Pharma Intelligence Applied to the Research and Development of Cell and Cell-Free Therapy: A New Era of the Medicine

Abstract

Cellular and cell-free therapies have provided novel therapeutic opportunities for treating various incurable diseases. This is because the mesenchymal stromal/stem cells (MSCs) produce a plethora of bioactive molecules able to target different biological pathways through extracellular vesicle-mediated paracrine mechanisms simultaneously. However, to share transcriptomic signatures with their origin tissue, it is expected that each MSC population has a unique molecular profiler. In this sense, to analyze the transcriptome of therapeutic cells, it is crucial to identify the molecular profiler of these cells to predict the potential clinical benefits promoted by these cells. Herein, we discuss the application of high-throughput RNA-sequencing (RNA-Seq) in the Pharma Intelligence Era, discussing and exemplifying how the combination of molecular biology with Analytics can revolutionize the Research, Development, & Innovation (RD&I) of advanced medicinal products.

1. Introduction

Since their discovery, mesenchymal stromal/stem cells (MSCs) have been proposed as the best candidate for treating non-curable diseases [1, 2, 3, 4, 5, 6, 7].

Stem cells can be obtained from both embryonic and adult tissues. Embryonic stem cells (ESCs) are derived from the inner cell mass of the human blastocyst at the pre-implantation stage. However, hESC are pluripotent cells and can differentiate into all cell types, which compose adult organisms. Therefore, when transplanted in an animal model, these cells form teratomas—benign tumors that contain derivate of all three germ layers, the most typical ones being bone, cartilage, neuronal rosettes, and epithelium of the airways and gut [8]. By contrast, human adult stem cells (hASCs) comprise a population of undifferentiated precursor cells that naturally reside in all tissues and organs. However, hASCs show limited differentiating capability (multipotent cells), and they do not confer any risk of forming teratomas. This is one of the reasons why hASCs are the most studied type of stem cell for therapeutic purposes.

The bone marrow hASCs are classified as stromal cells or mesenchymal stem cells (MSCs). Stromal cells are progenitors of the hematopoiesis-supporting stroma and are components of skeletal tissue such as bone, cartilage, and adipocytes. MSCs can be isolated from numerous adult and perinatal tissues, such as bone marrow (BM), umbilical cord (UC) vein, Wharton’s jelly, adipose and placental tissues, menstrual blood (MB), liver, spleen, and dental pulp (DP) [2, 3, 9, 10]. Moreover, MSCs can be easily expanded in vitro and in accordance with Good Manufacturing Process (GMP) conditions, which is crucial to guarantee a large-scale production of therapeutic cells for therapeutic purposes.

For a long time, it has been suggested that the therapeutic properties of stem cells were based on the capacity of these cells to replace dead cells in injured tissues [3, 11]. However, studies showed that transplanted MSCs remain in the injury site for a few days and, subsequently, are not found in the tissue [11, 12, 13]. These results were later confirmed by numerous independent studies [10, 14, 15], which reported that less than 1% of transplanted MSCs survive for more than 1 week after systemic administration [16, 17]. Historically, these studies broke the paradigm that the therapeutic action of these cells is based on their capability to replace dead cells. Despite the therapeutic properties and the facilities to isolate and produce these cells, the mechanism of action (MoA) of the MSCs remains not completely understood.

Supporting this breaking paradigm, Takahashi et al. [18] showed that injection of MSC conditioned culture medium (CCM) increases the capillary density, reduces the damage size, and improves the cardiac function in murine models for heart attack. This result, combined with previous evidence that less than 1% of MSCs survive for more than 1 week after systemic administration, provided strong evidence that the therapeutic properties of MSCs are mediated by the “secretome” of these cells, which is composed of a plethora of bioactive molecules [2, 19, 20].

To date, several bioactive molecules have already been identified in the MSCs’ secretome, such as chemokines cytokines, interleukins, growth factors, lipid steroids, nucleotides, nucleic acids, ions, and metabolites [2, 21, 22]. These molecules can be found in free form (when directly released into the extracellular space) or within extracellular vesicles (EVs), which are recognized as a key component of paracrine secretion [6, 22, 23].

