Molecular harvesting with electroporation for tissue profiling

Recent developments in personalized medicine are based on molecular measurement steps that guide personally adjusted medical decisions. A central approach to molecular profiling consists of measuring DNA, RNA, and/or proteins in tissue samples, most notably in and around tumors. This measurement yields molecular biomarkers that are potentially predictive of response and of tumor type. Current methods in cancer therapy mostly use tissue biopsy as the starting point of molecular profiling. Tissue biopsies involve a physical resection of a small tissue sample, leading to localized tissue injury, bleeding, inflammation and stress, as well as to an increased risk of metastasis. Here we developed a technology for harvesting biomolecules from tissues using electroporation. We show that tissue electroporation, achieved using a combination of high-voltage short pulses, 50 pulses 500 V cm−1, 30 µs, 1 Hz, with low-voltage long pulses 50 pulses 50 V cm−1, 10 ms, delivered at 1 Hz, allows for tissue-specific extraction of RNA and proteins. We specifically tested RNA and protein extraction from excised kidney and liver samples and from excised HepG2 tumors in mice. Further in vivo development of extraction methods based on electroporation can drive novel approaches to the molecular profiling of tumors and of tumor environment and to related diagnosis practices.

Personalized medicine is the optimization of care on an individual basis. Personalized medicine, based on molecular profiles of tumors and other tissues, has greatly developed over recent decades. In cancer therapy and care, a clear potential in several cases was demonstrated for the personalized approach as compared to traditional therapies [1][2][3] . Accurate diagnosis is a critical component of this approach. An important component of molecular diagnostics in diseased tissues, including tumors, is the profiling of DNA, RNA, proteins or metabolites, to identify molecular biomarkers that are predictive of tumor type [4][5][6][7] and of patient response 8,9 . To enable tumor profiling, current methods use liquid 10 or tissue biopsy. Tissue biopsy involves resection of a small tissue sample, a procedure which leads to localized tissue injury, bleeding inflammation, neural damage, fracture, and stress 11,12 , all increase the potential for tumor growth and metastasis [13][14][15] . The impact of this stress on tumor development is not well understood 15 . In addition, only a few biopsies can be performed at a time, limiting the scope of the spatial mapping of the sampled site, and leading to misdiagnosis if the tumor is missed or if a less relevant clonal sub-population is probed. Furthermore, some studies even concluded that due to tumor heterogeneity, information from a single biopsy is not sufficient for guiding treatment decisions in prostate cancer 16,17 .
Indeed, recent literature identified the limited support, by current technology, for characterizing tumor molecular heterogeneity 18 , as a major limitation of the personalized medicine approach in cancer 19 . Significant genomic evolution often occurs during cancer progression, creating variability within primary tumors as well as between the primary tumors and metastases 17,20,21 . Indeed, recent studies show that a positive result (both successful biopsy and a decisive detection of markers) appear to reliably indicate the presence of the high-risk disease 16 . However, a negative result does not reliably rule out the presence of high-risk clones. This is partly because a harvested tissue sample may not capture the most aggressive clone of a given tumor or tumor site 16 . Despite the significant improvement of molecular characterization technologies, during recent decades, thanks to the introduction of new high-resolution sequencing and bioinformatics methods, these technologies remain limited by tissue sampling methods 17,22 . Thus, tissue sampling remains a critical limitation to personalized medicine 16,17,23 . Therefore, new approaches to molecularly probing and characterizing several regions in the tumor, termed molecular cartography, are called for 24 .
To extend the state of the art of technologies that enable precision therapy, we developed a novel approach for tissue sampling with molecular biopsy using electroporation 25 . Electroporation-based technologies have been successfully used to non-thermally irreversibly and reversibly change permeabilization of the cell membrane Transcriptomics and proteomics differences detected with e-harvesting in HepG2 human tumor model compared to the normal liver in the mouse. We next tested our e-harvesting protocol by profiling HepG2 tumors grafted into murine livers (Fig. 3a). Histological examination clearly shows abnormal cells and tissue structures at the tumor area vs a normal liver structure (Fig. 3b). Tumors area showed an expected higher expression of Glypican 3 49 and Ki67 50 . For the comparison to normal mouse liver, we profiled 7 tumor samples and 14 normal mouse liver samples from 5 mice.
We found that in the e-extracts from the HepG2 liver model in mice, RNA encoding for PLK_1, S100P, TMED3, TMSB10, and KIF23 were expressed at significantly higher levels than RNA for these genes e-extracted from the normal liver (Figs 4 and S2).
