   Pan-cancer analysis indicated the aberrant expression of LSM2 in human
   cancers (Additional file [134]2: Figure S2). The results showed that
   LSM2 expression was significantly increased in 10 tumors, including
   diffuse large B-cell lymphoma (DLBC), glioblastoma multiforme (GBM),
   lower-grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung
   squamous cell carcinoma (LUSC), ovarian cancer (OV), pancreatic
   adenocarcinoma (PAAD), SKCM, testicular germ cell tumors (TGCT), and
   thymoma (THYM). Meanwhile, LSM2 expression was decreased in kidney
   chromophobe (KICH) and acute myeloid leukemia (LAML). To explore the
   effect of LSM2 expression on the prognosis of 33 types of cancer,
   univariate Cox regression analysis indicated that LSM2 overexpression
   was inversely correlated with poor OS in SKCM (p = 0.013), LIHC
   (p = 0.005), kidney renal clear cell carcinoma (KIRC) (p = 0.006),
   kidney renal papillary cell carcinoma (KIRP) (p = 0.005), OV
   (p = 0.024), rectum adenocarcinoma (READ) (p = 0.017), sarcoma (SARC)
   (p = 0.017), TGCT (p = 0.017), THYM (p = 0.013), uveal melanoma (UVM)
   (p = 0.003), and lung adenocarcinoma (p < 0.001). Therefore, LSM2
   overexpression in SKCM, LIHC, OV, TGCT, and THYM had similar negative
   effects on the prognosis. Detailed results are shown in forest charts
   (Additional file [135]3: Figure S3).

   Cox regression analysis was conducted to further investigate the role
   of LSM2 in SKCM (Fig. [136]12 A). In the univariate analysis, age,
   race, melanoma ulceration, TNM stage, pathologic stage, and LSM2
   expression were significantly related to the prognosis of SKCM in the
   TCGA database. In the multivariate analysis, race, T stage, N stage,
   pathologic stage II, pathologic stage III, and LSM2 expression were
   independent prognostic factors. Incomplete clinical information or the
   limited number of samples might have resulted in the above differences.
   To explore the prognostic ability of the above factors, we constructed
   a nomogram model for 1-, 3-, and 5-year OS prediction (Fig. [137]12B).
   The calibration curve displayed a desirable prediction of the nomograms
   for 1-, 3-, and 5-year clinical outcomes (Fig. [138]12C). Combined with
   the above data, LSM2 could be a useful factor for predicting prognosis
   in SKCM.

Fig. 12.

   [139]Fig. 12
   [140]Open in a new tab

   The Cox regression analysis of LSM2 in SKCM and the nomogram model
   established based on the expression of LSM2 in SKCM. A. The results of
   univariate and multivariate Cox regression analyses of clinical
   characteristics and LSM2 expression in the OS of patients with SKCM are
   represented by forest plots. B. A nomogram based on LSM2 expression for
   predicting the probability of 1-, 3-, and 5-year OS in patients with
   SKCM. C. The efficiency of nomograms for OS was validated using
   calibration plots

Functional enrichment analysis of LSM2 in SKCM

   Figure [141]13A and B displayed the top 50 genes positively and
   negatively associated with LSM2, respectively. LSM2 was positively most
   co-expressed with SNRPC (R = 0.806, p = 7.5226E-109), CSNK2B
   (R = 0.776, p = 9.592E−96), and ZNRD1 (R = 0.757, p = 7.861E−89)
   (Fig. [142]13C–E), and negatively most co-expressed with SLC41A2
   (R = −0.432, p = 8.435E−23), TMEM87B (R = −0.421, p = 1.256E−21), and
   SPAG1 (R = −0.410, p = 1.538E−20) (Fig. [143]13F, G, H). LSM2 and
   co-expression genes were enriched in 71 terms of GO BP, including viral
   gene expression, viral transcription, nuclear-transcribed mRNA
   catabolic process, nonsense-mediated decay, and nuclear-transcribed
   mRNA catabolic process, etc. LSM2 and co-expression genes were enriched
   in 63 terms of GO CC, including ribosomal subunit, ribosome, cytosolic
   ribosome, and small ribosomal subunit, etc. There were 19 terms of GO
   MF of those genes involved in, including structural constituent of
   ribosome, ribonucleoprotein complex binding, catalytic activity, acting
   on RNA, and rRNA binding (Fig. [144]13I). The KEGG [[145]35] analysis
   displayed that LSM2, and related genes were significantly enriched in
   ribosome, spliceosome, and ribosome biogenesis in eukaryotes
   (Fig. [146]13J)

Fig. 13.

