Pan-Cancer Analysis of the COVID-19 Causal Gene SLC6A20

Genome-wide association studies demonstrated that the chromosome 3p31.21 locus was linked to the severity of COVID-19 disease. The SLC6A20 gene was reported to be one of the critical causal genes regulated by this locus. Various studies focused on demonstrating the severity of COVID-19 in cancer patients and reported that elevated SARS-CoV-2-associated gene expression might contribute to increased susceptibility for COVID-19 in cancer patients. Given that pan-cancer association for the COVID-19 causal gene SLC6A20 is lacking, we aimed to perform systematic profiling of SLC6A20 in different malignancies. Human Protein Atlas, UALCAN, and Hepatocellular Carcinoma (HCCDB) databases were used to assess SLC6A20 gene expression changes in The Cancer Genome Atlas samples with respect to their normal counterparts. GEPIA and TIMER2.0 databases were used to determine the correlation between SLC6A20 and COVID-19-associated genes. Different databases were used for identification of the correlation of SCL6A20 with infiltrating immune cells. The canSAR database was utilized to determine the association of SCL6A20 with immune profiling in different malignancies. The STRING database was utilized to determine the protein network interacting with SLC6A20. Here, we showed SLC6A20 mRNA expression in pan-cancer samples and their normal counterparts. Increased SCL6A20 expression was associated with tumor grade, and there was a positive correlation with SARS-CoV-2-associated genes. Furthermore, SLC6A20 expression was positively correlated with infiltrating neutrophils and immune-related signatures. Lastly, SLC6A20 expression was found to be associated with the angiotensin converting enzyme 2 homologue, TMEM27, suggesting a potential link between SLC6A20 and COVID-19. Taken together, these results suggest that elevated SLC6A20 levels might be partly responsible for increased susceptibility of cancer patients to COVID-19 disease. Therapeutic intervention strategies against SLC6A20 in cancer patients, alongside other treatment modalities, might offer a benefit in delaying COVID-19 disease.


■ INTRODUCTION
SARS-CoV-2 outbreak started in 2019, and the World Health Organization (WHO) declared a pandemic in 2020. As of 30 December 2022, there have been approximately 663,380,366 reported cases with a total of 6,691,567 deaths (https://www. worldometers.info/coronavirus/). Viral envelope-mediated or direct fusion of the virus with cell membrane is facilitated by certain proteases whereby the viral spike protein (S1) is activated. 1 The interaction between the SARS-CoV-2 spike protein and the angiotensin converting enzyme 2 (ACE2) receptor mediates the entry of the virus into a cell. 2 Upon this interaction, protease cleavage within the spike protein is initiated. Within this process, various proteases including cathepsin L, cathepsin B, furin, and transmembrane protease serine 2 (TMPRSS2) are involved. 3 TMPRSS2 is one of the critical proteases in this process, activating the S protein for the entry of the virus into a cell. 4 Following the release of viral genome into the host cell, viral replicase proteins are translated, and subsequently these replicase proteins initiate the viral replication of genomic RNA. 5 Finally, the genomic RNA is translated into structural protein subunits of virus particles. 6 Cancer is a complex inflammatory disease and linked to an impaired immune system. 7−10 Several studies have shown that cancer patients can be more susceptible to SARS-CoV-2 infection than the general population. 11,12 Although in-depth understanding of the susceptibility of cancer patients to COVID-19 is still unclear, COVID-19-associated genes, ACE2, TMPRSS2, and TMPRSS4 were upregulated and coexpressed in different cancer types. 13 −15 In addition to pancancer studies, COVID-19 susceptibility in cancer patients was explained with elevated levels of endosomal entry and recycling proteins for SARS-CoV-2 infection. 16 Given the accumulating evidence documenting COVID-19 susceptibility in cancer patients, characterizing additional genes in these patients could contribute to identifying new targets.
