AAV-Mediated CRISPRi and RNAi Based Gene Silencing in Mouse Hippocampal Neurons

Uncovering the physiological role of individual proteins that are part of the intricate process of cellular signaling is often a complex and challenging task. A straightforward strategy of studying a protein’s function is by manipulating the expression rate of its gene. In recent years, the Clustered Regularly Interspaced Short Palindromic Repeat (CRISPR)/Cas9-based technology was established as a powerful gene-editing tool for generating sequence specific changes in proliferating cells. However, obtaining homogeneous populations of transgenic post-mitotic neurons by CRISPR/Cas9 turned out to be challenging. These constraints can be partially overcome by CRISPR interference (CRISPRi), which mediates the inhibition of gene expression by competing with the transcription machinery for promoter binding and, thus, transcription initiation. Notably, CRISPR/Cas is only one of several described approaches for the manipulation of gene expression. Here, we targeted neurons with recombinant Adeno-associated viruses to induce either CRISPRi or RNA interference (RNAi), a well-established method for impairing de novo protein biosynthesis by using cellular regulatory mechanisms that induce the degradation of pre-existing mRNA. We specifically targeted hyperpolarization-activated and cyclic nucleotide-gated (HCN) channels, which are widely expressed in neuronal tissues and play essential physiological roles in maintaining biophysical characteristics in neurons. Both of the strategies reduced the expression levels of three HCN isoforms (HCN1, 2, and 4) with high specificity. Furthermore, detailed analysis revealed that the knock-down of just a single HCN isoform (HCN4) in hippocampal neurons did not affect basic electrical parameters of transduced neurons, whereas substantial changes emerged in HCN-current specific properties.


Introduction
Information processing in the nervous system relies on the coordinated activity of electrical and chemical signals. At the level of individual neurons as well as in neuronal networks, the depolarization of the membrane voltage leads to the generation of action potentials, which finally cause the release of transmitters that transduce the signal to downstream target cells. Many ion channels that actively contribute to the electrical properties of neurons have been intensively studied by electrophysiological, biochemical, pharmacological, as well as molecular biological methods (for reviews see: [1][2][3][4]). One gene family has been uncovered encoding ion channels with exceptional activation properties,

Heterologous Expression of HCN Channel Subunits
Human embryonic kidney cells (HEK293; #85120602) were obtained from ECACC/Sigma Aldrich (Taufkirchen, Germany) and grown in a low glucose containing medium (M10, MEM + Glutamax TM , 10% (v/v) fetal calf serum (FCS), 1% antibiotics/antimycotics, and 1% (v/v) non-essential amino acids (all from Gibco/Thermo Fisher Scientific, Darmstadt, Germany). The cells were propagated in 9 cm petri dishes at 37 • C, 5% CO 2 , and~95% relative humidity. Twice a week, when the cells reached approximately 90% confluency, they were trypsinized and seeded at densities of 1-1.4 × 10 6 cells onto new petri dishes. Transfections were performed with a modified calcium phosphate co-precipitation method in order to establish cell lines constitutively expressing individual HCN channel subunits (HCN1, 2, 4) [18]. The cell clones were selected in the presence of Geneticin (0.8-1 mg mL −1 in M10 medium). The functional expression of HCN channels and homogeneity of cell clones was examined electrophysiologically and by immunological staining, respectively.

Immunocytochemistry
Tables 1 and 2 list the primary and secondary antibodies used for immunocytochemistry (ICC), respectively. Cells or tissue slices were rinsed with PBS and then fixed in PFA (4% (w/v) paraformaldehyde in PBS) for 10 min at room temperature (RT). After several rinses with PBS, unspecific binding sites were blocked for 1 h at RT in blocking solution (CT: 0.5% (v/v) Triton X-100, 5% (v/v) ChemiBLOCKER (Merck, Darmstadt, Germany). Subsequently, the samples were incubated with primary antibodies (Table 1) in CT at 4 • C over night or at RT for 4 h, rinsed for several times with PBS, and then incubated with secondary antibodies (Table 2) in CT at RT for 1 h. Finally, the samples were washed with PBS, before mounting the coverslips containing cells or PTFE membranes containing slices in Aqua-Poly/Mount (Polysciences, Eppelheim, Germany) on microscopy slides. The fluorescent images were obtained with an inverted confocal microscope (TCS SP5II; Leica, Wetzlar, Germany).