The rediscovery of EVs and the discovery that these vesicles mediate cell-to-cell communication led to the conceptualization of the paracrine stimulation theory [11, 24]. This theory postulates that MSCs release a variety of bioactive molecules into the intercellular space, playing a key role in cross-talk communication between the cells and the surrounding tissues [1, 2, 3, 10, 25]. Thus, the MSCs’ secretome is the key mediator of the regenerative action of these cells [6, 11, 21, 26]. For this reason, in 2017, Prof. Dr. Arnold Caplan, who isolated for the first time the MSC population for therapeutic purposes, proposed to rename these cells to signaling cells [27].

The paracrine stimulation theory not only brought novel horizons to elucidate the MoA of stem cells but also allowed us to explore the biotechnological potential of these EVs in a new therapeutic modality known as cell-free therapy [2, 6, 22, 28].

2. Cell-free therapy: novel horizons for the treatment of non-curable diseases

The discovery that the MSC conditioned culture medium (CCM) elicits therapeutic properties similar to the MSCs [18] aroused the interest in exploring the content of this CCM as an alternative to cell therapy. However, CCM contains components such as bovine fetal serum, antibiotics, and pH indicators that limit the direct application of this medium for clinical purposes. For this reason, the CCM is discarded as a subproduct of the MSC manufacturing process. Thus, biopharmaceutical companies are discarding a valuable source of feedstock rich in bioactive molecules that could be explored for clinical purposes in cell-free therapy [6, 28].

Although the MSC-derived bioactive molecules could be found in a free form diluted in the CCM, generally, in this form these bioactive molecules are easily degraded or oxidized. By contrast, when released within EVs, these bioactive molecules are preserved, increasing the half-life of these molecules [6, 28]. Therefore, it is not surprising that cell-free therapy is based on the therapeutic use of these EVs, justifying the increased interest in these vesicles, as illustrated by the number of publications involving EVs shown in Figure 1.

Figure 1.

Number of publications involving extracellular vesicles from 1956 to October 10, 2023 (A). Numbers show an increased interest in these vesicles, which has increased since 2008. Results also show an increased interest in exosomes from 2004 (B), a specific class of EV. However, the number of publications using exosomes derived from MSCs has been reduced since 2010 (B).

Although the interest in these vesicles had emerged two decades ago (Figure 1), the EVs were discovered in the second half of the 1940s, when Chargaff, studying blood coagulation, observed small “membrane debris” sedimented at high-speed centrifugation of plasma supernatant [29, 30]. Twenty years later, after Chargaff’s report, Peter Wolf confirmed the existence of corpuscles, “platelet dust,” through electron microscopy [31]. However, the biological nature of these corpuscles remained unknown until 1974, when Nunez et al. [32] identified structures with a size under 1000 nm, called multivesicular bodies (MVBs), opening up the path in the identification of a subtype of EVs that later was called exosomes or small EVs (30–150 nm) [33].

The EVs comprise a heterogenous group of small cell-released particles [34], which are classified according to their size and biogenesis in exosomes (30–200 nm), microvesicles (150–1500 nm), and apoptotic bodies (1000–5000 nm) [6, 35, 36]. Among these EVs, exosomes are the most studied and useful type of EV for therapeutic purposes [6, 36]. For this reason, we will focus on this class of EV.

Exosomes are nanosized vesicles, surrounded by a phospholipid membrane, containing cholesterol, sphingomyelin, ceramide, lipid rafts, and evolutionarily conserved markers such as tetraspanins (CD9, CD63, and CD81), heat shock proteins (HSP60, HSP70, and HSP90), major histocompatibility component (MHC) classes I and II, Alix, TSG101, lactadherin, and lysosome-associated membrane glycoprotein 2 [6].

Exosomes constitutively originate from late endosomes, which are formed by inward budding of the limited multivesicular body (MVB) [6, 37, 38].

MVBs and late endosomes are a subset of specialized endosomal compartments rich in intraluminal vesicles (ILVs) [39]. During the ILV formation process, certain proteins are incorporated into the invaginating membrane, while the cytosolic components are engulfed and enclosed within the ILV [38, 40, 41].