Genome-wide association studies (GWASs) identified hostspecific genetic factors contributing to COVID-19. 17,18 Among these, it was reported that chromosome 3p21.31 is associated with COVID-19 18 Therefore, the chromosome 3p21.31 locus was investigated to delineate the causal gene for COVID-19 in this locus using both genome and epigenome editing. 19 The SLC6A20 gene was identified as one of the causal genes in the chromosome 3p21.31 locus for the severity of COVID-19. 19 In addition, it was demonstrated that SLC6A20 gene participates as a putative causal gene for the severity of COVID-19, using a genome-wide clustered regularly interspaced short palindromic repeats (CRISPR) loss-of-function study. 20 Given the contribution of SLC6A20 gene in COVID-19 severity and increased COVID-19 susceptibility observed in cancer patients, it was aimed to study the SLC6A20 expression profile and association with SARS-CoV-2 infection genes and immune modulation in pan-cancer data sets.
Here, we show systematic profiling of SLC6A20 expression in pan-cancer tumor and healthy samples together with correlation with SARS-CoV-2 infection genes ACE2, TMPRSS2, and TMPRSS4 and immune filtration in tumor samples. The SLC6A20 interacting protein TMEM27, an ACE2 homologue, and its pan-cancer expression profiling analysis suggests the involvement of SLC6A20 in providing increased susceptibility of COVID-19 disease in cancer patients. These findings suggest a potential therapeutic intervention strategy for SLC6A20 in cancer patients with the COVID-19 disease.

SLC6A20 Expression in Healthy Samples.
To investigate the tissue-specific SLC6A20 gene expression, a Human Protein Atlas (HPA) data set was explored. SLC6A20 gene was predominantly expressed in brain, gastrointestinal track, liver and gallbladder, and pancreas more than other tissues ( Figure  1A). Next, a GTEx data set from HPA was assessed, and SLC6A20 expression was detected at high levels at small intestine, pancreas, and kidney ( Figure 1B). Finally, using the Hepatocellular Carcinoma (HCCDB) database, 21 SLC6A20 expression was confirmed in a number of healthy tissues. The HCCDB database demonstrated that SLC6A20 was abundant in small intestine and nerve tissue, followed by kidney tissue ( Figure 1C). Collectively, these data show that SLC6A20 is expressed in a variety of normal tissues.
Expression Profile of SLC6A20 in Pan-Cancers. Using the pan-cancer The Cancer Genome Atlas (TCGA) data set, the SCL6A20 gene expression profile was investigated. Between the TCGA tumor samples, SCL6A20 was highly expressed in pancreatic adenocarcinoma (PAAD), kidney renal papillary cell carcinoma (KIRP), rectal adenocarcinoma (READ), stomach adenocarcinoma (STAD), and colon adenocarcinoma (COAD) ( Figure 2A). Next, SCL6A20 expression in different malignancies compared with normal samples was examined using UALCAN database 22 ( Figure 2B). SCL6A20 was found to be elevated in COAD (p < 0.001), KIRP (p < 0.001), liver hepatocellular carcinoma (LIHC) (p < 0.001), prostate adenocarcinoma (PRAD) (p < 0.001), READ (p < 0.001), STAD (p < 0.05), thyroid carcinoma (THCA) (p < 0.001), and uterine corpus endometrial carcinoma (UCEC) (p < 0.001) compared with the expression in healthy samples ( Figure 2B). On the other hand, SLC6A20 expression was downregulated in glioblastoma multiform (GBM), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), lung adenocarcinoma (LUAD), and lung squamous carcinoma (LUSC) samples when compared to normal ( Figure 2B). Taken together, these findings indicated that SLC6A20 was predominantly expressed in gastrointestinal tumors, which is one of the most frequently seen sites for SARS-CoV-2 infection in cancer patients. 14,23,24 SLC6A20 Expression Might Predict High Tumor Grade. Given the high expression of SLC6A20 gene in a variety of human cancers, the UALCAN database was utilized. Expression of SCL6A20 was found to be statistically higher in stage 1-to-4 in COAD and READ samples compared to normal samples ( Figure 3A,B). In KIRP, SCL6A20 expression was found to be higher only in stage 1 and stage 4 samples in comparison to normal samples, while the mRNA expression of SCL6A20 in stage 2 and stage 3 samples was not significantly upregulated ( Figure 3C). Furthermore, elevated SCL6A20 expression in ESCA samples was detected only in stage 1 and stage 3 samples but not in