Colocalization Analysis
HEK293 cell lines constitutively expressing individual HCN channel subunits (HCN1, 2, 4) were transfected with shRNA encoding plasmids using PEI transfection [22]. The cells were plated on coverslips in 24-well plates (Ibidi, Martinsried, Germany) that were precoated with poly-L-lysine (0.1 mg mL −1 ) with a density of 300 cells mm −2 . Cultures were kept in a humidified CO 2 incubator (5% CO 2 , 95% relative humidity) at 37 • C and maintained for two days before fixation. After immunocytochemistry and image acquisition, the images were analyzed using the ImageJ Coloc 2 analysis plugin [23].

Quantification of Gene Expression by Real-Time qPCR
Total RNA was isolated from PHNs or organotypic hippocampal slice cultures (OHSCs) after 14-15 d.i.v. using the RNeasy Mini Kit (Qiagen) according to the supplier's protocol. First-strand cDNA synthesis was performed according to the methods that were described in [27]. In brief, Oligo-dT primers (Qiagen) and Moloney Murine Leukemia Virus reverse transcriptase (M-MLV-RT, Life Technologies/Thermo Fisher Scientific) were used according to the supplier's protocol. Thermocycling was performed in a LightCycler 1.5 (Roche, Mannheim, Germany) using the QuantiTect SYBR Green PCR Kit (Qiagen). qPCR reactions were performed on 2 µL aliquots of first-strand cDNA samples in a total volume of 20 µL. The results were analyzed using the C t method. Gene-specific primers (Table 3) were designed in silico and synthesized by Eurofins (Ebersberg, Germany). The specificity of primers was confirmed via BLAST analysis. For normalization, gapdh was used as a reference gene. The gapdh primers were designed to bind in exons that are separated by an intron of 134 bp to check for genomic impurities.

Validation of Knock-Down Efficiencies for shRNAs/sgRNAs
ShRNA-encoding constructs were transfected by calcium phosphate co-precipitation into HEK293 cell lines constitutively expressing individual HCN channel subunits (HCN1, 2, 4). For each target gene, up to five different shRNA sequences (see Table 4 and Supplementary Table S1) were independently tested. The total RNA was isolated from HEK293 cells using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen) according to the supplier's protocol. SgRNA-encoding constructs were packed into viral particles of serotype 2 and PHNs were transduced with a MOI of 2 × 10 4 . For each target gene, up to four different sgRNA sequences were tested independently. Total RNA was isolated from PHNs using the RNeasy Mini Kit (Qiagen), according to the supplier s protocol. Those constructs that induced a robust and reproducible knock-down were chosen for further analysis.

Whole-Cell Patch-Clamp Analysis
Whole-cell patch-clamp recordings were performed at RT following the methods that were described by Hamill et al. [29]. Electrodes with tip resistances between 2.5 and 4 MΩ were fashioned from borosilicate glass with an inner diameter of 0.86 mm and an outer diameter of 1.5 mm (Harvard Apparatus, Holliston, MA, USA) with a temperaturecontrolled pipette puller (P1000, Sutter Instrument, Novato, CA, USA). For HEK293 cells and PHNs, the pipettes were filled with intracellular saline solution containing 10 mM KCl, 10 mM NaCl, 120 mM KGluconate, 10 mM EGTA, 10 mM HEPES, 4 mM MgATP, and 0.3 mM NaGTP, adjusted to pH 7.3 with KOH and an osmolality of~310 mOsm/L. During the experiments, the cells were constantly superfused with extracellular saline solution containing 150 mM NaCl, 4 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , and 10 mM HEPES, adjusted to pH 7.4 with NaOH and adjusted to 330 mOsm/L with glucose. Wholecell voltage-clamp and current-clamp recordings were performed while using an EPC10 patch-clamp amplifier (HEKA-Elektronik, Lambrecht, Germany) that was controlled by the program Patch Master (version 2.5; HEKA-Elektronik). Electrophysiological data were sampled at 20 kHz and low pass filtered at 2.9 kHz with a four-pole Bessel-filter. Offset potentials, electrode capacity, and membrane capacity were manually compensated. The liquid junction potential between intracellular and extracellular solutions was calculated and also compensated by adjusting the offset potential. Series resistance was compensated between 60 and 80% with a time constant (τ) of 100 µs.

