Simultaneous detection of multiple mRNAs and proteins in bovine IVD cells and tissue with single cell resolution

Abstract

Objectives.

Interactions of cells with their neighbors and influences by the surrounding extracellular matrix (ECM) is reflected in a cells transcriptome and proteome. In tissues comprised of heterogeneous cell populations or cells depending on ECM signalling cues such as those of the intervertebral disc (IVD), this information is obscured or lost when cells are pooled for the commonly used transcript analysis by quantitative PCR or RNA sequencing. Instead, these cells require means to analyse RNA transcript and protein distribution at a single cell or subcellular level to identify different cell types and functions, without removing them from their surrounding signalling cues.

Results.

We developed a simple, sequential protocol combining RNA is situ hybridisation (RISH) and immunohistochemistry (IHC) for the simultaneous analysis of multiple transcripts alongside proteins. This allows one to characterize heterogeneous cell populations at the single cell level in the natural cell environment and signalling context, both in vivo and in vitro. This protocol is demonstrated on cells of the bovine IVD, for transcripts and proteins involved in mechanotransduction, stemness and cell proliferation.

Conclusions.

A simple, sequential protocol combining RISH and IHC is presented that allows for simultaneous information on RNA transcripts and proteins to characterize cells within a heterogeneous cell population and complex signalling environments such as those of the IVD.

Introduction

The intracellular generation of a messenger RNA transcript precedes its translation into a protein. Many diagnostic and research procedures build on detecting RNA transcripts through reverse transcription and polymerase chain reaction (RT-PCR), often in the quantitative real-time manner and primarily on material from pooled cells (Kolodziejczyk et al. 2015; Minogue et al. 2010a; Minogue et al. 2010b; van den Akker et al. 2017; Wang et al. 2020). Cell pooling is a common practice to achieve the critical amount of material for molecular investigation, yet comes at the price of potentially masking cellular dissimilarity within a chosen cell population. Combing methods for single-cell (sc) isolation and culture with advances in scRNA sequencing have pushed the field of sc transcriptomics in recent years. Due to the complex nature of the proteome similarly powerful sc proteomics approaches are still undergoing development (Hedlund and Deng 2018). In heterogeneous cell populations, single cell resolution or knowledge of spatial transcript or protein distribution is essential. However, removing cells from their environmental signalling cues to achieve single cell analysis can similarly bias results (Kraus et al. 2019a) to a point where results might no longer be biologically meaningful (Marx 2019). Transcript detection through RNA in situ hybridization (RISH) provides single cell resolution without disrupting the cellular context. RISH is also popular in areas of research where immunohistochemical (IHC) analysis of proteins is not feasible; either because of a lack of specific antibodies or, as seen for noncoding RNAs, proteins are never made (Kraus et al. 2013; Kraus et al. 2019b). The early history of nucleic acid detection by in situ hybridization is well reviewed by Levsky and Singer (Levsky and Singer 2003). Initially radioactive nucleotides were incorporated to label oligonucleotide probes detecting complementary DNA or RNA molecules (Gall and Pardue 1969). A range of synthetic fluorophores has been developed over the years (Martynov et al. 2016) and labelling technology advanced via fluorochrome labelled RNAs (Bauman et al. 1980) to end-labelled oligonucleotides, permitting the detection of multiple transcripts (Levsky et al. 2002). At the same time antibodies were developed specific for non-radioactive uracil-tri-phosphate (UTP) labelling with digoxygenin (DIG), biotin or fluorescein (FITC) epitopes (Chen et al. 2009; George et al. 2018), themselves recognized by secondary antibodies linked to enzymes for chromogenic detection methods or conjugated with different coloured fluorophores (Chatterjee et al. 2014; Holtke and Kessler 1990; Kraus et al. 2018; Panchuk-Voloshina et al. 1999; Schaeren-Wiemers and Gerfin-Moser 1993; Sivakamasundari et al. 2017; Tsukamoto et al. 1991). Trending nowadays is the use of single labelled oligonucleotide probes of 15-25 nucleotides (nts) in length to avoid repetitive elements in the target sequence. Reduced probe length however comes at the expense of reduced signal intensity and typically requires the use of multiple different oligonucleotides for transcript tiling (George et al. 2018). Instead, we present a PCR-tailored compromise, that allows for longer probes, with sufficient labelling intensity, also avoiding repetitive elements. A simple modification of the “classic” RISH protocols employing fluorophore conjugated antibodies allows for the simultaneous detection of different mRNAs in combination with protein IHC. Here, we introduce a detailed, flexible and economic tool set that can be tailored to any gene of interest. While this was established on cells and tissue of the bovine intervertebral disc (IVD), a focus of our research, this method can be easily expanded to human cells where the flexibility for commercially available primary and secondary antibodies is far superior to bovine samples.

