LPS Administration Impacts Glial Immune Programs by Alternative Splicing

We performed transcriptome analysis in the hippocampus 24 h after lipopolysaccharide (LPS) administration. We observed glial-specific genes, comprised of two-thirds of all differentially expressed genes (DEGs). We found microglial DEGs that were the most numerous in LPS group. On the contrary, differential alternative splicing (DAS) analysis revealed the most numerous DAS events in astrocytes. Besides, we observed distinct major isoform switching in the Ptbp1 gene, with skipping of exon 8 in LPS group. Ptbp1 usually considered a pluripotency sustaining agent in brain embryonic development, according to the previous studies. Analyzing the splicing tune-up upon LPS exposure, we came to a supposition that the short Ptbp1 isoform de-represses immune-specific response by Ptbp1 adjusted splicing architecture. Additionally, the Ptbp3 (NOD1) immune-specific splicing factor has apparently been de-repressed by the Ptbp1 short isoform in glial cells. Notably, both the Ptbp1 and Ptbp3 genes express primarily in microglial/endothelial brain cells. We also report immune-related genes, altering their major isoforms upon LPS exposure. The results revealed immune modulating role of alternative splicing in brain.


Introduction
An increasing interest in the study of neuroinflammation is explained by its association with different neurodegenerative diseases. Inflammatory markers were detected in the rat brain even 2 years after the insult that can induce the neuropathology of Alzheimer's disease type [1]. In order to understand the molecular changes that contribute to the development of these diseases, lipopolysaccharide (LPS), the component of the outer membrane of Gram-negative bacteria, is widely used for induction of the central inflammatory process [2]. Injections of LPS also resulted in gene expression changes in brain structures, including the hippocampus [3], brain region that was implicated in the development of neurodegenerative-related psychopathology. Inflammatory-and apoptosis-related genes were revealed among these genes.
Alternative splicing manifests the flexible and rapidly evolving mechanism [4,5] of expanding the proteome diversity. The diversity is particularly vivid in the nervous system [5]. Neuroinflammation related CD44 alternative splicing was shown to affect the response of the hippocampus in the Alzheimer disease [6]. The synaptic proteins of Nrxn1−3 were shown in response to neuroinflammation by altering their isoforms ratio [7].
Along with the mature neurons, there are stage specific splicing events featuring embryonic brain development. In particular, the previous studies have identified splicing factor Ptbp1 as one of the major factors affecting the transition from neural stem cells (NSC) to neural progenitor cells (NPS) in the course of brain-specific embryogenesis [8][9][10].
Principal component analysis (PCA) has been performed by commercial XLSTAT software (accessed on 10 May 2021; https://xlstat.com). As a distance matrix, a Pearson pairwise correlation matrix of DEGs expression profiles was used. Agglomerative hierarchical clustering (AHC) was employed using XLSTAT software. The AHC parameter set was: (similarity: Pearson correlation coefficient; agglomeration method: unweighted pair-group average; center: no; reduce: no; truncation: automatic-inertia). Heatmap construction was performed with a self-organizing map (SOM) algorithm employing XLSTAT software.