Data Analysis
Electrophysiological data were analyzed using FitMaster (version 2; HEKA-Elektronik) or Igor Pro (version 6; Wavemetrics, Lake Oswego, OR, USA). The data are represented as mean ± s.d. (standard deviation) or as box and whisker plots. The two-tailed unpaired Student's t test was applied for the calculation of p values using GraphPad Prism (version 5; Graphpad Software Inc., La Jolla, CA, USA). A p value of <0.05 was considered to be significant.

Generation of AAV-Based Gene-Interference Tools
In addition to transgenic approaches, gene expression can be manipulated alternatively by the cell-autonomous defense mechanism of RNA inhibition (RNAi) [30]. By forming homologous RNA double strands, the degradation of the targeted mRNA is initiated. Consequently, the protein level decreases due to an impaired de novo protein biosynthesis. The core component necessary for RNAi is a short hairpin RNA (shRNA) that is expressed, e.g., under the control of constitutively active human U6 (hU6) promoter. In this study, different shRNA molecules were designed complementary to the target HCN channelencoding mRNA molecules. For the identification of transduced cells, an eGFP-reporter expressed either under the control of the neuron-specific CaM kinase II (CKII) promoter or the CMV promoter was used ( Figure 1(A a ,A b )). In all experiments that were conducted with neurons, constructs were delivered by recombinant Adeno-associated viruses (rAAVs). (B b ) Schematic representation of sgRNA and dSaCas9-encoding constructs used in this study. Expression of the eGFP reporter could be driven by the CMV-or CKII promoter. For control experiments, variants with dSaCas9 fused to eGFP were generated. The modulatory design of the vector also allows to exchange dSaCas9 to SaCas9 for conventional CRISPR experiments. Abbreviations: ITR, inverted terminal repeat; hU6, human U6 promoter; shRNA, short-hairpin RNA; CKII, CaM kinase II promoter; CMV, cytomegalovirus promoter; eGFP, enhanced green fluorescent protein; Ori, origin of replication; Amp r , ampicillin resistance cassette; NLS, nuclear localization site; dSaCas9, nuclease deficient Staphylococcus aureus Cas9; HA, human influenza hemagglutinin tag; KRAB, Krüppel-associated box; sgRNA, short-guidance RNA; Kan r , kanamycin resistance cassette.
As an alternative strategy, a modified version of the "Clustered Regularly Interspaced Short Palindromic Repeats" (CRISPR) technique was applied and compared with the RNAi strategy. The CRISPR technique typically leads to a gene knock-out initiated by enzymatically active Cas9 proteins, cutting the genomic DNA. The DNA double strand break triggers cellular repair mechanisms, which eventually cause a gain or loss of nucleotides that ultimately disturb a genes' open reading frame [31][32][33][34][35]. In contrast, the CRISPR interference (CRISPRi) strategy blocks RNA synthesis by the occupation of the target gene's promoter and thereby impairs protein de novo biosynthesis [36][37][38]. The main component of CRISPRi is a nuclease-deficient version of Cas9 (dSaCas9), which is fused to the Krüppel-associated box (KRAB) repression domain, resulting in a transcriptional interference protein complex (dSaCas9-KRAB; Figure 1(B a ,B b )) [37,39]. The expression of this fusion protein was driven either by the CKII-or the CMV promoter. Because of the flanking nuclear localization sites, the dSaCas9-KRAB complex translocates into the nucleus, where, assisted by short guidance RNAs (sgRNAs), it binds to the transcriptional start site (TSS) of the target gene and interferes with the transcription machinery. Similar to shRNAs, the sgRNAs were expressed under control of the hU6 promoter. For the identification of transduced cells, an HA-tag was fused to the dSaCas9 protein. We generated an AAV-vector backbone in which the sgRNA expression scaffold, the HA-tag, and the KRAB motif can be exchanged for an eGFP-encoding cassette that is directly fused to the Cas9-encoding element in order to achieve more flexibility and allow direct proof of Cas9 expression ( Figure 1(B a ,B b )).