Degeneration of the IVD through aging or trauma is a common reason for severe and chronic lower back pain (Cheung et al. 2009; Luoma et al. 2000; Waterman et al. 2012). Identifying the spatio-temporal distribution of RNA transcripts and proteins within a tissue or a cell can provide important cues for normal and abnormal developmental processes and cellular responses during injury and recovery. The IVD is the largest avascular organ in the vertebrate body (Moore 2006; Urban et al. 1977). The mature IVD is structured in an outer annulus fibrosus (AF) at the periphery that transitions into the central nucleus pulposus (NP) (Pooni et al. 1986). Cells of the NP are spherical and embedded in vast amounts of less densely structured, aggrecan and collagen II rich extra cellular matrix (ECM). AF cells however, are located in a densely packed, lamellar organized, collagen I rich ECM (Errington et al. 1998; Eyre and Muir 1976; Kraus et al. 2019a; Lama et al. 2018; Li et al. 2019a). The heterogeneous nature of IVD cell populations requires single cell assessment in vivo and in vitro (Kraus et al. 2019a; Li et al. 2019a; Li et al. 2019b). Identifying NP progenitor cells could be useful in restoring degenerative IVDs through cell based regenerative medicine (Kraus and Lufkin 2017; Risbud et al. 2007). Cell identification can be achieved with single cell resolution through methods such as RISH and IHC as demonstrated here for cell markers related to mechanotransduction, sternness and cell proliferation. LMNA encodes for Lamin A and C as the two major protein isoforms identified as alternative splice products of the human LMNA gene. Lamins are associated with the nuclear lamina and belong to the type V group of intermediate filaments (Dittmer and Misteli 2011; Ho and Lammerding 2012; McKeon et al. 1986). Yes-associated protein (YAP) and tafazzin (TAZ) encode for transcriptional coactivators of the Hippo signalling pathway (Hong and Guan 2012). We further employed the cell proliferation marker Ki67, a non-histone protein present during the active cell cycle (Gerdes et al. 1991; Gerdes et al. 1983) and the cell cycle marker cyclin-dependent kinase 4 (CDK4), a member of the serine/threonine protein kinase family active during the G1/S phase of the cell cycle (Sheppard and McArthur 2013). The bovine coccygeal IVD is a generally accepted model system to study the IVD of a young healthy human adult (Beckstein et al. 2008; Demers et al. 2004; Oshima et al. 1993; Rodrigues-Pinto et al. 2013), however antibodies for bovine samples are not as readily available as they are for murine or human tissue or have not been tested for cross reactivity. Therefore we developed RNA transcript based alternatives as described in detail in the protocol presented here.

Materials and methods

All procedures were performed in compliance with the ethical standards of Clarkson University. No human material was included in this study. All blocking and antibody dilutions were made in undiluted SuperBlock™ (PBS) Blocking Buffer (Thermo Fisher Scientific). Antibodies and their dilutions are listed in Table 1. All steps including blocking or antibody incubation were carried out at 4 °C.

Table 1.