GO DEGs Analysis
Due to the brain invasion and consequent post-operational inflammation in both samples (SAL and LPS), we filtered the DEGs sample for the genes maintaining abs (log2 fold) >2. This resulted in final DEGs set of 87 entries (Table S1, Supplementary Materials).
Analysis of the gene ontology (GO) in the biological process yielded 284 entries (Table S2, Supplementary Materials). To obtain the core DEGs set, we applied a GO clustering algorithm GOMCL (Wang et al., 2020), which yielded just one GO supercluster consisting of 37 distinct DEGs, encompassing the 284-fold initial GO BP annotation list (Table S1, Supplementary Materials). This way we were able to see the non-redundant DEGs network, represented in Figure 1.
We employed PCA analysis ( Figure 2a) outlining three clusters depicted in Figure 1 to proceed on details of each cluster. We found that Hcrt, attenuated in the LPS group, refers to hormone activity (Figure 2b), while two other clusters manifest inherent distinct coordinated responses to LPS exposure (Figure 2c,d). Elucidating glial-specific genes in the cluster using brain cells expression atlas, presented in [22] we were able to relate 24 genes as glial-specific ones (Figure 2e,f).
We found Shox2 DEG, induced by an outlier value in the SAL1 sample, equaling 7.4 FPKM, while five other samples maintaining it as less that 0.07 (avg = 0.05; stddev = 0.03), so we discarded it.
Overall, we came to 36 DEGs as the core DEGs spanning the GO spectra. Microglialspecific genes were prevalent (22 DEGs) followed by endothelial genes (Fos, Sele; Figure 2e) and astrocytes display 2 genes (Serping1, Fcnb). Notably, Fos DEG expresses virtually in all cell types, most prominently in astrocytes [22] implying the assignment of this gene to epithelial cells may be due to Sele DEG co-variation. Other DEGs were not immediately assigned to glial cells; still, more than half of DEGs (24 entries) were assigned to be glial rather confidently.  Table S1. Colors on the plot correspond to those on the figure. Number of DEGs per GO term attached as bar labels. (b) Heatmap of 37 DEGs, clustering underlines three clusters: small one (Hcrt and Shox2), attenuated in LPS group, and two others (specific to the LPS group).
We employed PCA analysis (Figure 2a) outlining three clusters depicted in Figure 1 to proceed on details of each cluster. We found that Hcrt, attenuated in the LPS group, refers to hormone activity (Figure 2b), while two other clusters manifest inherent distinct coordinated responses to LPS exposure (Figure 2c,d). Elucidating glial-specific genes in the cluster using brain cells expression atlas, presented in [22] we were able to relate 24 genes as glial-specific ones (Figure 2e,f).  Table S1. Colors on the plot correspond to those on the figure. Number of DEGs per GO term attached as bar labels. (b) Heatmap of 37 DEGs, clustering underlines three clusters: small one (Hcrt and Shox2), attenuated in LPS group, and two others (specific to the LPS group).

DAS Genes Analysis
We identified 49 differentially, alternatively spliced genes with ES event alteration FDR <0.05 between SAL and LPS groups, presented in Table S3, Supplementary Materials. Only one DAS event per gene was observed. While no significantly enriched GO terms were featured for the set, we ascribed them manually.
A total of 49 genes, featuring DAS ES events, maintain 9 median isoforms per gene (Table S4, Supplementary Materials), as annotated in the NCBI repository. We observed 3 median isoforms per gene for the same 49 DAS genes in our RNA-seq data, when including all ES events detected. The Pearson correlation of the RNA observed and expected (NCBI) numbers is quite significant (equaling r = 0.665, df = 45 (excluding three non-annotated RNA), p < 1 × 10 −5 ) and implying we maintain good sensitivity with our AS detection algorithm.
Pursuing further analysis and keeping in mind the previous studies on neuroinflammastion, we suppose that: (a) the major role in LPS response attributed to glial cells; (b) non-random DAS ES events invoke the corresponding genes expression modulation (up or down).  Based on [22], the brain cell-specific expression profiles we assessed showed 49 for cell specific expression ( Figure 4). We identified the gene as a cell specific on when it maintains the highest expression among seven cell types. As expected, the most AS events took place in neuron specific genes (13-16 DAS genes). Still, we observed distinct glia specific genes in Figure 4 (colored gene names in the first column). 9 genes were not found in Atlas. In particular, cep126, cep295 are absent, while 21 cep genes family was found highly expressed specifically in endothelial cells (order of magnitude higher than in any other cell type), but further were omitted from the analysis due to high non-specificity to LPS issue. No noncoding (ncRNA)/unannotated ones were found in Atlas, in particular: Spaca, AABR07019088.1, AABR07039316.1, AABR07069473.1. Ogt was used for annotation of Eogt, Stk30 for Mok. Both Stk30 and Ogt are astrocyte specific genes. Thus, we maintained 42 genes as annotated for cell specificity.