Functional Testing of Knock-Down Constructs
In order to check for the functionality of the different knock-down constructs, we transfected HEK293 cells and examined the expression and localization of the eGFP reporter as well as HA-tagged dSaCas9. Figure 2(A a ) shows the expression of eGFP under control of the CMV promoter. The distribution of eGFP was not restricted to specific cellular compartments. Figure 2(A a -A c ) show the expression of the dSaCas9 either fused to an HA-tag (2A b ) or to an eGFP (2A c ). The protein was preferentially located in the nucleus, due to nuclear localization sites (NLS) flanking the dSaCas9 cassette (see Figure 1(B a ,B b )).
shRNA-encoding constructs targeting individual HCN channel isoforms were transfected in HEK293 cells constitutively expressing homomeric HCN channels 1, 2, or 4 in order to validate the specificity and efficacy of the RNAi-mediated knock-down on the protein level. Two days after transfection, the cells were fixed and stained with specific antibodies (Figure 2(B a -B c )). Capturing confocal images of HCN channel fluorescence and eGFP fluorescence allowed for calculating Pearson R values for the colocalization of eGFP and HCN channel signals. The immunofluorescence signal of the HCN channel will decrease in those cells in which HCN channel expression is downregulated by shRNAs, whereas the eGFP-reporter signal remains constant. Consequently, the Pearson R value decreases as a measure for colocalization of both signals. Normalized colocalization analyses are shown in Figure 2(C a-C c ). Only shRNA1 (sh1), which is complementary to hcn1 mRNA downregulated HCN1 channel protein expression (Figure 2(C a )), whereas the control shRNA (shScr) and shRNAs complementary to hcn2 and hcn4 mRNAs did not change the expression level of HCN1 (shScr: 1.0 ± 0.159; sh1: 0.149 ± 0.115). Similarly, only shRNA2 (sh2), which is complementary to hcn2 mRNA (shScr: 1.0 ± 0.189; sh2: 0.122 ± 0.042) (Figure 2(C b )), and shRNA4 (sh4), which is complementary to hcn4 mRNA (shScr: 1.0 ± 0.361; sh4: 0.152 ± 0.062) (Figure 2(C c )) induced the specific downregulation of their target transcripts and a reduction in channel protein expression. (C a -C c ) Colocalization analysis by comparison of Pearson's R values for HEK293 cells (C a ) constitutively expressing HCN1 channels and different shRNAs (shScr, sh1, sh2 and sh4), (C b ) constitutively expressing HCN2 channels and different shRNAs (shScr, sh1, sh2 and sh4), and (C c ) constitutively expressing HCN4 channels and different shRNAs (shScr, sh1, sh2 and sh4). The data were obtained from indicated numbers of fluorescent images from at least five independent transfections. Pearson's R values were normalized to shScr controls and results are depicted as mean ± standard deviation. Schematic of AAV-delivered constructs are displayed above the merged immunofluorescence images.
From all of the constructs tested (see Table 3 and Supplementary Table S1), only those shRNAs reliably demonstrating the highest specificity and efficacy were selected to generate rAAVs. Primary cultures of hippocampal neurons that were prepared from C57BL/6J mice ( Figure 3A) were transduced with these rAAVs. The vectors either encoded shRNA and eGFP ( Figure 3(B a )), dSaCas9-HA ( Figure 3(B b )), or dSaCas9-eGFP (Figure 3(B c )). Twelve days after transduction, neurons were fixed and the expression of reporter proteins was studied by immunocytochemistry. Similar to the expression pattern in HEK293 cells, in neurons expressing shRNAs, eGFP expression was not restricted to a specific cellular compartment. The dSaCas9 fused either to the HA-tag or to eGFP was located in the nucleus.  (D a ,D b ), or eGFP-tagged dSaCas9 (D c ). The eGFP, HA-tag, and the neuron-specific microtubule-associated protein 2 (MAP2) protein were immunostained with specific anti-GFP, anti-HA, and anti-MAP2 antibodies combined with fluorescently labeled secondary antibodies (eGFP and HA-tag, green; MAP2, red). Nuclei were labeled with TOPRO (blue). Schematic of AAV-delivered constructs are displayed above the merged immunofluorescence images.