Antibodies used in this study

Name	Dilution	Company	Product number
IgG Fraction Monoclonal Mouse Anti-Digoxin 1.2 mg/ml	1:100	Jackson Immuno Research	AB_2339005
IgG Fraction Monoclonal Mouse Anti-Fluorescein (Chen et al.) 1.2 mg/ml	1:100	Jackson Immuno Research	AB_2313645
Ki-67 Polyclonal Antibody 0.8 mg/ml	1:100	Thermo Fisher Scientific	PA5-19462
CDK4 Polyclonal Antibody 1.0 mg/ml	1:100	Thermo Fisher Scientific	PA5-80462
Alexa Fluor® 488 AffiniPure Goat Anti-Mouse IgG (H+L) 1.5 mg/ml	1:1000	Jackson Immuno Research	AB_2338840
Alexa Fluor™ 594 goat anti-mouse IgG(H+L) 2.0 mg/ml	1:1000	Thermo Fisher Scientific	A11032
Alexa Fluor™ 568 goat anti-rabbit IgG(H+L) 2.0 mg/ml	1:1000	Thermo Fisher Scientific	A11011
Alexa Fluor™ 647 goat anti-rabbit IgG(H+L) 2.0 mg/ml	1:1000	Thermo Fisher Scientific	A21244

PCR template generation for RISH probes

All templates for the RISH RNA probes were amplified by polymerase chain reaction (PCR) from genomic DNA as described (Kraus et al. 2017). The transcript target sequence was optimized by omitting repetitive elements with the help of ISB Repeat Masker (http://www.repeatmasker.org). High specificity for primers was ensured using the primer design software NCBI Primer (https://www.ncbi.nlm.nih.gov/nucleotide). All target sequences and primers used for the generation of the probe templates are listed in Table 2. Enzymatic activity of T7-RNA-polymerase facilitated labelling of the RNA antisense probes with DIG (Kraus et al. 2017; Li et al. 2019b) or FITC epitopes according the manufactures protocol: Essentially, 200 ng PCR product were labelled for 4 h at 37 °C with 1x transcription buffer (Roche or Promega), 40 U of T7-RNA-polymerase (Roche or Promega), 40 U of RNAse inhibitor (Roche) and a 1x concentration of the respective labelling mix (Roche #11277073910 (DIG) or #11685619910 (FITC)) in a final volume of 20 μL, purified through EtOH precipitation, resuspended in 20 μL RNAse-free water and stored at −20 °C. Prior to their use, 1 μL was tested via gel electrophoresis and used at −500-800 ng/mL in the hybridization.

Table 2.

Primers used for the generation of bovine RISH probes used in this study

Name	Primer	PCR product	Sequence ID
LMNA exon 2	5’-CTATTAGAGCCTTTGCCCAGGA-3’	585 bp	XM_005203621.1
	*5’-CCTTGAACTCCTCTCGCACT-3’
LMNA exon 13	5’-GCATCATGTAACCTGGGACCT-3’	858 bp	XM_005203621.1
	*5’-GCAAGGGGCTCCTTAGTGTT-3’
YAP	5’-GCCGCCACCAAGCTAGATAA-3’	1160 bp	XM_024975708.1
	*5’-ACACTACCCCAACCGGATTT-3’
TAZ	5’-TCCAGCTGCCTTTTGGACTT-3’	518 bp	XM_002699714.5
	*5’-TTGGACTGGGCTCCCCTTAT-3’
Ki67	5’-CGAGCCTCAGAGCTGAAGTG-3’	915 bp	XM_015469655.1
	*5’-GACTGGCTCCGGTTGAGAAG-3’

^*The following sequence was added to the 5’end of this primer to allow for T7-RNA polymerase driven in vitro transcription to generate an antisense probe: 5’CCAAGCTTGTAATACGACTCACTATAGGGC-3’