stage 2 and stage 4 samples in comparison to normal samples ( Figure 3D). Moreover, while SCL6A20 expression was not upregulated in any of the stages in PAAD samples, only stage 4 STAD samples were found to be upregulated in SCL6A20 expression in comparison to their normal control groups ( Figure  3E,F). In addition, SLC6A20 mRNA expression was upregulated in all of the stages in THCA samples and in stages 1, 2, and 3 in UCEC samples compared to their normal counterparts ( Figure  S1A, B). Collectively, these findings indicated that SCL6A20 expression might predict high stage risk in COAD, READ, ESCA, STAD, THCA, and UCEC patients. SLC6A20 Expression Is Correlated with ACE2, TMPRSS2, and TMPRSS4. Given the prominent roles of ACE2, TMPRSS2, and TMPRSS4 in SARS-CoV-2 infection, the relationship between these genes and SLC6A20 expression was investigated. The TIMER2.0 database 25 was utilized for the assessment of co-expression of SLC6A20 with ACE2, TMPRSS2, and TMPRSS4 in TCGA pan-cancer samples and showed that among the gastrointestinal cancers, ESCA, PAAD, and STAD together with UCEC and UCS exhibited positive correlation with SLC6A20 gene expression ( Figure 4A). This result was corroborated through the analysis of TGCA tumor and normal samples using the Gene Expression Profiling Interactive Analysis (GEPIA) database. 26 The GEPIA database demonstrated that SLC6A20 in ESCA samples was correlated with ACE2 (weak, R = 0.17, p = 0.019), TMPRSS2 (moderate, R = 0.52, p = 5.8 × 10 −15 ), and TMPRSS4 (weak, R = 0.35, p = 3.7 × 10 −7 ) ( Figure 4B). For PAAD samples, SLC6A20 was positively correlated with ACE2 (weak, R = 0.38, p = 1.4 × 10 −7 ), TMPRSS2 (moderate, R = 0.47, p = 2 × 10 −11 ), and TMPRSS4 (moderate, R = 0.4, p = 2.8 × 10 −8 ) ( Figure 4C). For STAD samples, SLC6A20 expression was correlated weakly with ACE2 (weak, R = 0.096, p = 0.044), TMPRSS2 (weak, R = 0.21, p = 7.2 × 10 −6 ), and moderately with TMPRSS4 (moderate, R = 0.33, p = 4.5 × 10 −13 ) ( Figure 4D). Additionally, in COAD samples, SLC6A20 was positively correlated with ACE2 (weak, R = 0.26, p = 2.3 × 10 −11 ), TMPRSS2 (weak, R = 0.1, p = 0.011), and TMPRSS4 (weak, R = 0.37, p = 0) ( Figure S2A). Moreover, SLC6A20 expression was correlated positively with ACE2 (moderate, R = 0.45, p = 0) and negatively with TMPRSS2 (weak, R = −0.23, p = 1.4 × 10 −5 ) and TMPRSS4 (very weak, R = −0.087, p = 0.1) in KIRP samples ( Figure S2B). Collectively, these results indicated that SLC6A20 was positively correlated with SARS-CoV-2 infection genes ACE2, TMPRSS2, and TMPRSS4 mostly in gastrointestinal tumors, namely ESCA, PAAD, and STAD. SLC6A20 Expression Is Detected in Innate and Adaptive Immune Cells. Given that the COVID-19 disease is linked with a cytokine storm and inflammatory associated complications, 27 the expression of SLC6A20 gene was investigated in various immune cells. The Monaco data set showed that neutrophils and basophils exhibited the highest levels of SLC6A20 expression, while the rest of the innate and adaptive immune cells had detectable amounts of SLC6A20 expression ( Figure 5A). Since SLC6A20 expression was detected at the highest level in neutrophils, TGCA pan-cancer tumor    samples were further examined using a TIMER2.0 data set. A correlation analysis using this data set to investigate a relationship of SLC6A20 expression with neutrophil infiltration levels was performed. This analysis demonstrated a significant correlation of SLC6A20 with neutrophil infiltration in a variety of tumors including ESCA, LIHC, PAAD, PRAD, BRCA, LUAD, KIRP, MESO, STAD, and UCEC ( Figure 5B). To support the link between SLC6A20 expression and COVID-19, the canSAR data set, providing information about the immune landscape of pan-cancer tumors, was utilized. 28 This analysis demonstrated the association of increased SLC6A20 expression with interferon-λ (IFN-λ), an immune related marker, and the inflammation signature particularly in THYM, READ, COAD, ESCA, and OV samples ( Figure 5C). Given that IFN-λ acts as an independent risk factor in COVID-19 infection 29 and inflammation is strongly associated with COVID-19 disease, 27 this result suggests a contribution of SLC6A20 expression for COVID-19 infection. Taken together, these findings indicate that elevated SLC6A20 expression might play a role in immune infiltration in a variety of human tumor samples.