AAV-Mediated Knock-Down of hcn Gene Expression in Neurons
Cultured PHNs express at least three HCN channel isoforms that can be quantified by qRT-PCR ( Figure 4A). The analyses showed that hcn2 transcripts were most abundant (72 ± 12.3%), followed by hcn1 and hcn4 transcripts (hcn1: 22 ± 2.6%; hcn4: 6 ± 0.9%). Even individual neurons expressed all three isoforms, as illustrated by immunocytochemistry ( Figure 4(B a -B c )). Therefore, we aimed to downregulate each of the channel isoforms specifically and independently by RNAi and CRISPRi in order to determine the strategy best suited for hcn gene knock-down in postmitotic cells. For CRISPRi-mediated knockdown, sgRNAs were designed to target regions between −50 and + 300 bp relative to the predicted TSS. The knock-down efficacies of these sgRNAs were compared to a scrambled sgRNA (sgScr), which was designed not to bind to any endogenous promoter. In order to evaluate the CRISPRi efficacy, we also determined the effects of RNAi using the shRNAs that were previously validated in the colocalization experiments. For each construct, rAAVs representing serotype 2 and 9 were generated. While rAAV2 is considered to transduce a broad range of cell types and tissues, rAAV9 is more efficient in transducing neurons [40]. In neurons that were transduced with rAAV2 virions, each of the previously identified shRNAs reduced the mRNA levels of the target gene in comparison to untreated wildtype and shScr-treated controls ( Figure 4(C a -C d )). Sh1 reduced the expression of hcn1 transcripts to 58.3 ± 11.2% of the shScr control level. Similarly, sh2 reduced the expression of hcn2 transcripts to 42.3 ± 18.3% of the shScr control level and sh4 reduced expression of hcn4 transcripts to 60.1 ± 32.9% of the shScr control level. Notably, only one sg construct (sg1) caused a robust reduction of hcn1 mRNA to a level of 51.8 ± 23.6% of the sgScr control ( Figure 4(C a -C d )). Using rAAV9 virions for transduction ( Figure 4(D a -D d )), the knockdown efficiencies were: sh1: 65.2 ± 13% of the shScr control level; sh2: 24 ± 15.5% of the shScr control level; and, sh4 25.2 ± 8.1% of the shScr control level). The knock-down efficiencies of sg1, sg2, and sg4, were markedly improved when delivered by rAAV9 virions (sg1: 71.1 ± 21.5% of the sgScr control level; sg2: 66.3 ± 6.6% of the sgScr control level; sg4 83.9 ± 4.1% of the sgScr control level). However, when compared to RNAi, CRISPRi inhibited hcn1, 2, and 4 gene expression moderately.
While primary hippocampal neurons are perfectly suited to study a neurons' physiology on the single cell level, organotypic hippocampal slice cultures (OHSCs) are widely used to study network properties. Therefore, we aimed to examine the effects of HCN channel knock-down in OHSCs that were prepared from C57BL/6J mice ( Figure 3C). After two weeks of cultivation and 12-13 days after transduction with rAAV9 expressing shRNA and eGFP ( Figure 3(D a ,D b )) or dSaCas9-eGFP (Figure 3(D c )), the samples were fixed, and the expression of reporter proteins was examined immunohistochemically. In agreement with the previous experiments performed on PHNs, the transduction of OHSCs with rAAV9 virions encoding shRNAs yielded a high number of transduced neurons. Furthermore, rAAVs inducing RNAi and rAAVs causing CRISPRi both reproduced the localization pattern of reporter proteins previously observed in PHNs. The eGFP reporter that was encoded by shRNA containing constructs was homogenously distributed in the cell, whereas eGFP-dSaCas9 was restricted to the nucleus (Figure 3(D a -D d )).
OHSCs express hcn1, hcn2, and hcn4 transcripts, as well as HCN1, HCN2, and HCN4 channel proteins (Figure 4(E,F a ,F b )). qRT-PCR and immunohistochemistry both indicated that HCN2 is the most abundant channel isoform in OHSCs (qRT-PCR quantification: hcn1 5.3 ± 2.5%; hcn2 92.1 ± 28.4%; hcn4 5.5 ± 1.3%). Thus, we decided to exemplarily manipulate HCN2 channel expression in OHSCs. sh2 and sg2 both reduced hcn2 mRNA levels when compared to untreated wildtype and shScr-or sgScr-treated controls (sh2: 52.1 ± 18.9%; sg2: 62.7 ± 21.9%) (Figure 4(G a ,G b )). Therefore, we conclude that both of the techniques are suitable for manipulating the expression levels of hcn transcripts in a variety of culture systems. . The transcript expression levels were normalized to gapdh. Values shown are calculated to 1 as the sum of all hcn transcripts. cDNA was prepared from five culture inserts, each containing three individual slices. In total, slices were obtained from three different animals. (F a ,F b ) Representative immunofluorescence images showing the expression of HCN-channel isoforms 1 (green) and 2 (blue). The isoforms were stained using specific antibodies. Enlargements show HCN-isoform expression in hippocampal CA1 (1) and CA3 (2) subfields. (G a ,G b ) qRT-PCR knock-down analysis of hcn2 mRNA levels in organotypic slices after transduction with shRNA or sgRNA/dSaCas9 expressing rAAV9. The data were obtained from indicated numbers of culture inserts, each containing three individual slices. In total, slices were obtained from at least three different animals. The results are depicted as mean ± standard deviation. Statistical significance was assessed using the unpaired two-tailed Student's t test, * p < 0.05, ** p < 0.01, *** p < 0.001.