Along the biogenesis, the endosomal sorting complex required for transport (ESCRT), which consists of ESCRT-0, -I, -II, -III, combined with the vacuolar protein sorting 4-vesicle trafficking 1 (VPS4-VTA1), and some accessory proteins such as the ALG-2-interacting protein X (ALIX) homodimer [42] promotes the sorting of specific biomolecules to the exosomes [43].

Upon maturation, the MVBs can be transported to the plasma membrane via the cytoskeletal and microtubule network and undergo exocytosis post-fusion with the cell surface, whereby the ILVs get secreted as exosomes [40, 44]. Alternatively, MVBs can also follow a degradation pathway either by direct fusion with lysosomes or by fusion with autophagosomes followed by lysosomes [39]. Although both the secretory and degradative MVB pathways coexist, the mechanism that influences these pathways remains unclear [39].

Once released into the extracellular space, exosomes interact with the extracellular matrix and recipient cells, serving as natural vehicles for the nanodelivery of nucleic acids, proteins, metabolites, and lipids [6, 36, 39].

Considering that all cell types naturally produce and secrete exosomes, all cell types can be considered a source of exosomes [6, 28]. However, for therapeutic purposes, the MSCs emerge as the most important source of therapeutic exosomes. This is because, as previously discussed, MSCs can be easily isolated and produced on a large scale. In addition, these cells naturally produce a plethora of bioactive molecules that can elicit therapeutic responses. Thus, considering that during the biogenesis, the ESCRT complex promotes the sorting of specific molecules that may be driven to the lumen of these nanosized vesicles, it is expected that the molecular cargo of exosomes changes according to their origin. Therefore, no prototype MSCs can serve as a universal source of exosomes to treat all types of diseases. For this reason, knowing the molecular profiler (transcriptomic signature of MSCs) can help us to identify appropriate exosomes for the treatment of a set of diseases. In this regard, RNA-Seq emerges as a useful tool to identify the molecular profiler of these cells and their respective exosomes and, with this, to predict the clinical application of these cells and/or exosomes.

3. RNA-Seq analysis applied to identification of molecular profiler of ATMPs

The advent of high-throughput techniques presents a multitude of opportunities to the biopharmaceutical industry [45, 46]. Among these techniques, transcriptome, which includes RNA sequencing (RNA-Seq) and single-cell RNA sequencing (scRNA-Seq), stands out due to their capability to assess qualitative and quantitatively the gene expression levels, serving as important tools to characterize MSCs and MSC-derived exosomes with therapeutic interest. While RNA-Seq can be applied to characterize a cell or exosome population, scRNA-Seq is useful to analyze the heterogenicity of cell type in the MSC population. This chapter will focus on RNA-Seq since this technique is most useful to characterize MSC or MSC-derived exosome populations and predict their therapeutic properties.

Due to the complexity of the RNA-Seq data, several recommendations have been proposed to guarantee the quality of the results obtained with this technique [46, 47, 48, 49]. In general, these recommendations are focused on experimental questions, such as the number of samples, sequencing depth, quality control, read alignment, and quantification of gene levels [47, 48, 49]. However, there is no single analysis pipeline to be used in all cases. In this sense, the multiple bioinformatic tools, especially those applied to assess the differential expression, make the RNA-Seq data analysis a laborious task, requiring basic knowledge of genetics, bioinformatics, and statistics [46].

Statistical analyses comprise a mandatory step to identify differentially expressed genes (DEGs), which can be used to characterize the molecular profile and predict the therapeutic potential of MSCs or MSC-derived exosomes. Although the statistical techniques required for these purposes are included in the packages used to analyze the RNA-Seq, the choice for inappropriate controls and/or the lack of the most powerful analytics techniques, particularly those based on artificial intelligence (AI), can negatively affect the power of the transcriptome to predict the therapeutic potential of cell and cell-free products.

However, the combined use of traditional statistical methods with Analytics tools to analyze Pharma Big Data such as those generated by the RNA-Seq has revolutionized both pharmaceutical and biopharmaceutical industries. The combination of these techniques applied to analyze (bio)pharmaceutical data led to the advent of Pharma Intelligence, which has revolutionized the research, development, & innovation (RD&I) in life sciences [46].