   [147]Fig. 13
   [148]Open in a new tab

   LSM2 co-expression genes and functional enrichment analysis in SKCM. A
   and B Heatmap of the top 50 genes that were positively and negatively
   related to LSM2. C–E. The top three positively co-expressed genes in
   the LSM2. F–H. The top three negatively co-expressed genes in the LSM2.
   I Top four GO terms enriched in BP, CC, and MF. J KEGG enrichment of
   LSM2 co-expression genes, including ribosomes, spliceosomes, and
   ribosome biogenesis in eukaryotes. The genes involved in each pathway
   are displayed in the network

Discussion

   SKCM is the most aggressive type of skin cancer. Although SKCM
   represents a minority of all types of skin cancers, it causes the
   majority of skin cancer-related deaths worldwide [[149]2, [150]37,
   [151]38]. These data and the increasing number of cases make it
   imperative to identify more specific and sensitive prognostic
   biomarkers of SKCM. According to previous studies, the LSM family genes
   were involved in the various types of cancer [[152]26, [153]31,
   [154]32]. We hypothesized that LSM family members could serve as novel
   biomarkers for SKCM. However, no study has provided an overview of the
   entire LSM family in SKCM. To our knowledge, this study is the first to
   comprehensively explore the expression profile of LSM family members
   and to elucidate the relationships between LSM family members and the
   prognosis of patients with SKCM.

   In this study, we found that LSM2, LSM4, LSM8, LSM10, and LSM12 were
   highly expressed in SKCM tissues compared with normal skin tissues,
   whereas LSM1, LSM3, LSM5, LSM7, LSM11, LSM14A, and LSM14B were not
   significantly upregulated or downregulated. Consistently, another study
   revealed that LSM2 was overexpressed in basal-like primary tumors
   [[155]39]. Upregulated LSM4 expression promotes the growth and
   metastasis of HCC cells by regulating the cell cycle and focal adhesion
   pathways [[156]33]. In addition, LSM4 overexpression has been
   associated with rapid cancer progression in BC [[157]31, [158]40].
   Recently, LSM12 was found to be overexpressed in colorectal cancer
   (CRC) and to play an important role in cancer metastasis and
   progression [[159]41]. LSM1 downregulation is involved in prostate
   cancer progression [[160]27], whereas amplification of the LSM1 gene in
   luminal BC is significantly related to poor clinical outcome [[161]42]
   and LSM1 expression was significantly related to tumor stage in BC
   [[162]31]. LSM1 is overexpressed in 87% of human pancreatic tumor
   samples, and LSM1 knockout has been shown to have therapeutic efficacy
   in mouse pancreatic cancer models [[163]26]. LSM3 expression was
   correlated with progression-free survival and strongly associated with
   the metastatic phenotype of cervical carcinoma [[164]43]. LSM7
   overexpression is highly accurate in the preoperative diagnosis of
   thyroid nodules [[165]44]. These studies displayed a reverse comparison
   with our expression profile analysis of SKCM. This inconsistency may be
   due to tumor heterogeneity. Therefore, further in vitro and in vivo
   studies are required. The different expression characteristics of LSM
   family genes in cancers can be explained by factors such as genetic
   alterations, epigenetic changes, and tumor microenvironment (TME)
   [[166]45–[167]47]. In particular, RNA splicing, transcription,
   post-translational modification of proteins, and other interactions can
   lead to different phenotypes [[168]48–[169]50].

   Furthermore, in the HPA dataset, LSM2, LSM4, and LSM12 proteins were
   confirmed to be overexpressed in SKCM tissues compared with normal skin
   tissues. There are no data on LSM1 and LSM10 expression in the HPA
   dataset. However, in lung cancers and mesotheliomas, LSM1 protein was
   overexpressed compared to adjacent normal lung tissue, as measured by
   immunohistochemistry and western blot analysis [[170]51]. This
   indicates that the results of the dataset analysis need to be further
   validated by in vivo and in vitro experiments. In addition, we
   demonstrated that upregulated expression of LSM2 and LSM4 was
   significantly associated with shorter OS, whereas LSM6 overexpression
   was associated with longer OS. Additionally, univariate Cox regression
   analysis revealed that LSM2, LSM4, and LSM6 expression levels were risk
   factors influencing the prognosis of SKCM. Multivariate analysis
   concluded that LSM2 expression and TNM stage were related to adverse
   clinical outcomes in patients with SKCM. Moreover, a prognostic
   nomogram model based on TNM stage and LSM2 expression could increase
   the accuracy of identifying high-risk patients. Moreover, LSM2 and LSM4
   showed relatively high accuracy (AUC > 0.8) in predicting the prognosis
   of patients with SKCM. LSM2 was also an independent predictor of poor
   prognosis in other tumors [[171]28]. LSM2 overexpression in LIHC, OV
   TGCT, and THYM had similar negative effects on prognosis. The aberrant
   DNA methylation of LSM2 in the OV might lead to increased LSM2
   expression, thereby promoting tumor development [[172]28]. In terms of
   DFS, only LSM2 overexpression was associated with DFS in patients with
   SKCM. In addition, high expression of LSM2 was markedly related to high
   tumor stage. Another study also confirmed a significant relationship
   between upregulated LSM2 expression and advanced tumor stage [[173]31].