Proteins were also extracted from the samples, following e-harvesting, and then profiled using unlabeled proteomics by LC/MS-MS (see Methods) (Table S8). For 2782 proteins from the HepG2 and the normal liver, identified using this approach, we performed gene ontology analysis with the associated genes (on the ranked list of differently expressed proteins, Table S9) using GOrilla 42 , annotating the ranked gene list to the mouse genome. We also analyze the molecular weight (MW) of the extracted proteins and observe a lognormal distribution (Table 1) as further described in Supplementary Information 1.
Analysis of gene ontology terms by processes showed that the processes of macromolecules metabolism, nucleic acid metabolism, regulation of cellular processes and macromolecule biosynthesis were higher in the HepG2 than in the normal liver (Fig. 5, Tables 5 and S10).
Analysis of the function terms shows multiple significant functional differences between the HepG2 as compared to the normal murine liver, including nucleic acid binding, protein binding, and oxygen binding, all expressed higher in the tumor (Tables 6 and S11).
Analysis by component showed large differences in the levels of proteins from cytosolic parts, from the protein-containing complexes and from the ribonucleoprotein complex extracted from HepG2 vs the normal liver (Tables 7 and S12).

Discussion
In this work, we tested molecular harvesting by electroporation from normal tissue samples as well as from tumor samples and assessed the molecular profiles of RNA and proteins obtained by this procedure. We showed that e-harvested RNA and proteins from HepG2 liver tumor in mouse liver, normal murine liver, and normal murine kidney are tissue-specific. These results suggest that e-extraction could be used for sample handling that can lead to informative molecular signatures.
Molecular extraction is a starting point in any molecular diagnostic assay 51 . The procedures include tissue disruption, cell lysis, sample pre-fractionation, and separation 52 . Although chemical, enzymatic and mechanical methods, including grinding, shearing, beating, and shocking to obtain tissue permeabilization for molecular extraction are well developed 53,54 , the extraction of molecules for analysis and for diagnosis at the point of care is still very challenging 55 . In addition, most of the current methods are very low-throughput, require individual sample manipulation and are not suitable for rapid extractions. The later is often required when the sample is sensitive and degrades rapidly 54,56 . To address these challenges, electric fields have been investigated in the recent decade as potentially supporting improved molecular extraction 54 . High-voltage, pulsed electric fields that lead to tissue electroporation is a specific example of these emerging technologies. Previous works demonstrated the extraction of genomic DNA 57 , RNA 58 , and proteins 59 from cells and tissues by electroporation. However, to the best of our knowledge, no previous work investigated molecular signatures, nor did any previous work perform differentiation expression analysis, as shown in this work. The e-extract of the kidney contained higher expression Tmem27, Umod and Slc34a1 and the extract from the liver contained higher expression Apoa5, F12, and Abcb11 ( Fig. 1), which aligns with literature reports 60 . This demonstration of differential expression measured in e-extracts was also corroborated with the study on the RNA extraction from the HepG2 tumor model in the mice liver. We found that in RNA e-extracted from the HepG2 liver model in mice RNA molecules encoding for PLK_1, S100P, TMED3, TMSB10, and KIF23 were significantly overexpressed as compared to RNA molecules of these genes e-extracted from the normal liver (Fig. 4). The higher expression of these genes in HepG2 was shown previously in the RNA-seq data from these tumors 61,62 .

Figure 2.
Annotation of the identified proteins to processes. The annotation applies to all identified proteins (Tables S1 and S2) and is performed by GOrilla 42 using a list ranked by (LFQ_liver-LFQ_Kidney) values.

GO term
Description P-value* FDR q-value** Enrichment (N Table 2. Gene ontology by a process of the differently expressed proteins in the liver and the kidney extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. ***Enrichment (N, B, n, b) is defined as follows: N -is the total number of genes. B -is the total number of genes associated with a specific GO term. n -is the number of genes in the top of the user's input list or in the target set when appropriate. b -is the number of genes in the intersection. Enrichment = (b/n)/(B/N). www.nature.com/scientificreports www.nature.com/scientificreports/ Taken together, our data suggest that electroporation extraction maintains some of the relative expression of RNA and allows for the faithful detection of differential expression between kidney, liver, and tumor in the liver.
However, the impact of the pulsed electric fields on RNA integrity, which is important for the downstream analysis 63 is still not known and is the subject of current invetigation 64,65 . Importantly, our previous work on the impact of electroporation type pulsed electric fields on the supercoiled DNA showed that electric fields cause the change in the size of the supercoiled DNA molecules, probably by nicking 29 . Future protocols development should focus on the extraction of RNA with high integrity, which could be used for sequencing.