Sampled DAS Genes
Based on Figure 4 we defined 3 glia specific DAS groups comprising 17 genes. To further confirm their specificity within glial cells we built PCA plot for 3 glial cells ( Figure 5). Most of the DAS genes were observed in astrocytes, according to Figure 5. Overall, we compiled 24 glial-specific DAS genes presented in Table 1.

Cell-Specific Expression of DAS Genes
First, we assessed the distribution of 49 DAS gene (DAS FDR < 0.05) expression profiles across SAL and LPS samples ( Figure 3). Upon analysis of Figure 3 we may state there is certain preference of DAS genes to SAL/LPS samples, based on their PCA of expression profiles, depicted in Figure 3. In particular, 17 glial DAS genes augmented their expression in the LPS group ( Figure 3; bold typed genes, orange circles), while only 6 glial DAS genes attenuated it in LPS group (Figure 3; blue circles).

Endothelial Cells Microglia Astrocytes
Based on [22], the brain cell-specific expression profiles we assessed showed 49 for cell specific expression ( Figure 4). We identified the gene as a cell specific on when it maintains the highest expression among seven cell types. As expected, the most AS events took place in neuron specific genes (13-16 DAS genes). Still, we observed distinct glia specific genes in Figure 4 (colored gene names in the first column). 9 genes were not found in Atlas. In particular, cep126, cep295 are absent, while 21 cep genes family was found highly expressed specifically in endothelial cells (order of magnitude higher than in any other cell type), but further were omitted from the analysis due Ogt was used for annotation of Eogt, Stk30 for Mok. Both Stk30 and Ogt are astrocyte specific genes. Thus, we maintained 42 genes as annotated for cell specificity.

Sampled DAS Genes
Based on Figure 4 we defined 3 glia specific DAS groups comprising 17 genes. To further confirm their specificity within glial cells we built PCA plot for 3 glial cells ( Figure  5). Most of the DAS genes were observed in astrocytes, according to Figure 5. Overall, we compiled 24 glial-specific DAS genes presented in Table 1.

Endothelial Cells
Microglia

Rhog Gene
Rhog (Ras homolog gene family, member g (rho g)) coordinates response to 'bacterial invasion of endothelial cells' as is underlined by GO enrichment term ( Figure 6).

Rhog Gene
Rhog (Ras homolog gene family, member g (rho g)) coordinates response to 'bacterial invasion of endothelial cells' as is underlined by GO enrichment term ( Figure 6). We observed Rhog DAS ES event as 5′ UTR exon 4 skipping ( Figure 6; right panel) implying it expression elevation in LPS group ( Figure 3). Additionally, we observed distinct elevated expression of long isoform in LPS compared to SAL group, along with higher expression of short one ( Figure 6; right panel). Exon 4 skipping is manifested as a major isoform in both SAL/LPS groups ( Figure 6).  We observed Rhog DAS ES event as 5 UTR exon 4 skipping ( Figure 6; right panel) implying it expression elevation in LPS group ( Figure 3). Additionally, we observed distinct elevated expression of long isoform in LPS compared to SAL group, along with higher expression of short one ( Figure 6; right panel). Exon 4 skipping is manifested as a major isoform in both SAL/LPS groups ( Figure 6). Figure 7 demonstrates non-linear coordinated elevation/downturn of Atp13a2, Afmid isoforms expression (Figure 3).