4. Pharma Intelligence applied to the identification of transcriptomic signature of different types of MSCs

Pharma Intelligence (PI) is a generic term employed to refer to the combined use of mathematics, statistics, Data Analytics (or Analytics), Business Intelligence (BI), bioinformatics, and artificial intelligence (AI) to analyze and extract valuable information from big data produced in the multi-Omics Era [46].

Pharma Intelligence emerged as a need of the pharmaceutical sector that, over the last two decades, evidenced numerous disadvantages of the traditional empirical drug development model [46]. The traditional empirical drug development model is based on meeting regulatory requirements to guarantee the investigational product (IP) gets market access instead of providing the most useful details about the IP, such as the eventual side effects and even the mechanism of action (MoA), which could predict the drug repositioning, accelerating the RD&I in the pharmaceutical area [46]. For this reason, historical evidence of the traditional empirical drug development model not only increases the time to obtain the registry of new (bio)pharmaceutical products for the treatment of NCDs but also causes economic losses to the entire pharmaceutical supply chain. In this regard, technologies that could predict the eventual risk of the IP, and MoA are crucial to mitigate the failure rate of the IP during the clinical trials.

On the one hand, the advent of “Omics” technologies (genomics, transcriptome, proteome, and metabolome) revolutionized the RD&I in the (bio)pharmaceutical sector; on the other hand, these multi-Omics tools have contributed to the generation of high amount of data (Big data) that are not efficiently used by (bio)pharmaceutical industry to generate valuable insights [46, 47, 50, 51]. Moreover, due to the complex nature of the pharmaceutical data, which combine structured (tabular data) and non-structured data (image bank, sound, and video recording from clinical trial), the conventional statistical techniques are not sufficient to efficiently explore these datasets [46]. This occurs because standard statistical tests assume that a single hypothesis is tested. However, when multiple hypotheses are tested simultaneously, there is a higher probability of a type I error (false-positive)—an incorrect rejection of a true null hypothesis [46, 52]. However, PI has allowed us to overcome the challenges imposed by the Pharmaceutical Big Data Era, helping to select variables of interest, revealing unexpected correlations among these variables, and significantly reducing the time and cost of drug discovery [46, 52, 53].

In addition, the PI possibilities the reanalyze of datasets available on public data banks, such as the Sequence Read Archive (SRA) of the National Institute of Health (NIH, https://www.ncbi.nlm.nih.gov/sra), facilitating the RD&I of ATMPs.

5. Pharma intelligence applied to prediction of therapeutic potential of MSCs and MSC-derived exosomes

To share a transcriptomic signature with their origin tissue, each MSC population is expected to possess different molecular profiles. Thus, it is not surprising that there is no universal MSC that can confer therapeutic benefits for all diseases. For this reason, identifying the molecular profile (or transcriptomic signature) of each MSC population is crucial to predicting the therapeutic potential, as well as the mechanism of action (MoA) of these cells.

In this sense, the SRA can serve as a useful source of RNA-Seq datasets of MSCs and MSC-derived exosomes to predict the appropriate use of each MSC or MSC-derived exosome for target disease.

Generally, standard analysis of RNA-Seq, using conventional packages (DSeq2, NOISeq, edgeR, etc.) to assess the genes differentially expressed, provides noise results. This is because these packages are based on statistical tests that provide p-values. Thus, all gene that shows a p-value <0.05 is recognized as differentially expressed. However, slight biological expression differences result in a significant p-value (p-value <0). Although commonly employed, analysis using this analytical approach can result in the identification of numerous genes/transcripts with low biological relevance. Alternatively, some bioinformaticians have calculated the log2 fold-chance (log2FC) and adopted a cutoff value that could justify a phenotypic difference (i.e., a biological response), assuming that this cutoff represents a statistically significant difference. However, using appropriate analytical strategies based on PI, it is possible to combine the log2FC to statistical tests to select genes of interest accurately for downstream prediction analyses.