IVD tissue and cell lines

Fresh adult bovine tails obtained from local abattoirs served as source for coccygeal IVDs. The isolated IVDs were fixed in 4% (w/v) paraformaldehyde (PFA) for 24 h, dehydrated through a gradient of 30% (v/v), 70% (v/v), 90% (v/v) and 2x 100% (v/v) Ethanol (EtOH) baths, followed by one bath in equal volumes of Histochoice (VWR) and 100% (v/v) EtOH and finally three baths with Histochoice prior to their embedding in paraffin (Richard-Allen Scientific). Sections of 7 μm were cut on a rotary microtome (KEDEE) and mounted on VistaVision™Histobond glass slides (VWR). On the morning of the experiment, the tissue sections were dewaxed with Histochoice (VWR) and rehydrated by reversing above gradients as previously described (Kraus et al. 2015) followed by a fixation step for a least 20 min with 4% (w/v) PFA at room temperature (RT).

Primary cell lines were isolated from the nucleus pulposus (NP) of bovine IVDs and cultured essentially as described (Kraus et al. 2017). Cells were trypsinized with 0.05% Trypsin/EDTA (Gibco) for a maximum of 3 min and plated at low density on 12 mm round, glass coverslips (NeuVitro), which were previously sterilized with 70% EtOH and coated with 0.1% (w/v) gelatine (Sigma) diluted from sterile filtered, non-autoclaved 1% stock solution. The seeded coverslips were maintained in a 24-well culture plate (Greiner) for 12 h to several days prior to fixation with 4% (w/v) PFA for at least 20 min or until the morning of the experiment. Storage occurred at 4 °C, protected from dehydration.

Single and double fluorescent RNA in situ hybridization (RISH)

The simplified sequential RISH procedure for one or two RNA targets including an approximate timeline is outlined in Fig.1a and described in more detail below. All procedures were carried out in a RNAse-free environment. The following hybridization method can be applied to paraffin tissue sections after rehydration or cells grown on glass coverslips. All washing steps were carried out at RT for 5 min each unless stated otherwise. The 4% (w/v) PFA fixed samples were washed three times with 1x phosphate buffered saline (PBS) followed by 2 h of prehybridization in hybridization buffer and overnight (o/n) hybridization including the labelled RNA probes diluted in fresh hybridization buffer. Both steps were carried out at 62 °C in a hybridization oven. The prehybridization solution consisted of 50% formamide (Amresco), 5x SSC (comprised of 0.75 M sodium chloride and 0.075 M sodium citrate dehydrate), 1x Denhardt’s (comprised of 0.02% (w/v) ficoll, polyvinylpyrrolidone and bovine serum albumin each), 0.1% (v/v) Tween 20 (Sigma), 0.1 mg/ml tRNA (Roche) and 0.05 mg/ml Heparin (Alfa Aesar). To avoid dehydration of the tissue sections or cells, a humid chamber was generated with equal volumes of 5x SSC and 100% formamide. The protocol for a single template RISH was previously described in detail (Kraus et al. 2015). For double template RISH, two probes corresponding to different transcripts were exclusively labelled with either DIG or FITC and added simultaneously to the sample for hybridization. Stringent washes as described in (Kraus et al. 2015; Kraus et al. 2017) removed all unbound probe: Three changes of RISH wash buffer A for 20 min each at 62 °C consisting of 50% (v/v) formamide, 5x SSC and 1% (v/v) sodium dodecyl sulphate (SDS, VWR), were followed by three changes of RISH wash buffer B (0.5M NaCl, 0.01M Tris/HCl pH7.5 and 0.1% (v/v) Tween 20) and one incubation in equal volumes of RISH wash buffers B and C, prior to three changes of RISH wash buffer C at 58 °C for 20 min each, consisting of 50% (v/v) formamide, 2x SSC and 0.2% (v/v) SDS. This was followed by three washes with 1x PBS and a 2 h blocking step. RNA probes labelled with DIG were first detected by a mouse anti-digoxin monoclonal antibody for a minimum of 3 h followed by three washes with 1x PBS to remove any unbound primary antibodies. A 2 h blocking step and the addition of a goat anti-mouse Alexa 488 secondary antibody followed for a minimum of 3 h in the dark. All further steps were carried out omitting light as much as possible. If two RNA probes were analysed simultaneously, another three rounds of 1x PBS washes followed by 2 h of blocking preceded the addition of the mouse anti-FITC monoclonal antibody for a minimum of 3 h. After three washes with 1x PBS and another 2 h blocking step, a goat anti-mouse Alexa 594 secondary antibody was added for a minimum of 3 h (Fig. 1a, day 3). After three washes with 1x PBS and one wash with deionized water the nuclei were stained with 2 μg/mL 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific) for 15 min at RT followed by three more washes with sterile deionized water. Slides were then mounted with Shandon Immu-Mount™ (Thermo Fisher Scientific), sealed and imaged with a DMi8 confocal microscope (Leica). Expression intensities could further be quantified using IMAGE J as shown in (Kraus et al. 2019a; Li et al. 2019a; Li et al. 2019b). If desired, a primary antibody raised in rabbit specific for a target protein could be added together with the last RISH related secondary antibody, which would add another round of three 1x PBS washes, 2 h of blocking and 3 h incubation with a goat anti rabbit Alexa 647 secondary antibody prior to the DAPI staining as illustrated in Fig. 1a,c.