GO term
Description P-value* FDR q-value** Enrichment (N, B, n, b)***  Table 3. Gene ontology by a function of the differently expressed proteins in the liver and the kidney extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. ***Enrichment (N, B, n, b) is defined as follows: N -is the total number of genes. B -is the total number of genes associated with a specific GO term. n -is the number of genes in the top of the user's input list or in the target set when appropriate. b -is the number of genes in the intersection. Enrichment = (b/n)/(B/N).  Table 4. Gene ontology by a component of the differently expressed proteins in the liver and the kidney extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. www.nature.com/scientificreports www.nature.com/scientificreports/ The proteomic analysis of the electroporation extracted samples showed that proteins extracted from tissues are aligned with literature 45 in terms of relative tissue abundance (Figs 2 and 5). Gene Ontology (GO) analysis performed on the ranked lists of the extracted proteins showed significant differences for process, function, and component terms of proteins extracted from the kidney, liver and HepG2 tumor model in the murine liver. Our previous studies on proteins extraction with electroporation from the chicken breast muscle also showed that using Gene Ontology analysis antioxidant function of the extracted proteins could be predicted 59 . The current work expanded that preliminary observation and shows that the extracted proteins are tissue-specific and potentially could allow differential expression measurements in various tissues including tumors. In the aforementioned previous work, we also showed that electroporation leads to selective extraction of proteins and ash from the seaweed 36,37 . Therefore, future studies of e-harvesting in tissue samples should determine the properties of the extractable proteins 37,38 . These properties depend on the tissue structure, using pulsed electric fields protocols and the extraction solvent 66 . This work does not address the physicochemical properties of the extracted proteins, which predicate their extractability and solubility 67 and additional studies in this direction are needed 68 . The combined knowledge of the physicochemical properties of the extractable proteins, the structure and chemical properties of the analyzed tissue could provide new ways for optimizing the pulsed electric field parameters such as electric field strength, pulse duration, pulse number, and frequency. Optimized parameters are critical for the further development of this method for molecular harvesting from various tissue types and for detailed molecular profiling.

GO term
Molecular harvesting with electroporation, as introduced in this work, is a new concept for tissue molecular profiling. The feasibility of permeabilization by electroporation is established either to deliver molecules to tissues and cells (drugs, vaccines, etc.) or to directly kill cells 25 . Temporary permeabilization of tissue to facilitate www.nature.com/scientificreports www.nature.com/scientificreports/ molecular harvesting has not been proposed until this work and to the best of our knowledge, devices that allow for the harvesting of molecules from tissues do not exist. In this study, we show that e-harvesting allows for the extraction of tissue-specific markers from excised tissue samples. This approach could potentially be used for molecular biopsy when the exact location of the tumor is not known. Future studies are needed to develop devices and methods for multiple probing of tissues and tumors in a single multipart procedure. Multiple probing will enable precise molecular cartography of physiological regions, which, in turn, could provide new essential information related to tumor heterogeneity, paving new roads for personalized medicine.  www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Animals. All animal procedure was approved by the Israel National Council for Animal Experimentation (Israel Ministry of Health number IL-18-3-92). Eight, 8-week old female Athymic Nude mice weighting ~20 g were provided by the Science in Action Ltd. CRO. Five animals were used for the HepG2 model and 3 additional animals were used for the differential expression comparison between normal tissues. The animals were housed in cages with access to food and water ad libitum and were maintained on a 12 h light/dark cycle in a room temperature of around 21 °C and a relative humidity range of 30 to 70%. All in vivo experiments were conducted by a professional veterinary in accordance with the Israel National Council for Animal Experimentation guidelines and regulations.
In vivo human HepG2 liver tumor model. Human HepG2 cell line was purchased from ATCC. The cells were grown on the base medium for this cell line (ATCC-formulated Eagle's Minimum Essential Medium, Catalog No.  supplemented with a fetal bovine serum to a final concentration of 10%. After 5 passages, approximately 2 weeks of cultivation, 10 6 HepG2 cells (50 µL) were directly injected into the mice liver lobe during surgery, similar to 69 . In brief, the animals were anesthetized with Ketamine/Xylazine. The abdomen was shaved and the skin was cleaned with ethanol (70%). The small incision was made on the skin up to the liver. One lobe was exposed and the cells were injected with 0.5 mL syringe and 29G needle. After the injection the skin was sutured with 0/5 thread. Four to five weeks after the injection of the cells, the mice were euthanized with CO 2 and the tissues were immediately harvested for extraction with pulsed electric fields. The tumor was induced in 5 animals.