This strategy was currently used to assess the transcriptional profile of human immature dental pulp stem cells (hIDPSCs) produced on a large scale [54]. In order to increase the predictor power of this method, the study compared the hIDPSCs transcriptional profile with the transcriptome signature of other MSCs obtained from adipose tissue (AD-MSC, BM-MSC (HD-MSC, hepatocyte-derived-MSC), menstrual blood (MB-MSC), UC-MSC and vertebral tissue (vMSC) [54]. For this, the study employed a dataset comprised of more than 130 samples from the SRA [54]. Results obtained by the PI showed that the hIDPSCs have a unique transcriptome when compared with other analyzed MSCs [54]. The study also identified that the hIDPSCs express 375 unique genes (Table 1), which are not expressed by other MSCs [54]. These unique genes were subjected to the biological functional enrichment analysis. Interestingly, the results of these analyses suggest that the hIDPSCs have a higher capacity to mediate cell communication when compared to other MSCs and that the genes exclusively expressed by the hIDPSCs regulate metabolic processes and neurogenesis [54]. To confirm experimentally these results, primary motor neurons from transgenic mice for amyotrophic lateral sclerosis (ALS) were treated with three different concentrations of exosomes isolated from the conditioned culture medium of these cells for up to 14 days. Results showed that the exosomes derived from the hIDPSCs were able to promote both neurite and axon guidance length in a concentration- and time-dependent manner [54]. These data provide evidence that the PI applied to the analysis of RNA-Seq of therapeutic cells has an accurate predictor power.

	Number of genes
MSC population	Total identified	Commonly expressed (%1)	Exclusively expressed
AD-MSC	7581	5912 (77.98%)	13
BM-MSC	6912	5912 (85.53%)	0
HD-MSC	7739	5912 (76.39%)	93
hIDPSC	8127	5912 (72.75%)	375
MB-MSC	9593	5912 (61.62%)	832
UC-MSC	7011	5912 (84.32%)	0
vMSC	7970	5912 (74.17%)	114

Table 1.

Genes commonly e exclusively expressed in each MSC population.

¹

In relation to the total genes identified/MSC population.

Using this same analytical approach (based on PI), herein we reanalyzed these data, showing that the MSCs share the expression of 5912 genes in common, which represents between 61.62% (MB-MSC) to 85.53% (BM-MSC) of the genes encoded by these cells (Table 1).

Aiming to predict the biological pathways in which these 5912 genes commonly expressed by all MSCs, it was employed the same functional enrichment technique used to predict the hIDPSCs therapeutic properties. Results showed that 98.34% (5814/5912) genes commonly expressed by all MSCs encode proteins, which are involved in (i) metabolic process (4246) and (ii) biological regulation (3877), (iii) response to stimulus (2811), (iv) cellular component organization (2557), and (v) cell communication (1968) (Figure 2). The high number of cell communication-related genes identified as commonly expressed by the MSCs reinforces the paracrine stimulation theory, suggesting that the therapeutic effects of MSCs can be mediated by the bioactive molecules naturally produced and secreted by these cells within EVs.

Figure 2.

Biological functional enrichment in terms of biological process of the 5814 genes commonly expressed by the MSC populations analyzed. Results were obtained using the WEB gestalt tool, available from: https://www.webgestalt.org/.

Biological function enrichment also showed that the genes commonly expressed by the MSCs are involved in (i) transcriptional and (ii) translational regulation, being associated with protein folding and polyubiquitination (Table 2). Interestingly, the products of these genes are also involved in (iii) mitochondrial regulation, (iv) cell survival, (v) angiogenesis, and (vi) vesicle-mediated transport (Table 2). These data show that MSCs have a complex mechanism of action since the proteins encoded by these cells can act synergically and simultaneously in multiple biological pathways, which are found deregulated in most diseases.