PPI Network Analysis of SLC6A20 and Pan-Cancer Assessment of TMEM27 Expression. To investigate the interplay between SLC6A20 and its associated proteins, protein−protein interaction (PPI) network analysis was constructed using the STRING database. 30 This analysis revealed a total of 10 expected edges with PPI enrichment pvalue of 0.016 ( Figure 6A). Most of the remaining SLC6A20associated proteins in the same analysis included SLC family genes, SLC13A4, SLC7A9, SLC1A7, and SLC36A2 and additional proteins such as HAND1, XYLT2, KCNV1, and UGT8 ( Figure 6A). Among the list of proteins interacting with SLC6A20, TMEM27 (collectrin), a homologue of ACE2 gene, 31 was identified. Next, the involvement of TMEM27 was assessed in pan-cancer tumor samples using the UALCAN data set. This analysis revealed that TMEM27 expression was upregulated in BLCA, COAD, GBM, HNSC, LIHC, LUSC, PRAD, STAD, and THCA samples in comparison to their normal counterparts ( Figure 6B). Furthermore, the interplay between TMEM27 and SARS-CoV-2 infection genes ACE2, TMPRSS2, and TMPRSS4 was investigated using the TIMER2.0 database. The correlation analysis performed using this database demonstrated that TMEM27 was positively correlated with ACE2 in all malignancies except COAD, ESCA, MESA, READ, and TGCT ( Figure 6C). The same correlation analysis demonstrated that TMEM27 expression was positively correlated with TMPRSS2 in CHOL, GBM, KICH, PRAD, KIRP, LGG, LIHC, LUAD, LUSC, SKCM, BRCA, and UCEC ( Figure 6C). The same analysis exhibited a negative correlation between TMEM27 and TMPRSS4 in a number of pan-cancer samples including KIRC, KIRP, PAAD, SKCM, TGCT, and THCA, while there was a weak positive correlation in HSNC and THYM samples ( Figure 6C). Taken together, these findings indicated a potential involvement of SLC6A20 with COVID-19 susceptibility in different types of human cancers via the association between SLC6A20 and the ACE2 homologue TMEM27.

■ DISCUSSION
Rapid increase in the SARS-CoV-2 infection rates worldwide resulted in an immediate response to understand the underlying mechanisms of the COVID-19 disease, including in cancer patients. Cancer patients are considered as more prone to develop COVID-19 especially because of the impaired immune system response seen in them against the SARS-CoV-2 vaccine. 32,33 Studies demonstrated that ACE2, TMPRSS2, and TMPRSS4 expression were upregulated and facilitated SARS-CoV-2 infection in tumor samples. 14,34 In addition, certain hostspecific genetic factors have been shown to play a critical role in increased susceptibility to COVID-19 in cancer patients. 35 One of the causal genes identified to be linked with COVID-19 severity was SLC6A20 gene located in the chromosome 3p21.31; 19,20 however, its link with pan-cancer tumor samples remained to be elucidated. Here, we performed systematic analysis of SLC6A20 expression in different malignancies and demonstrated that SLC6A20 was upregulated in a wide number of human pan-cancer samples, and SLC6A20 expression was associated with ACE2, TMPRSS2, and TMPRSS4 genes. In addition, our findings indicate that there is an interplay between SLC6A20 expression and immune infiltration especially via increased amounts of neutrophils in a number of human tumor samples. Last, SLC6A20 protein interacting with TMEM27, an ACE2 homologue, suggests the potential involvement of SLC6A20 and TMEM27 in COVID-19 in cancer patients.