Histology. Specimens were harvested immediately after the treatment and fixed in 10% formalin. Samples in plastic cassettes were dehydrated through ascending ethanol concentrations, transferred into xylene and then paraffinized into paraffin, by an automated machine. Next, the samples were manually embedded into paraffin blocks. The paraffin blocks were sectioned at approximately 3-5 microns thickness. Sections were put on glass slides. Slides were stained with Hematoxylin & Eosin (H&E) and covered by an automated machine. Pulsed electric field application for biomolecules extraction ex vivo. First, 250-300 mg of tissue was excised and loaded into electroporation cuvette (BTX electroporation cuvettes plus, 2 mm, Model No. 620,

GO term
Description P-value* FDR q-value** Enrichment (N, B, n, b) Table 5. Gene ontology by a process of the differently expressed proteins in the HepG2 the normal liver extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. www.nature.com/scientificreports www.nature.com/scientificreports/ Harvard Apparatus, MA). The cuvette was inserted into a custom-made electroporation cuvette holder and connected to the electric field pulse generator, (BTX830, Harvard Apparatus, MA). Electroporation was performed using a combination of high-voltage short pulses with low-voltage long pulses, which was shown to increase the efficiency and area of permeabilized tissues [70][71][72] , as follows: 50 pulses 500 V cm −1 , 30 µs, 1 Hz, and 50 pulses 50V cm −1 , 10 ms, delivered at 1 Hz. After the PEF treatment, 300 µl Nuclease-free water was added to the cuvette for "juice" dilution and then liquids transferred to 1.5 ml tubes.
RNA isolation and amplification from the pulsed electric field extracted juice. Normal mouse liver and normal mouse kidney. All samples were collected in fresh conditions and transferred on ice from the surgery room to the laboratory. The extract from 3 mice tissues was separated to 27 normal liver and 18 normal kidney samples after the RNA extraction.
The total RNA was extracted using Water-saturated phenol and 1-Bromo-3-chloropropane (Biological Industries, Beit Haemek Ltd). The cDNA used for PCR was synthesized from total RNA using the GoScript ™ Reverse Transcription System (Promega Corporation, Madison, WI, USA). To select genes for PCR profiling we downloaded mouse liver and kidney RNA-seq data from Newman et al. (2017. PMID:28877458) (GEO ID: GSE101657) 60 , which covers five mice per tissue type. Normalization and differential expression (DE) analysis were done using DESeq2 73 . We considered a gene to be DE if it has a corrected p-value < 0.01 and log2(fold-change) > |1| and if it also has average read coverage >100 normalized reads. The selected DE genes: Slc34a1, Umod, Tmem27, Apoa5, F12, and Abcb11 were also manually checked to see if their human orthologs are also liver/kidney-specific according to the human protein atlas (https://www.proteinatlas. org/). To measure expression differences between normal tissues (kidney vs liver) we used PCR with 6 pairs of specific primers for the selected genes. The primers were designed according to the mouse transcriptome (Table S13).
The PCR amplification protocol was 95 °C for 30 s, 40 cycles of 95 °C for 5 s, 55 °C for 10 s, and 72 °C for 30 s.
HepG2 cells compared to normal mouse liver. To select genes for comparative PCR measurements in HepG2 vs normal murine liver we downloaded RNA-seq of HepG2 cellular carcinoma and matched normal samples from TCGA (TCGA LIHC, PMID: 28622513) 74 . Normalization and DE analysis were done using DESeq2. We considered a gene as DE, if it has corrected p-value < 0.01 and log2(fold-change) > |1|. The list of genes considered as up-regulated in cancer (the genes with log2(fold-change) > 1) was further filtered to include only genes for which in both HepG2 RNA-seq data from Solomon et al. (2017, PMID:29129909) 61 and HepG2 RNA-seq data from the ENCODE project 62 , an expression level higher than 75% of the expressed genes is observed (reads per kilobase million > 10 (RPKM > 10) in both studies). Using the human protein atlas, we manually checked that the selected genes are considered as elevated in cancer but underexpressed in normal liver. The selected genes were PLK1, TMED3, TMSB10, S100P, and KIF23. To measure expression differences between tumor and normal liver tissue we used PCR with 5 pairs of specific primers for the selected genes. The primers were designed according to the human transcriptome (Table S14).