	Enrichment result	FDR	p-value	Gene set size	Overlap
Geneontology (GO)	Regulation of catabolic process (GO:0009894)	0.000	0.000	875	510
	Positive regulation of catabolic process (GO:0031331)	0.000	0.000	348	204
	Vesicle-mediated transport (GO:0016192)	0.000	0.000	1942	899
	Translational initiation (GO:0006413)	0.000	0.000	192	154
	RNA binding (GO:0003723)	0.000	0.000	1603	1056
	RNA processing (GO:0006397)	0.000	0.000	487	337
	Protein targeting (GO:0006605)	0.000	0.000	412	264
	Protein folding (GO:0006457)	0.000	0.000	210	131
	Protein polyubiquitination (GO:0000209)	0.000	0.000	255	155
	Ubiquitin ligase complex (GO:0000151)	0.000	0.000	273	166
	Mitochondrial protein complex (GO:0098798)	0.000	0.000	266	167
Pathway (KEGG, Pnather, Reactome, Wikipathay)	Protein processing in endoplasmic reticulum (has04141)	0.000	0.000	165	125
	Ubiquitin-mediated proteolysis (hsa04120)	0.000	0.000	136	100
	Ubiquitin proteasome pathway (P00060)	0.000	0.000	44	37
	p38 MAPK pathway (P05918)	0.000	0.000	34	24
	RAS pathway (P04393)	0.000	0.000	70	48
	PI3 kinase pathway (P00048)	0.000	0.000	47	30
	PDGF signaling pathway (P00047)	0.000	0.000	125	72
	Integrin signaling pathway (P000340	0.000	0.000	166	95
	P53 pathway (P00059)	0.000	0.000	79	44
	Transcription regulation of TP53 (R-HAS-3700989)	0.000	0.000	365	221
	Vesicle-mediated transport (R-HAS-5653656)	0.000	0.000	667	397
	Cellular response to stress (R-HAS-2262752)	0.000	0.000	426	242
	PTEN regulation (R-HAS-6807070)	0.000	0.000	140	104
	mRNA processing (WP411)	0.000	0.000	126	108
	Electron transport chain (OXOPHOS system in mitochondria) (WP111)	0.000	0.000	103	71
	VEGFA-VEGFR2 signaling pathway (WP3888)	0.000	0.000	431	280
	EGF/EGFR signaling pathway (WP437)	0.000	0.000	162	109

Table 2.

Biological functional enrichment of the genes commonly by all analyzed MSC population.

FDR – false discovery ratio; GO – Geneontology; has – KEGG; P – Panther; R-HAS – Reactome; WP – Wikipathway.

Using the same approach, we also analyzed the set of genes identified as exclusively expressed by the MB-MSCs, since this MSC population showed the highest number of unique genes, followed by the hIDPSCs (Table 1). Results showed that 96.39% of the genes exclusively expressed by the MB-MSCs (802/832) encode proteins, which are related to (i) metabolic process (480), (ii) biological regulation (410), (iii) response to stimulus 251), (iv) component organization (223), and other biological process, including (v) cell communication (Figure 3).

Figure 3.

Biological functional enrichment in terms of biological process of the genes exclusively expressed by the MB-MSCs (A) and hIDPSCs (B) results were obtained using the WEB gestalt tool, available from: https://www.webgestalt.org/.

Biological function enrichment analyses of the genes exclusively expressed by the MB-MSCs showed that most of the products encoded by these genes bind to the DNA, acting as transcriptional factors, regulating cellular responses to DNA damages, or are involved in DNA replication, as part of the replication fork (Table 3).

	Enrichment result	FDR	p-value	Overlap
Geneontology (GO)	RNA modification (GO:0009541)	0.000	0.000	41
	ncRNA processing (GO:0034470)	0.000	0.000	38
	DNA repair (GO:0006281)	0.000	0.000	43
	Cellular response to DNA damage stimulus (GO:0006974)	0.000	0.000	62
	DNA replication (GO:0006260)	0.000	0.000	27
	Replication fork (GO:0005657)	0.034	0.000	10
	DNA-binding transcriptional factor activity (GO:0003700)	0.000	0.000	114
	Methyltransferase activity (GO:0008168)	0.000	0.000	27
Pathway	Chromosome maintenance (H-SRA-73886)	0.000	0.000	17
	Ubiquitin-mediated proteolysis (hsa04120)	0.000	0.000	100
	Mitotic spindle checkpoint (H-SRA-69618)	0.000	0.000	16
	Cell cycle (H-SRA-1640170)	0.000	0.000	49
	Gene expression (transcription) (H-SRA-1640170)	0.000	0.000	49

Table 3.

Biological functional enrichment of the genes exclusively expressed by MB-MSC population.

FDR – false discovery ratio; GO – Geneontology; R-HAS – Reactome.