Ptbp1 Exon 8 Skipping Preference
Strikingly, we also observed significant exon 8 skipping in neurospecific splicing factor, Ptbp1 (Figure 8), responsible for splicing alteration of several thousand genes [23]. Along with its expression elevation in LSP (Figure 9), its splicing skews towards to exon 8 skipping ( Figure 5) implies putative immunomodulation in astrocytes/microglia/endothelial cells in LPS1 species, since Ptbp1 short isoform weakens its polypyrimidine tract binding [23]. As exon 8 skip in Ptbp1 alters target exon inclusion rate based on Ptbp1 binding preference [23], this way altering exons inclusion rate (both up and down), we may speculate it serves as a switch to turn on immune response by altering exon inclusion/skipping events ratio in specific genes. Based on annotation of 4 microglial cell specific genes we may state that DAS underscores specific major isoform enhancing. Three genes (Ptbp1, Atp13a2, and Rhog) have their

Ptbp1 Exon 8 Skipping Preference
Strikingly, we also observed significant exon 8 skipping in neurospecific splicing factor, Ptbp1 (Figure 8), responsible for splicing alteration of several thousand genes [23]. Along with its expression elevation in LSP (Figure 9), its splicing skews towards to exon 8 skipping ( Figure 5) implies putative immunomodulation in astrocytes/microglia/endothelial cells in LPS1 species, since Ptbp1 short isoform weakens its polypyrimidine tract binding [23]. As exon 8 skip in Ptbp1 alters target exon inclusion rate based on Ptbp1 binding preference [23], this way altering exons inclusion rate (both up and down), we may speculate it serves as a switch to turn on immune response by altering exon inclusion/skipping events ratio in specific genes.

Ptbp1 Exon 8 Skipping Preference
Strikingly, we also observed significant exon 8 skipping in neurospecific splicing factor, Ptbp1 (Figure 8), responsible for splicing alteration of several thousand genes [23]. Along with its expression elevation in LSP (Figure 9), its splicing skews towards to exon 8 skipping ( Figure 5) implies putative immunomodulation in astrocytes/microglia/endothelial cells in LPS1 species, since Ptbp1 short isoform weakens its polypyrimidine tract binding [23]. As exon 8 skip in Ptbp1 alters target exon inclusion rate based on Ptbp1 binding preference [23], this way altering exons inclusion rate (both up and down), we may speculate it serves as a switch to turn on immune response by altering exon inclusion/skipping events ratio in specific genes. Based on annotation of 4 microglial cell specific genes we may state that DAS underscores specific major isoform enhancing. Three genes (Ptbp1, Atp13a2, and Rhog) have their  Based on annotation of 4 microglial cell specific genes we may state that DAS underscores specific major isoform enhancing. Three genes (Ptbp1, Atp13a2, and Rhog) have their expression rate increased along with altering isoforms ratio (Figure 3), while Afmid decreased its expression in LPS (Figure 3) featuring skipping of coding exon and thus being a subject to nonsense mediated decay (NMD) due to ORF disruption.

Endothelial Specific DAS Genes
Herein we annotate the endothelial cells specific DAS genes (Table 2): Cast, Phactr4, Nebl, Lrp8, and Pkn2. Figures 9-13  Phactr4 gene was shown to be involved in mediation of transient response potential regulation upon inflammatory response (Figure 10a) featuring short isoform expression elevation in LPS sample (Figures 3 and 10b).   Pkn2 (Serine/threonine-protein kinase N2) is involved in endothelial cell migration (GO:0010594), also featuring intracellular signal transduction in response to stress ( Figure  11). DAS ES event in this gene features its activation by elevation major long isoform in LPS group leading to its enhanced expression (Figure 3).

Ndufb3 vs. AABR07019088 Genes
Ndufb3 is a member of MWFE subunit of the mitochondrial NADH-ubiquinone oxidoreductase (complex I) is a small, essential membrane protein of 70 amino acids, which is made in the cytosol, imported into mitochondria, and assembled without further proteolytic processing. AABR07019088 is antisense RNA of Ndufb3 (Figure 12a). Both genes elevated their long isoforms expression in LPS group (Figures 3 and 12b), Ndufb3 isoform featuring alternative translation start site (TSS; Figure 12a).