Fig. 1. Illustration of RISH/IHC procedures.

Simplified sequential procedure for single or double RISH with IHC (a) or RISH with double IHC (b) alongside an approximate timeline for the individual modules of the procedure. A schematic of a cell illustrates the different components and their targets after double RISH and IHC (c) or RISH and double IHC (d). 488,568,594 and 647 indicate the different Alexa-fluorophore conjugated antibodies; t1/t2: different RNA transcripts; px/py: different target proteins; mad: mouse anti-DIG antibody; maf: mouse anti-FITC antibody; gam: goat anti-mouse antibody; rax: rabbit-anti protein x antibody; ray: rabbit anti-protein y antibody: gar: goat ant-rabbit antibody; W/B: wash and blocking; W: wash.

Combination of RISH and IHC

A simplified sequential procedure for the combination of RISH and IHC including an approximate timeline is outlined in Fig. 1b. Cells or tissue sections were fixed and processed in a RNAse-free environment as described above. Samples were first subjected to RISH as described above until the last RISH related blocking step for the final RISH 2^nd antibody was completed (Fig. 1a–b, green bar). Typically, the RISH procedure has unmasked the antigen of the target protein sufficiently for epitope recognition by a primary antibody. A specific primary antibody for the target protein raised in rabbit (or any species other than mouse not cross-reacting with the sample tissue) was then added together with the last secondary antibody of the RISH procedure for a minimum of 3 h as described above. After three washes with 1x PBS and another 2 h blocking step a goat anti-rabbit Alexa 647 antibody was added for a minimum of 3 h followed by DAPI staining. Slides were mounted and subjected to imaging with a DMi8 confocal microscope (Leica) as described above. To detect multiple proteins alongside a RNA transcript another round of three 1x PBS washes and 2 h blocking was included as indicated in Fig. 1b, followed by the addition of a rabbit raised primary antibody against the second target protein for 3 h, three washes with 1x PBS and a 2 h blocking step prior to adding the final secondary goat anti-rabbit Alexa 568 antibody and proceeding with the DAPI staining. The outcome is illustrated in Fig. 1b,d.

Results and Discussion

We describe a sequential protocol for the simultaneous analysis of two RNA transcripts alongside proteins with subcellular single cell resolution. This method, while established in the bovine model, is easily adaptable to human tissue and cells. The flexibility of PCR based RISH probe design allows for the detection of non-coding RNAs (Kraus et al. 2013; Kraus et al. 2019b), a field becoming increasingly more important in our understanding of the complex gene regulation events in context of cancer, aging and tissue degeneration. To demonstrate the usefulness of the protocol, we describe the application thereof in the context of genes and cells relevant to the IVD.