Interestingly, the reanalyzing of the 375 genes exclusively expressed by the hIDPSCs showed that, although the unique genes found in hIDPSCs are involved in the same biological process regulated by the unique genes found in MB-MSCs (Figure 3), they govern distinct biological pathways (Table 4), conferring different therapeutic properties for these two populations of MSC. This is because the unique genes expressed by hIDPSCs are involved in neurogenesis and synapsis (Table 4), suggesting that these cells can be explored for the treatment of neurological disorders, such as neurodegenerative diseases.

	Enrichment results	FDR	p-value	Overlap
GO	Neurogenesis (GO:0022008)	0,000	0,000	51
	Neuron spine (GO:0044309)	0,007	0,000	10
	Dendritic spine (GO:0043197)	0,0000	0,007	10
	Neuron to neuron synapsis (GO:0098984)	0,019	0.000	13

Table 4.

Biological functional enrichment of the genes exclusively expressed by hIDPSC population.

FDR – false discovery ratio; GO – Geneontology.

Confirming the predictor power of the Pharma Intelligence applied to cell therapy, preclinical studies involving the intravenous administration of the hIDPSCs in rat models for Huntington’s (HD, neurodegenerative disorder characterized by a progressive loss of medium spiny neurons (MSNs) in the striatum, causing motor, cognitive and neuropsychiatric disorders) provided evidence that these cells can cross the brain-blood barrier and engraft within the brain areas involved in the HD pathophysiology. Besides this, the study showed that the hIDPSCs administration increases the expression of MSN biomarkers, suggesting a neuroregenerative action [55]. In another independent study, using a rat model for Parkinson’s disease, it was also observed that the intravenous administration of these cells ameliorated the motor function of these animals only 3 days after the cell transplantation [56]. Similar results were also demonstrated in phase I clinical trial for HD, which provided evidence that the transplantation of hIDPSCs can improve the motor function of patients with HD [57].

6. Perspectives of the use of pharma intelligence in cell and cell-free therapy

Although already commonly used by pharmaceutical companies to accelerate the RD&I in drug discovery, the complex nature of MSCs and/or MSC-derived exosomes still has limited the application of Pharma Intelligence as part of the characterization and, as a tool to predict the therapeutic potential of cells and exosomes. However, fortunately, today, we already have specialists and facilities providing innovative solutions that combine mathematical, statistics, analytics, and bioinformatic tools to extract valuable information from multi-Omics, preclinical, and even clinical datasets. Although the number of specialists and companies dedicated to employing Pharma Intelligence in RD&I of cell and cell-free therapy remains reduced when compared to the number of companies offering services dedicated to drug discovery, PI is no longer a promise for the future of these therapies. PI is the New Era of advanced medical therapies.

Moreover, the capability of the algorithms to predict results that can drive the correct use of each MSC population or even each MSC-derived exosome has improved, particularly with the combined use of AI. Section 6 exemplifies how the appropriate use of PI can drive the correct use of MSC populations for clinical purposes.

However, the recent emergence of ChatGPT (which is a type of AI) and the idealization of the use of AI brought by science fiction (sci-fi) could lead to a false interpretation of the power of these techniques, creating an over-expectation about the PI.

In this sense, PI may be understood as a powerful predictor tool to characterize molecular cells and exosome candidates for clinical purposes that can bring many advantages to the R&D of cell and cell-free therapies. Thus, PI serves as a tool to facilitate decision-making, allowing us to identify the best application or predict the MoA of a specific MSC population or MSC-derived exosome.

On the one hand, the PI brings novel opportunities, including job offers. On the other hand, it requires qualified professionals who have expertise in life science (particularly with expertise in cell and cell-free therapy) and analytics. This is because the predictive value of any result obtained using PI depends on many factors, such as (i) the data quality, (ii) data organization and availability, advanced knowledge in (iii) programming, (iv) statistics, (v) bioinformatics, (vi) molecular and cellular biology, and in (vii) the pathophysiology of the target disease. Despite this complexity, which requires a multidisciplinary effort to extract valuable information from data, PI can be considered a revolutionary tool for the biopharmaceutical sector, accelerating the time to obtain the registry of an investigational product and reducing costs with non-necessary experiments needed to bring the proof of concept (POC) required for Regulatory Agencies.