Cast Gene
Cast is calcium-dependent cysteine protease. Based on annotation (Figure 9), we may state that cast is involved in 'degradation of extracellular matrix' and may be implicated in immune response processes.

Phactr4 Gene
Phactr4 gene was shown to be involved in mediation of transient response potential regulation upon inflammatory response (Figure 10a) featuring short isoform expression elevation in LPS sample (Figures 3 and 10b).

Pkn2 Gene
Pkn2 (Serine/threonine-protein kinase N2) is involved in endothelial cell migration (GO:0010594), also featuring intracellular signal transduction in response to stress ( Figure 11). DAS ES event in this gene features its activation by elevation major long isoform in LPS group leading to its enhanced expression ( Figure 3).
We observed the two most significant DAS ES events (Table S3, Supplementary Materials) in Ank2 (DAS FDR < 0.006) and Eogt (DAS FDR <0.00014) genes altering its major isoforms upon LPS exposure ( Figure 14). Both these genes also express in neurons ( Figure  4). Notably, while Ank2 increased its expression in LPS, Eogt was attenuated (Figure 3), due to 5′ UTR related exon skipping implying glycosylation process downturn.

Ndufb3 vs. AABR07019088 Genes
Ndufb3 is a member of MWFE subunit of the mitochondrial NADH-ubiquinone oxidoreductase (complex I) is a small, essential membrane protein of 70 amino acids, which is made in the cytosol, imported into mitochondria, and assembled without further proteolytic processing. AABR07019088 is antisense RNA of Ndufb3 (Figure 12a). Both genes elevated their long isoforms expression in LPS group (Figures 3 and 12b), Ndufb3 isoform featuring alternative translation start site (TSS; Figure 12a).
We observed the two most significant DAS ES events (Table S3, Supplementary Materials) in Ank2 (DAS FDR < 0.006) and Eogt (DAS FDR <0.00014) genes altering its major isoforms upon LPS exposure ( Figure 14). Both these genes also express in neurons (Figure 4). Notably, while Ank2 increased its expression in LPS, Eogt was attenuated (Figure 3), due to 5 UTR related exon skipping implying glycosylation process downturn.
We observed the two most significant DAS ES events (Table S3, Supplementary Materials) in Ank2 (DAS FDR < 0.006) and Eogt (DAS FDR <0.00014) genes altering its major isoforms upon LPS exposure (Figure 14). Both these genes also express in neurons ( Figure  4). Notably, while Ank2 increased its expression in LPS, Eogt was attenuated (Figure 3), due to 5′ UTR related exon skipping implying glycosylation process downturn.
Biomolecules 2022, 12, x 18 of 26 Figure 14. Switching major isoforms in two highly significant DAS genes. Ankyrine 2 (Ank2) gene increases short isoform expression upon LPS exposure, while Eogt Glycosyltransferase gene expressed, mostly in astrocytes, is switched off by DAS altering 5′UTR region.
The rest of DAS genes referring mainly to transcription process and apparently featuring metabolism intensity increase in LPS (Figure 3), all manifesting long isoforms in LPS group, are presented in Figure 15. The rest of DAS genes referring mainly to transcription process and apparently featuring metabolism intensity increase in LPS (Figure 3), all manifesting long isoforms in LPS group, are presented in Figure 15.

Fnbp1 Gene
Fnbp1 is Formin binding protein 1. It is required to coordinate membrane tubulation with reorganization of the actin cytoskeleton during the late stage of clathrin-mediated endocytosis. (DAS FDR < 0.027). We observed its major isoform elevation, accompanied with attenuation of alternative skipped isoform (Figure 16).