Double RISH procedure for markers related to mechanotransduction and stemness of IVD cells

We have previously used single RISH to identify IVD biomarkers (Li et al. 2019a; Li et al. 2019b). Here we show the simultaneous analysis of two different RNA transcripts identified by two different fluorophores reflecting the DIG or FITC labelled RNA probes. As an example, we investigated the colocalization of RNA transcripts encoding for the bovine YAP and TAZ proteins. Both proteins are known to bind to other transcription factors, acting as main regulators of downstream target genes such as cyclins and growths factors by turning gene expression on, if Hippo signalling is in the off-mode (Boopathy and Hong 2019; Chatterjee et al. 2010; Johnson and Halder 2014). The subcellular localization of the YAP and TAZ proteins has been studied extensively as the evolutionary conserved Hippo signalling pathway plays an important role in modulating organ size and cancer by controlling stem cell maintenance and proliferation in residual stem cell populations (Johnson and Halder 2014; Pan 2010; Ramos and Camargo 2012; Sng and Lufkin 2012). Similar emphasis was given to transcriptome analysis of YAP/TAZ downstream targets (Kim et al. 2015). The identification of these targets along with loss-of-function studies carried out in mouse or HEK293 cells suggested redundant yet also partially separate functions of YAP and TAZ (Pan 2010; Plouffe et al. 2018). A cytoskeleton mediated control of YAP/TAZ localization was described in Drosophila and Hela-cells (Sansores-Garcia et al. 2011). YAP was also tied to cellular mechanosensory mechanisms apparently independent of the Hippo signalling cascade that regulate nuclear distribution of the YAP protein, identifying YAP as a mechanotransduction protein (Koushki et al. 2020). The authors further describe differences in lamin A and YAP association with the nuclear lamina or nucleoplasm in response to cell tension. Cells under high tension showed lamin A and YAP proteins evenly distributed throughout the nucleoplasm, while in a more relaxed cell with a less deformed nucleus the concentration of nuclear YAP was reduced and lamin A found in the nuclear periphery. Interestingly, the A and C isoform of lamin are also known as mechanotransducive components (Gonzalez-Cruz et al. 2018), with transcription increasing with nuclear stiffness: LmnA transcripts are found more abundant in cells of cartilage and bone than neurons or fat (Swift et al. 2013). Members of the A-type nuclear lamins have been discussed as tumor biomarkers and are considered stem cell differentiation markers as they are typically associated with differentiated cells (Constantinescu et al. 2006; Ho and Lammerding 2012). Mutations in LMNA in human are further associated with the premature aging disease progeria (Dittmer and Misteli 2011; Taimen et al. 2009). Lamin A-type proteins are initially present in the zygote (Stewart and Burke 1987), possibly as a maternal factor. They are not produced in the preimplantation mouse embryo and disappear until the mid-gestation stage when organogenesis is well on the way (Rober et al. 1989; Stewart and Burke 1987). Given the role