Two recent studies have characterized the 3p21.31 locus in the severity of COVID-19 disease. 19,20 These studies utilized the genome and epigenome editing approaches to reveal COVID-19 causal genes regulated by the 3p21.31 locus. Among the genes regulated by this locus, SLC6A20, CXCR6, and CCR9 were identified as causal genes to regulate COVID-19 risk in humans. Since SLC6A20 gene was the only gene independently identified in these two studies for COVID-19 risk, we have focused on SLC6A20 in human tumor samples to investigate SLC6A20 expression in human pan-cancer samples and its association with SARS-CoV-2-related genes and immune infiltration. Our findings indicate that SLC6A20 is highly expressed in a variety of human tumor samples, and there is a positive correlation with SARS-CoV-2 infection genes, ACE2, TMPRSS2, and TMPRSS4, suggesting that SLC6A20 might be involved in modulating COVID-19 disease and hence increased COVID-19 susceptibility for cancer patients.
Subsequently, the relationship between SLC6A20 and immune response in different malignancies was analyzed. The analysis demonstrated that SLC6A20 was highly correlated with neutrophil infiltration in majority of pan-cancer samples including ESCA, KIRP, PAAD, PRAD, BRCA, LIHC, MESO, LUAD, STAD, and UCEC. As elevated immune infiltration is linked with prognosis of human cancers 36,37 and found as an early indicator of COVID-19, 38,39 SLC6A20 might be involved in modulating COVID-19 in cancer patients. Moreover, the positive association found between SLC6A20 expression and neutrophil infiltration prompted us to think whether SLC6A20 might be involved in regulating the cytokine storm in neutrophils as they are known to be one of the sources of cytokine release in COVID-19 patients. 40,41 However, to support this claim, further functional studies will be needed.
In addition, immune subtypes inflammatory (C3), IFN-λ (C2), wound healing (C1), immunologically quiet (C5), lymphocyte depleted (C4), and TGF-β dominant (C6) analysis revealed an association between SLC6A20, particularly with inflammatory (C3) and IFN-λ (C2) signatures in THYM, READ, COAD, ESCA, and OV samples, suggesting a potential involvement of SCL6A20 in immune modulation in these cancer types. Given that cancer and COVID-19 share common modalities in terms of being inflammatory 42,43 and the role of IFN-λ 44,45 in both processes, our findings suggest the potential involvement of SCL6A20 in the pathophysiology of COVID-19 in cancer patients. Furthermore, efforts to treat severe COVID-19 patients with hypercytokinemia using emapalumab, an IFN-λ monoclonal antibody, is still underway with a registered clinical trial number of NCT04324021.
SLC6A20 gene encodes the sodium-dependent imino transporter 1 (SIT1) protein. 46 Importantly, it was reported that SIT1 was involved in viral entry and infection of SARS-CoV-2 in the human small intestine via heterodimerization with ACE2. 47 Furthermore, SIT1 was reported to regulate glycine and proline transport. 48 Moreover, SLC6A20 knockout mice exhibited elevated plasma levels of glycine, suggesting the regulatory role of SLC6A20 in glycine homeostasis. 49 On the contrary, ACE2 knockout mice demonstrated decreased levels of plasma glycine concentration, suggesting the potential use of glycine itself in the treatment of COVID-19 patients. 50 Indeed, it was proposed that glycine could intervene with anti-inflammatory cytokine storm seen in COVID-19 patients via interacting with its receptor GlyR to induce plasma membrane polarization and hence protection against the cytokine storm. 51 To support this mechanism, one of the drugs used in the treatment of COVID-19 patients was ivermectin, an agonist for glycine-gated chlorine channels. While additional studies are still needed to clarify the antiviral effect of ivermectin in COVID-19 disease, a number of studies demonstrated ivermectin in the treatment of COVID-19. 52,53 In this study, PPI enrichment analysis showed that TMEM27 was a SLC6A20-associated protein. TMEM27, a homologue of ACE2, 31 has previously been reported as being co-regulated with ACE2 in bronchial samples. 54 Given the prominent role of ACE2 in increased susceptibility in cancer patients and TMEM27 exhibiting 47% homology to ACE2 functions, we speculate that TMEM27 might be involved in the increased susceptibility in cancer patients for COVID-19. The evaluation of TMEM27 expression in pan-cancer samples resulted in elevated detection of TMEM27 mRNA expression in a number of cancer types such as COAD, GBM, HNSC, KIRC, KIRP, and PRAD. Alongside this observation, TMEM27 expression was found to be correlated predominantly with ACE2 and TMPRSS2 in pan-cancer samples, supporting the link between SLC6A20 binding protein TMEM27 and SARS-CoV-2 infection-related proteins, ACE2 and TMPRSS2 in different cancer types.