Splicing Factors (SF) in Brain
We assessed 15 key SFs expression in the brain using Barres Lab atlas for mature brain cell specific expression [22]. As can be seen from Figure 17, the majority of brainspecific SFs are neuron specific ones, as was reported earlier, in a range of publications starting from 2005 [24,25]. Still, there is a range of SFs, in particular, Ptbp1, expressing preferentially in microglia and endothelial cells ( Figure 13). As we observed, Ptbp1 exon 8 skipping as highly specific for LPS exposed species (Figure 4), we speculated it might affect immunomodulation in response to LPS. Figure 14. Switching major isoforms in two highly significant DAS genes. Ankyrine 2 (Ank2) gene increases short isoform expression upon LPS exposure, while Eogt Glycosyltransferase gene expressed, mostly in astrocytes, is switched off by DAS altering 5′UTR region.
The rest of DAS genes referring mainly to transcription process and apparently featuring metabolism intensity increase in LPS (Figure 3), all manifesting long isoforms in LPS group, are presented in Figure 15.

Fnbp1 Gene
Fnbp1 is Formin binding protein 1. It is required to coordinate membrane tubulation with reorganization of the actin cytoskeleton during the late stage of clathrin-mediated endocytosis. (DAS FDR < 0.027). We observed its major isoform elevation, accompanied  We assessed 15 key SFs expression in the brain using Barres Lab atlas for mature brain cell specific expression [22]. As can be seen from Figure 17, the majority of brainspecific SFs are neuron specific ones, as was reported earlier, in a range of publications starting from 2005 [24,25]. Still, there is a range of SFs, in particular, Ptbp1, expressing preferentially in microglia and endothelial cells ( Figure 13). As we observed, Ptbp1 exon 8 skipping as highly specific for LPS exposed species (Figure 4), we speculated it might affect immunomodulation in response to LPS. Figure 17. SFs expression preference in seven brain cell types [22]. Besides mostly neuron specific SFs (Nova, Ptbp2, Rbfox, SLM1, and 2) we may see four SFs preferential ones in glial cells (Ptbp1, Rod1, Mbnl1, and U2af2).
From Figure 18 we may state that glia specific SFs (Figure 18a) are elevated in LPS sample (Figure 18b), implying its involvement in LPS response.  Figure 17 across three glial cell types point to microglial and endothelial cells SFs activity, while astrocytes essentially lack SFs Figure 17. SFs expression preference in seven brain cell types [22]. Besides mostly neuron specific SFs (Nova, Ptbp2, Rbfox, SLM1, and 2) we may see four SFs preferential ones in glial cells (Ptbp1, Rod1, Mbnl1, and U2af2).
From Figure 18 we may state that glia specific SFs (Figure 18a) are elevated in LPS sample (Figure 18b), implying its involvement in LPS response.
We assessed 15 key SFs expression in the brain using Barres Lab atlas for mature brain cell specific expression [22]. As can be seen from Figure 17, the majority of brainspecific SFs are neuron specific ones, as was reported earlier, in a range of publications starting from 2005 [24,25]. Still, there is a range of SFs, in particular, Ptbp1, expressing preferentially in microglia and endothelial cells ( Figure 13). As we observed, Ptbp1 exon 8 skipping as highly specific for LPS exposed species (Figure 4), we speculated it might affect immunomodulation in response to LPS. Figure 17. SFs expression preference in seven brain cell types [22]. Besides mostly neuron specific SFs (Nova, Ptbp2, Rbfox, SLM1, and 2) we may see four SFs preferential ones in glial cells (Ptbp1, Rod1, Mbnl1, and U2af2).
From Figure 18 we may state that glia specific SFs (Figure 18a) are elevated in LPS sample (Figure 18b), implying its involvement in LPS response.