of nuclear lamins, most work is focused on a mechanistic understanding by employing protein immunostaining. However most commercially available antibodies do not distinguish between the lamin A and C or other isoforms (Gonzalez-Cruz et al. 2018) or are not suitable for non-standard model systems like bovine cells and tissue. RISH provides high specificity through nucleic acid hybridization that can be tailored to alternative splice products as seen above for the detection of two bovine LMNA transcripts (Table 2 and Fig. 2). Here we employed these two different bovine LMNA RNA probes in RISH (Fig. 2). One probe is directed against exon2, identifying the shorter LMNA transcript NM_001034053.1 encoding the lamin C isoform that ends in ~VSGARR and the longer transcript variant XM_005203621.1 encoding the pro-peptide isoform that ends in ~PQNCSIM. The other probe is directed against bovine LMNA exon 13 which detects only the longer isoform (see Table 2). LMNA transcripts remain expressed in adult tissue like the adult bovine IVD (Fig. 2a,c–e,g,h). The connection of YAP/TAZ and lamin A to mechanotransduction processes, their proposed impact on stemness and cell proliferation and the implication of LMNA mutations in degenerative diseases is of particular interest for cell types of the mature IVD. Based on the elongated cell and nuclei shape of AF cells (Fig. 2a–d), one would expect increased mechanical tension for AF cells over NP cells. The elongated AF cells are located in densely packed, lamellar organized ECM. Spherical NP cells lie embedded in vast amounts of less densely structured ECM, without obvious deformities of the nucleus (Fig. 2e–h) and (Errington et al. 1998; Kraus et al. 2019a; Lama et al. 2018; Li et al. 2019a). The in vivo subcellular localization of the RNA transcripts encoding for YAP, TAZ (Fig. 2b–d,f–h) and LMNA (Fig. 2a,c–e,g,h) in AF and NP cells of the mature bovine IVD revealed that all four transcripts were detected in both cell types. Differences in sequence length and uracil /epitope content of the RNA probes would not allow for direct quantitative comparison between different transcripts. Despite the stretched nucleus of AF cells (Fig. 2a–d), LMNA transcripts appeared more prominent in NP cells (Fig. 2e–h) compared to AF cells (Fig. 2a–d). Interestingly, in cells that showed abundant transcripts for LMNA, YAP or TAZ transcripts were also more prominent (Fig. 2g,h). Yet different LMNA, YAP or TAZ transcripts abundance might also be attributed to a difference in a cells stemness potential. It is widely accepted that cell populations in the mature IVD are heterogeneous (Chelberg et al. 1995; Kraus et al. 2017; Li et al. 2019a; Li et al. 2019b) like seen for the TAZ and LMNA exon13 probes in NP cells (Fig. 2h), where one cell shows the presence on LMNA transcripts and could therefore be a differentiated cell, while the other NP cell (arrowhead in Fig. 2h) is not showing LMNA expression and therefore potentially of progenitor cell potential. However type-A nuclear lamins, TAZ and YAP predominantly execute their function at the protein level, especially through their subcellular distribution, which is not addressed here.