GO term Description P-value* FDR q-value** Enrichment (N, B, n, b) Table 6. Gene ontology by function of the differently expressed proteins in the HepG2 the normal liver extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. ***Enrichment (N, B, n, b) is defined as follows: N -is the total number of genes. B -is the total number of genes associated with a specific GO term. n -is the number of genes in the top of the user's input list or in the target set when appropriate. b -is the number of genes in the intersection. Enrichment = (b/n)/(B/N).  Table S14. Primers used for the mouse liver and human HepG2 tumor in mouse liver differentiation by the RNA extracted with electroporation.
All samples were collected in fresh conditions and transferred on ice from the surgery room to the laboratory. The extract from 5 mice tissues was separated into 7 tumors and 14 normal kidney samples after the RNA extraction. The RNA was separated using 1.2% E-Gel electrophoreses system (ThemoFisher, CA). The gel images were captured with ENDURO TM GDS camera (Labnet Inc., NJ). Quantification was done with ImageJ (ver 1.52e, NIH, MA).
Isolating proteins from the pulsed electric field extracted juice. Proteins were isolated from the PEF extract using the protocol of EZ-RNA II kit (Biological Industries, Beit Haemek Ltd). Air-dried protein pellets were taken for proteomic analysis as described below.
identifying and quantifying proteins using Lc-MS/MS. Proteolysis. The samples were brought to 8 M urea, 400 mM ammonium bicarbonate, 10 mM DTT, vortexed, sonicated for 5′ at 90% with 10-10 cycles, and centrifuged. Protein amount was estimated using Bradford readings. 20 ug protein from each sample was reduced 60 °C for 30 min, modified with 37.5 mM iodoacetamide in 400 mM ammonium bicarbonate (in the dark, room temperature for 30 min) and digested in 2 M Urea, 100 mM ammonium bicarbonate with modified trypsin (Promega) at a 1:50 enzyme-to-substrate ratio, overnight at 37 °C. Additional second digestion with trypsin was done for 4 hours at 37 °C.
Mass spectrometry analysis. The tryptic peptides were desalted using C18 tips (Harvard) dried and re-suspended in 0.1% Formic acid. The peptides were resolved by reverse-phase chromatography on 0.075 × 180-mm fused silica capillaries (J&W) packed with Reprosil reversed-phase material (Dr. Maisch GmbH, Germany). The peptides were eluted with linear 180 minutes gradient of 5 to 28% 15 minutes gradient of 28 to 95% and 25 minutes at 95% acetonitrile with 0.1% formic acid in water at flow rates of 0.15 μl/min. Mass spectrometry was performed by Q-Exactive Plus mass spectrometer (Thermo) in a positive mode using repetitively full MS scan followed by collision induces dissociation (HCD) of the 10 most dominant ions selected from the first MS scan.
The mass spectrometry data from all the biological repeats were analyzed using the MaxQuant software 1.5.2.8 (Mathias Mann's group) vs. the mouse proteome from the UniProt database with 1% FDR. The data were quantified by label-free analysis using the same software, based on extracted ion currents (XICs) of peptides enabling quantitation from each LC/MS run for each peptide identified in any of the experiments. Statistical analysis. We statistically analyzed the functional groups of the extracted proteins using Gene Ontology (GO) analysis with GOrilla, annotating the ranked gene list to the mouse genome. GOrilla is based on the mHG statistics 42,48,75 .
Comparison of ranked lists: m 2 HG. To compare two ranked lists we compute the relative permutation of one against the other and then assess the significance of the size of the intersection of two prefixes of the lists by using GO term Description P-value* FDR q-value** Enrichment (N, B, n, b) Table 7. Gene ontology by component of the differently expressed proteins in the HepG2 the normal liver extracted with electroporation mapped with GOrilla 42 . *'P-value' is the enrichment p-value computed according to the mHG or HG model. This p-value is not corrected for multiple testing of 731 GO terms. **'FDR q-value' is the correction of the above p-value for multiple testing using the Benjamini and Hochberg (1995) method. Namely, for the i th term (ranked according to p-value) the FDR q-value is (p-value * a number of GO terms)/i. ***Enrichment (N, B, n, b) is defined as follows: N -is the total number of genes. B -is the total number of genes associated with a specific GO term. n -is the number of genes in the top of the user's input list or in the target set when appropriate. b -is the number of genes in the intersection. Enrichment = (b/n)/(B/N).