Discussion
We conducted RNA-seq analysis of rat hippocampal tissues upon LPS exposure compared to saline administered ones. Analysis revealed 609 genes with significant difference in expression rate at q-value < 0.05, which were further filtered with log2 fold > 2 criterion, due to surgical invasion procedure in both SAL/LPS groups and consequent background inflammation in both groups (Table S1, Supplementary Materials). That left 93 DEGs that were subject to gene ontology annotation, yielding 284 GO biological process terms. Using GOMCL toolkit [21], we elucidated single DEGs cluster based on ontology and consequent non-redundant 36 DEGs, encompassing all GO terms present in initial sample.
We present the key relevant events in Table 2, while full GO annotation can be found in Supplementary Table S2, Supplementary Materials. Based on Table 3 we may state we witness quite clear LPS effect within our sample compared to SAL control group. The 3 major clusters of coordinated genes. Based on DEG analysis, (Figures 1 and 2) we state that LPS response is quite robust (612 DEGs overall) and single directed (majority of DEGs are elevated in LPS group; Figure 2). The 36 core DEGs co-varied clusters are depicted in Figure 2. We should stress that the majority of DEGs are highly expressed glial genes. The cell specific analysis reveals the majority of DEGs are microglial specific ones following endothelial and astrocyte cells (Figure 2e,f).
Given AS in brain primarily refers to neurons primarily expanding their synaptic plasticity [26] and has been widely addressed therein; we decided on assessing AS-mediated LPS response specifically in glial cells for gaining the insight on it. Analysis of alternative splicing reveals a group of 50 structural and signaling genes being differentially alternatively spliced (see Table S3, Supplementary Materials). No particular abundance in GO categories were inferred for the DAS gene set. Still, while expanding the gene neighborhood of DAS genes there were relevant DAS genes networks affecting rho-gtpase activity, platelet-related genes (F8, wVag1; [27]) and a range of others (Figures 6-16).
With that, we should outline the basic challenges of splicing analysis we have observed. Due to relatively low read coverage in our study (about 30 mln reads per transcriptome), we might miss some DAS genes, along with maintaining DAS low confidence (sensitivity) in a range of cases. Recently reported tandem splice sites (TASS) persistence reported in RNA-seq data [28] were also observed in our data quite abundantly (about 1.15 thousand, across 10,000 ES events total average per sample) given only 20% proved being nonspurious/sensible splice sites [28], also distorted the overall picture. We applied the filter of minimal exon junction counts to be more than five, totaling both groups, compared.
Additionally, based on our preliminary assessments with protein database, at least half of the ES events maintain open reading frame (ORF) disruption and invoking nonsense mediated decay (NMD), thus being purely the means of prompt regulation of homeostatic balance of particular genes expression [29].
Another fundamental problem is that, in contrast to expression temporal course, the splicing dynamics manifests a much more velocity, implying there are large pools of unspliced transcripts in the pool [30] waiting for the splicing 'decision'. Splicing alteration may be performed in a matter of milliseconds and be restored back immediately, especially for many auto-/cross-regulated splicing factors and chromatin rearrangement machinery by utilizing NMD routine while altering (downing) gene expression would take relatively longer time.
Functional interpreting ES events is also a keystone. We approached this subject at least by validating coding/non-coding transcript variants resulted in the course of ES events. We revealed that only 8 DAS genes (Ptbp1, Fnbp1, Med12l, MegF11, Cast, Phc3, Cep295, and Ank2) maintain both coding isoforms, 19 DAS genes maintain long coding isoform only, no genes with skipped isoform coding only, and 23 remaining DAS genes maintain ES events in 3 /5 UTRs. Elaborating on coding ES renders protein interaction analysis and, while not being already annotated experimentally, is hard to exemplify. Still, new means of protein isoforms annotation are approaching with an advent of AlfaFold resource [31].