Fig. 2. Example of the double RISH procedure for transcripts of the mechanotransduction components YAP, TAZ and LMNA on sections through the coccygeal bovine IVD.

(a-d) shows cells of the AF. (e-h) shows cells of the NP. Alexa-488 conveyed fluorescence reflects a RNA probe targeting exon2 of LMNA in (a,c,e and g) or the TAZ transcript in (b,d,f and h). Alexa-594 conveyed fluorescence indicates the presence of a RNA probe targeting exon 13 of LMNA in (a,d,e and h) or the YAP transcript in (b,c,f and g). The white arrowhead in (h) points to a potential progenitor cell as indicated by the absence LMNA or TAZ transcripts. DAPI stains the nucleus

Identifying proteins alongside their RNA transcripts

IHC has a long history with applications in medical diagnostics and research. However, IHC is rather costly, can require harsh antigen unmasking procedures and protocols are often antibody-specific rather than universally applicable (Matos et al. 2010). Owing to the high cost of antibody production, antibodies directed against novel proteins or those of non-common target organism are often not commercially available. Cross-reactivity of antibodies to unknown reactive epitopes can further complicate procedures (Lufkin 1996). RISH with the high specificity based on nucleic acid hybridization can be used as an alternative to detect RNA transcripts instead. However, transcript stability and protein turnover can differ and so can their subcellular compartments. Here we apply our protocol to visualize key players during cell cycle progression namely Ki67 and CDK4 (Fig. 3). We demonstrate that RISH and IHC can be combined to identify both transcript and protein within the same cell. The use of Ki67 as a marker for cell proliferation is a widely practiced concept (Sun and Kaufman 2018). Generally, the cell proliferation marker Ki67 is assessed by IHC, but Ki67 mRNA levels are known to increase during the G1 phase of the cell cycle (Sobecki et al. 2017). With the advent of Next Generation Sequencing technologies, genome sequence information for most organisms is becoming readily accessible and allows for the design of highly specific RNA probes. With RISH the evaluation of cell proliferation can be expanded to alternate splice variants, non-standard model organisms and tissues such as those of primary NP cells in culture derived from the bovine IVDs and combined with IHC for previously established antibodies. We show the colocalization of Ki67 mRNA by RISH alongside Ki67 and CDK4 proteins (Fig. 3). While Ki67 transcripts are present in pre-metaphase and metaphase cells (Fig. 3a–e), Ki67 protein is by far more abundant in cells during later stages of the cell cycle (Fig. 3c–e) as seen for those in metaphase (white arrowhead, Fig. 3e). The immunogen for the CDK4 antibody employed here was recombinant human CDK4 protein according to the manufacturer (Table 1) detecting both the bovine epitope (Fig. 3i–j) and human CDK4 (data not shown), while it was absent in control experiments omitting the primary antibody (data not shown). Based on the combination of Ki67 mRNA which precedes its translation and the presence of CDK4, it appears that most of the NP cells cultured here were in the G1/S phase, with those showing increased Ki67 transcript having likely progressed further into G1 than others (green arrow head Fig. 3f,j) yet not as far as those with high Ki67 protein and transcript levels. The latter have likely advanced further in the cell cycle towards S phase (red arrow head, Fig. 3f,g,j).

Fig. 3. Example of a combined RISH and IHC procedure for cell cycle markers Ki67 and CDK4 on primary bovine NP cells in culture.

Alexa-488 conveyed fluorescence reflects the bovine Ki67 transcript, Alexa-568 conveyed fluorescence reflects the presence of Ki67 protein, while Alexa-647 conveyed fluorescence indicates presence of CDK4 protein. DAPI stains the nucleus. The white arrowhead in (e) points to a NP cell in metaphase. The green and red arrowheads in (f-j) point to Ki67 transcript and Ki67 protein, respectively

By now, several commercial systems have been developed to allow for multi-probe RNA in situ hybridization often based on direct probe-labelling with amine modified UTPs and amine reactive dyes for detection (Cox and Singer 2004). These probes can be derived by in vitro transcription of plasmid cloned template DNAs as used in the FISH Tag™ RNA Multicolor Kit (Thermo Fisher). Other technologies use single labelled oligonucleotides (Raj and Tyagi 2010), Stellaris® (LGC Biosearch Technologies) requires a minimum of 25 labelled oligonucleotides per transcript for reliable detection as recommended by the manufacturer (Orjalo and Johansson 2016). RNAScope® (Advanced Cell Diagnostics) uses a complex double-Z probe design strategy for target detection (Wang et al. 2012) and more recently in situ sequencing was commercialized by CARTANA, a technology introduced by the Church group (Lee et al. 2015). Some of these in situ protocols can be combined with IHC (Gross-Thebing et al. 2014; Shih et al. 2011). However, commercialized kits and systems can be uneconomic for non-routine research use (George et al. 2018) and others require complex probe design strategies or specialized equipment (Wang et al. 2012). The protocol we have described here is easy to follow, versatile and economic in small scale research settings.

Lastly, while (sc) transcriptomics and current proteomics approaches provide data on a range of cellular molecules, methods for contamination-free isolation of single cells that will not bias a cells transcriptome or proteome in the process are still under development (Hedlund and Deng 2018; Kraus et al. 2019a; Marx 2019). Combining RISH and IHC for multiple RNA transcripts and protein epitopes as outlined here, allows for assessment with single-cell resolution in a biologically meaningful context. Furthermore, combined with high resolution imaging, like confocal microscopy, studying subcellular transcript and protein localization in a qualitative and quantitative manner is feasible (Li et al. 2019a). Combining RISH and IHC for the same target could address important questions about RNA and protein turnover rates and further elucidate whether a cell population is truly phenotypically heterogeneous or if cells are at a different phase in the cell cycle instead (Marx 2019).