■ CONCLUSIONS
Collectively, our results suggest that SLC6A20 might be involved in the increased susceptibility of cancer patients to COVID-19. SLC6A20, known to be regulated by the 3p31.21 locus and linked with the severity of COVID-19 disease, was further examined in pan-cancer samples. The analysis of pancancer tumor samples in comparison to their normal counterparts exhibited the upregulation of SLC6A20 expression in a wide array of cancer types. In addition, SLC6A20 expression was positively correlated with COVID-19-associated genes, ACE2, TMPRSS2, and TMPRSS4, in a number of tumor samples. Given the role of ACE2, TMPRSS2, and TMPRSS4 in providing increased susceptibility to COVID-19 in cancer patients, our findings indicate that SLC6A20 might exhibit a similar function. Importantly, SLC6A20 expression was associated with increased immune modulation particularly through gained IFN-λ and inflammatory signatures in different cancer types. Last, SLC6A20 was found to interact with an ACE2 homologue protein TMEM27, and TMEM27 was upregulated in different cancer types along with its being positively correlated predominantly with ACE2 and TMPRSS2, which are known SARS-CoV-2-associated proteins. In conclusion, our study provides a systematic analysis of COVID-19 causal gene SLC6A20 in pan-cancer samples.

■ MATERIALS AND METHODS
Human Protein Atlas. The Human Protein Atlas is a webbased platform providing mRNA expression profiles of human genes. RNA-seq results obtained from 37 different types of human cancers and normal samples were visualized using the online available webpage.
UALCAN Database. The UALCAN database was used to analyze SLC6A20 gene expression in TCGA pan-cancer samples. The UALCAN database is an online webpage and provides an interactive environment to analyze the differential expression for a gene of interest in log 2 intensity in TCGA tumors including different pathological stages (stages 1, 2, 3, and 4). 22 HCCDB Database. The HCCDB is an interactive platform primarily for hepatocellular carcinoma but also includes RNAseq data for pan-cancer tumors. 21 Differential gene expression for SLC6A20 in tumor and normal samples was analyzed in log 2 intensity.
GEPIA Database. The GEPIA database provides an interface to perform gene expression correlation analysis between mRNA expression for selected genes using TCGA and normal samples. 26 Scatter plots were generated using the interactive GEPIA database.
TIMER2.0 Database. The TIMER2.0 database provides a systematic resource for cancer exploration and immune infiltrates in different malignancies in a webserver format. 25 The cancer exploration module was utilized to determine the correlation coefficients for SLC6A20 and ACE2, TMPRSS2, TMPRSS4 genes. In this analysis, the Spearman correlation was used, and when p-value was less than 0.05, it was significant. The immune gene module was used to determine the link with SLC6A20 and infiltration of immune cells in different malignancies. The partial correlation (cor) and p-values were obtained using the Spearman rank correlation test.

ACS Omega
http://pubs.acs.org/journal/acsodf Article PPI Network. The STRING database was utilized to determine SLC6A20 interacting proteins. 30 Using this database, the protein network interacting with SLC6A20 was identified.
Statistical Analyses. The results obtained from various TCGA expression databases provided a p-value, whereby *p < 0.5, **p < 0.01, and ***p < 0.001 were considered as significant. p-values were derived from the databases as follows. The UALCAN database utilized Welch's T-test for the significance between the difference in normal and tumor samples. The HCCDB database used the t-test function in R based on whether two groups had equal means or not followed by Benjamin− Hochberg correction. The UALCAN database performed t-test using a PERL script with the CPAN module. Pearson correlation analysis indicating ρ value: 0.00−0.19 (very weak), 0.20−0.39 (weak), 0.40−0.59 (moderate), 0.60−0.79 (strong), and 0.80− 1.0 (very strong) was followed. P-value less than 0.5 was considered as significant. GEPIA webtool provided the Pearson correlation coefficients based on performing pair-wise gene expression correlation analysis.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c00407. mRNA expression levels of SLC6A20 in different stages of TCGA tumor and the correlation analysis between SLC6A20 and ACE2, TMPRSS2, and TMPRSS4 (PDF) ■