   To explore the cause of the aberrant expression of LSM family genes in
   SKCM, we further analyzed the genetic variations, GO enrichment, and
   KEGG annotation of LSM genes in SKCM. These results confirmed that LSM
   family genes were altered. The amplification and mutation of these
   genes might be one of the reasons for their aberrant expression. This
   was also confirmed in another study [[174]52]. Previous studies have
   revealed that LSM proteins are RBPs and play important roles in RNA
   metabolism [[175]53]. Sm protein binding is essential for the
   biogenesis and stability of small nuclear ribonucleoproteins (snRNPs).
   In this study, GO analysis also revealed that the LSM family and
   related genes were RBPs and were mainly involved in RNA binding and
   splicing. KEGG analysis of these genes revealed spliceosome and RNA
   degradation. Our results also confirmed that LSM genes were most
   closely related to some snRNPs, such as SNRPD2, SNRPD3, SNRPF, and
   SNRPE. The copy number variation of SNRPD2 was significantly associated
   with gene expression and poor prognosis in patients with HCC [[176]54].
   SNRPD2 and SNRPD3 were identified as splicing factors and were also
   critical for various proliferations in BC patients by regulating
   effective sororin splicing [[177]55]. An RBP-based prognostic
   prediction model for LUAD also indicated that SNRPE participates in the
   occurrence and development of cancer [[178]56]. In addition, SNRPE was
   found to be involved in cell proliferation and the progression of
   prostate cancer by regulating androgen receptor mRNA expression in
   cells [[179]57]. SNRPD2 and SNRPF are the core proteins related to
   below-background radiation, and further affect the proliferation of
   well-differentiated laryngeal squamous cell carcinoma cells [[180]58].
   In addition, some LSM genes can influence circadian rhythms by
   regulating RNA processing in mammals and plants [[181]59]. LSM3 was
   identified as a major pathogenic gene of AD, which involved RNA
   silencing and degradation [[182]29]. Increasing evidence has shown that
   post-transcriptional regulation plays a critical role in gene
   expression. RNA splicing is efficiently and precisely conducted by the
   spliceosome. However, in many cases, the splicing process is flexible
   enough to produce alternative transcripts that affect protein
   transcript levels and diversity [[183]60]. Given the importance of RNA
   splicing, alterations in this process have been implicated in several
   human cancers. Mutations in some RNA splicing factors, such as SF3B1,
   SRSF2, and U2AF1, have been found to be involved in the pathogenesis of
   cancers, such as lung cancer, breast cancer, and pancreatic cancers
   [[184]61–[185]63]. Tumor cells often take advantage of mutations in RNA
   splicing to promote growth and invasion [[186]64]. BRAF, NRAS, CDKN2A,
   and TP53 are known to be significantly mutated in cutaneous melanoma.
   Mutations in SF3B1 often occur in mucosal melanoma [[187]65]. Recently,
   one study confirmed that SF3B1 mutations occur in SKCM [[188]66].
   Another study also demonstrated that mutations in RBP, including LSM
   genes, were correlated with splicing aberrations in SKCM [[189]52].
   Herein, we discovered that LSM genes in SKCM mainly participate in RNA
   splicing and degradation. But whether the different alterations of LSM
   genes influences RNA expression and RNA splicing in SKCM patients and
   their mechanism of action remain elusive.

   In particular, LSM2 had the highest alteration rate (3%) among the LSM
   family. Altered LSM2 expression was significantly associated with
   shorter OS in patients with SKCM. Similarly, genetic variants of LSM2
   are associated with lung cancer risk [[190]67]. Our findings also
   suggest that LSM2 is an independent risk factor for poor LUAD
   prognosis. Functional enrichment analysis further confirmed that LSM2
   mainly participated in viral gene expression and RNA metabolism in
   eukaryotes, similar to the LSM family. We found that the top three
   genes co-expressed with LSM2 were SNRPC, CSNK2B, and ZNRD1. Gene
   expression data revealed that CSNK2B was highly expressed in CRC.
   Moreover, in vitro and in vivo experiments indicated that upregulated
   CSNK2B promotes CRC cell proliferation [[191]68]. Another study
   confirmed that SNRPC is involved in the regulation of spliceosomes,
   ribosomes, and proteasome signaling. SNRPC promotes HCC cell motility
   [[192]69]. Overexpression of ZNRD1 can promote the multidrug-resistant
   phenotype of GC cells via upregulation of P-glycoprotein [[193]70,
   [194]71]. Therefore, we hypothesized that LSM2 and its co-expressed
   genes may play a synergistic role in promoting the initiation and
   development of SKCM.