Ptbp1 is one of the key neurogenesis factors, also used in reprogramming neural cells [8,[32][33][34] We elucidated it is primarily expresses in glial cells (Figure 4), while highly repressed in neurons. Intriguingly, recent study demonstrated morphing astrocytes into dopaminergic neurons when depressing Ptbp1 [34]. They did it with virus vectors while natural antagonist of Ptbp1 in embryogenesis is mir-124 [8]. With that, the highest expression of Ptbp1 is observed in microglia and endothelial cells.
As we found essential splicing factor Ptbp1 being significant DAS gene (FDR < 0.03) in our set, we decided exploring brain-specific splicing factors genes group, since these affect splicing the most and, while being altered and impact several thousand of relevant exons reported elsewhere [5,8,23]. Upon the analysis, we report glial SFs manifest augmented expression in LPS group ( Figure 13) emphasizing their relevance.
The mammalian specific Ptbp1 exon 8 (9 in human) skipping (27aa) manifests one of current enigmas not fully apprehended. It reported to be involved in embryonic stage tuneup after NSC -> NPC stage splicing tune up [23]. It is reported therein "exon 9 possesses splicing regulatory activity that is partially separable from the repressive activity conferred by RRM2 and that skipping of exon 9 reduces the negative and positive regulatory activities of PTBP1 without substantially affecting RNA binding activity", implying significant effect of the ES9 event.
Notably, according to GTEX v.7 data (gtexportal.org), Ptbp1 ES9 short isoform (ENST00000349038; NM_031991) is a major one with average expression rate of 31 transcripts per million (TMP) across 53 tissues, while exon 9-in most abundantly expressed variant (ENST00000350092; 5 UTR truncated isoform) manifests 16 TPM average, with other 16 ones manifesting 3.5 TPM and less, still comprising overall around 40% of Ptbp1 total expression rate per tissue. There is only one coding ES9 isoform in human Ptbp1 isoforms spectra, except for extremely rare transcript ENST00000627714.2, which simultaneously lacks all RNA recognition motifs (RRMs) coding exons and not detected in GTEX data. Two top expressed Ptbp1 isoforms co-vary with Pearson r = 0.81 (df = 52; p < 1 × 10 −7 ).
Based on analysis of Ptbp1 expression organism wide with GTEX human resource, we elucidated that it's a redundantly expressed gene observed in virtually all cells except for mature neurons, and immune competent organs including heart, whole blood, pancreas, liver (Table 3a). Oppositely, cancer-related tissues/cells manifest high Ptbp1 expression rates, as reported in studies [11,35]. Ptbp1 deletion enhances MHC II expression in dendritic cells [13], therefore attenuating its binding affinity by ES8 may do the work. Based on glial cells analysis, we may state that Ptbp1 is significantly expressed in epithelial cells, as well ( Figure 4).
Given Ptbp1 is primarily a glia specific gene among differentiated brain cells, we found its overall expression elevated specifically in LPS administered group, though not significantly (Figure 3. p-value < 0.33), at the same time featuring its major isoform, significantly altering toward ES8 one (Figure 8), implying some stress response activity. Expanded Ptbp1 isoforms spectra (17 entries) allows for proposing Ptbp1 isoform expression ratios, which may impact the immunogenic potential of glial cells, judging by the dynamics it manifests upon LPS response in the hippocampus brain region.

Conclusions
While many of DAS events relate to the modulation of expression rates (increasing/decreasing), we may underline the enhanced accuracy of splicing routine in the course of gene expression elevation (often non-significant one), since short isoform (usually noncoding one) preference often manifest attenuation in a vast range of cases. We report that, while DAS genes and DEGs do not overlap, as a rule (e.g., [36]), it was observed that NMD employed coding gene expression abrogation correlates with its expression rate downturn, implying certain feedback routines at the transcription layer [37,38]. To this end, we observe certain glial DAS genes change its expression rate in nonrandom manner (Figure 3).
While we cannot immediately, functionally interpret the AS ES events observed, besides overall expression rate alteration, we correspond that there are plenty of statistically significant AS ES alterations in glial genes, in particular. Underlining their abundance, further research may shed light on the functional meaning of DAS ES events, besides NMD-related ones. We stress the DAS of Ptbp1 as one of the potent events possibly leading to massive alteration of splicing landscape upon LPS administration.