Electrophysiological Characterization of HCN4 Channel Knock-Down
Because RNAi efficiently reduced HCN channel transcript levels in HEK293 cells, PHNs, and OHSCs, we decided to evaluate the effects of hcn4 gene knock-down on both basic neuronal parameters and I h -current characteristics while using standard electrophysiological patch-clamp experiments.
To obtain detailed insight into HCN4 subunit-specific functions, we measured I hcurrents at different membrane potentials and constructed current-voltage relationships, from both HEK293 cells expressing homomeric HCN channels and PHNs expressing heteromeric HCN channels ( Figure 5(A a -A c ,B a -B c )). Half-maximal activation voltages recorded from HEK293 cells confirmed the differences in activation potentials of homomeric HCN channel currents [9]. While homomeric HCN1 channels activate at more positive membrane potentials, HCN2 and HCN4 homomeric channels activate at more negative membrane potentials (HCN1: −96.14 ± 1.63 mV; HCN2: −108.5 ± 1.3 mV; HCN4: −119.3 ± 5.86 mV), thus indicating an activation sequence of HCN1 > HCN2 > HCN4 from more depolarized to very hyperpolarized potentials ( Figure 5(A a -A c )). In PHNs, the knock-down of HCN4 leads to a shift in half-maximal activation voltage of about 5 mV from hyperpolarized to more depolarized potentials as compared to shScr-treated control neurons (shScr: −111.3 ± 3.84 mV; sh4: −106.4 ± 3.42 mV) ( Figure 5(B a -B c )). We measured the activation time constants (τ) of homomeric HCN channels in HEK293 cells and of HCN channels endogenously expressed in PHNs in order to analyze activation kinetics of HCN channel-mediated I h -currents ( Figure 6(A a ,A b ); see also Supplementary Figures S5 and S6). Similar to the differences in activation potentials, homomeric HCN channel currents differed in their activation kinetics, as illustrated by differences in τ values derived from I h -currents measured at −130 mV. While homomeric HCN1 channels activated with a time constant of 71.6 ± 43.9 ms, homomeric HCN2 channels activated with a time constant of 269.7 ± 97 ms and homomeric HCN4 channels activated even slower with a time constant of 934.4 ± 183.2 ms. The differences in the activation kinetics were also apparent in the half-width of Sag-potentials. While halfwidths of Sag potentials from homomeric HCN1 channels were 37.9 ± 21 ms, half-widths of Sag potentials from homomeric HCN2 and HCN4 channels were 83.29 ± 38.9 ms and 298.6 ± 56.47 ms, respectively. The knock-down of HCN4 in PHNs increased the speed of activation of endogenous HCN channels when compared to shScr-treated neurons. This increase in kinetics resulted in faster activation time constants (shScr: 382 ± 181.1 ms; sh4: 238.1 ± 131.6 ms) ( Figure 6(A a ,A b )) and also in smaller Sag potential half-width values (shScr: 143.2 ± 76.27 ms; sh4: 92.06 ± 23.82 ms) ( Figure 6(B a ,B b )). Thus, the knock-down of HCN4 spares an I h current that has inherent characteristics of heterotetrameric currents that are mainly composed of HCN1 and HCN2 channel subunits.