   Moreover, the occurrence and progression of tumors are not only related
   to gene mutations, but also involve complex interactions between
   malignant cells and TME. Previous studies have indicated that the
   immune infiltration of NK CD56 bright cells significantly affects the
   survival of SKCM [[195]72]. This finding is consistent with the results
   of the present study. High expression of LSM2 and LSM4 was positively
   related to immune infiltration of NK CD56 bright cells and
   significantly affected prognosis. Tcm has excellent anti-tumor ability
   [[196]73]. Tcm and its derived clonal T cells are highly effective
   anti-tumor immune T cells [[197]74]. Tcm can be activated by tumor
   antigens to directly kill tumor [[198]75]. In a melanoma patient, Tcm
   cells were found to have the ability to self-renew and replicate, and
   survive in vivo for a long time to achieve persistence of anti-tumor
   effects [[199]76]. In this study, Tcm infiltration was positively
   correlated with LSM8 and LSM12 but negatively correlated with LSM2 and
   LSM4, which was consistent with the effect of genes on tumor prognosis.

   Immune checkpoint inhibitors (ICIs) are an important new strategy for
   tumor immunotherapy, and have gradually improved the prognosis of
   patients with a variety of cancers [[200]77–[201]79]. In addition, we
   analyzed the associations between LSMs (LSM2 and LSM4) and the
   expression of immunoinhibitors, immunostimulators, and MHC molecules in
   various types of human cancers. Interestingly, we found that both LSM2
   and LSM4 had significant associations with CD274, PDCD1LG2, TGFBR1,
   TNFSF14, IL2RA, HLA-DOA, and B2M in patients with SKCM. TGFBR1
   [[202]80], PDCD1LG2 [[203]79], CD274 [[204]81], and TNFSF14 [[205]80]
   are involved in tumor immune evasion and are effective targets for
   tumor immunotherapy. Therefore, this study provides new insights into
   the roles of LSM genes in immune cell infiltration and their potential
   as immunotherapy targets in SKCM.

   This study has some limitations. First, the possible expression profile
   and functions of LSM2 have not yet been validated in vivo or in vitro.
   Further research on the function of LSM2 in clinical and basic
   experiments is required. Second, this study preliminarily explored the
   expression profile, prognostic value, and biological function of LSM
   family genes, especially LSM2, in SKCM. Bioinformatics analysis of
   genes using different databases or analysis tools might result in
   different results. Therefore, it is necessary to validate the LSM genes
   in clinical samples and cell lines.

Conclusions

   In summary, diverse associations between LSM family genes and SKCM were
   identified for the first time. Our findings revealed that LSM family
   members were differentially expressed in patients with SKCM.
   Furthermore, the LSM family played important roles in RNA metabolism.
   Upregulated LSM2 is significantly associated with shorter OS and DFS in
   patients with SKCM, which may be caused by abnormal RNA metabolism,
   LSM2 alteration or the regulation of immune check point genes.
   Therefore, LSM2 might be a potential poor prognostic indicator and
   immunotherapy target for SKCM.

Supplementary Information

   [206]12920_2022_1395_MOESM1_ESM.docx^ (620.8KB, docx)

   Additional file 1. Supplementary Fig S1. The expression of LSM family
   members in SKCM was analyzed using univariate Cox regression as a
   predictor of OS. The results are presented as forest plots. LSM2, LSM4,
   and LSM6 significantly influenced the OS of patients with SKCM.
   [207]12920_2022_1395_MOESM2_ESM.docx^ (3.5MB, docx)

   Additional file 2. Supplementary Fig S2. Expression profile of LSM2 in
   pan-cancer. Red indicates significant upregulation, while green
   indicates downregulation. Black indicates no significant differences.
   [208]12920_2022_1395_MOESM3_ESM.docx^ (1.4MB, docx)

   Additional file 3. Supplementary Fig S3. The forest plot indicates the
   results of OS analysis of LSM2 in pan-cancers by univariate Cox
   regression.
   [209]12920_2022_1395_MOESM4_ESM.docx^ (17.8KB, docx)

   Additional file 4. Supplementary Table S1. The correlation coefficients
   of the LSMS family genes in SKCM were explored using Spearman’s test.
   [210]12920_2022_1395_MOESM5_ESM.docx^ (17.7KB, docx)

   Additional file 5. Supplementary Table S2. The relationship between LSM
   family genes in SKCM was investigated using Spearman’s test. The
   p-values are presented in this table.

Acknowledgements