Discussion
For studying the signal transduction and communication pathways in biological systems, there is a need for applying methodologies that allow less invasive approaches to assess the function of individual proteins in single cells and on the network level in tissue or in living organisms. In comparison to transgenic approaches that are time consuming and bear the risk of unexpected developmental defects, methods to impair transcript levels of a gene of interest, i.e., gene knock-down strategies, hold great promise in overcoming the limitations of genome changing or editing strategies.
Here, we examined two independent experimental approaches, i.e., RNAi and CRISPRi, in order to reduce the expression of HCN channel-encoding genes at the mRNA level without altering the gene's nucleotide sequence. The main finding of our study was that both procedures reduced hcn transcript levels in transgenic cell-lines, primary hippocampal neurons (PHNs), and organotypic hippocampal slice cultures (OHSCs) for each of the three targeted HCN isoforms. We used rAAVs as cargo vehicles to overcome constraints of construct delivery to neurons. Because rAAV serotypes differ in their transduction efficacy for certain cell types or tissues, we used rAAV2 virions that are known to transduce a broader spectrum of cell types, and rAAV9 virions that are known to be particularly suited for neuronal and glial cell transduction, respectively [41,42]. Furthermore, the knock-down of HCN4 channel subunits in PHNs unraveled unique insights into single cell I h -current related electrophysiological properties.
The main goal of this study was to examine the specificity and efficacy of two independent strategies to uncover a suitable system for knock-down applications of hcn gene expression in forthcoming studies. Initially, an all-in-one CRISPRi system was developed and tested in recombinant cell-lines and PHNs. In contrast to conventional CRISPR gene-editing tools [31][32][33][34][35], CRISPRi is based on a mutation in the Cas9 gene, leading to a loss of nuclease activity of the enzyme (dSaCas9) [17,[36][37][38]. The dSaCas9-encoding cDNA was fused to the Krüppel-associated box (KRAB) repression domain to generate a protein complex that is able to interfere with gene transcription [27,39]. The constructs were delivered via rAAVs, which are known to be non-immunogenic and they have been approved for therapeutic applications [43,44]. However, the maximal cargo size of AAV genomes is~4.5 kb [45,46]. Incorporating the native Streptococcus pyogenes Cas9 (SpCas9) gene (4.2 kb) [31] and a sgRNA-encoding cassette into the vector backbone exceeded the size limits of the virion's capsid. In order to overcome this limitation, the SpCas9-encoding gene was substituted for an orthologous gene, previously identified in Staphylococcus aureus (SaCas9). The SaCas9-encoding cDNA has a size of 3.2 kb and the inactive version of the enzyme has been generated (dSaCas9) [24]. Thus, the all-in-one cloning vector (see Figure 1) contained the enzymatically inactive dSaCas9 fused to the Krüppel-associated box repression domain (dSaCas9-KRAB) and it was flanked by nuclear localization sites facilitating nuclear translocation of the fusion protein. The expression of the dSaCas9-KRAB protein was driven by exchangeable RNA polymerase II promoters, with the CMV-promoter being active in a wide range of cell types, whereas the CKII-promoter is preferentially active in neurons. sgRNA-encoding elements controlled by an RNA polymerase III promoter (human U6-promoter) were integrated to achieve specific binding of dSaCas9-KRAB to the promoter region of a gene of interest. The modular vector design also allowed for expressing fluorescent reporters, like eGFP, to visualize the transduction success. Finally, enzymatically active SaCas9 can be substituted for dSaCas9-KRAB to perform conventional CRISPR gene-editing experiments. The versatility and functionality of the vector was confirmed in independent experiments showing that the dSaCas9-KRAB protein was restricted to the nucleus of transfected HEK293 cells (Figure 2(A a -A c )) and transduced PHNs or OHSCs (Figure 3(B a -B c /3D a -D c )).
Because the promoters of murine HCN channel genes have not been experimentally determined, we used the Eukaryotic Promoter Database (EPD) [25] to predict the TSS of the mouse hcn1, 2, and 4 genes. For each of these genes, we designed at least three independent sgRNAs (Supplementary Table S1). The transduction of PHNs with rAAV2 and rAAV9 virions led to three fundamental findings: (1) for each target gene, a sgRNA molecule was identified that caused reduction of transcript levels, (2) rAAV9 virions were better suited than rAAV2 virions for delivering the cargo into PHNs as well as OHSCs, and (3) the knock-down effects were gene-specific. However, in comparison to controls that were conducted with scrambled sgRNA versions, the hcn transcript levels were reduced in neurons (~20-40% reduction). An explanation for this moderate efficiency at hcn gene loci might be imprecise assignment of the TSS regions, the most challenging part in designing CRISPRi experiments. Upon binding of the dSaCas9-KRAB-sgRNA complex to the target sequence, KRAB induces heterochromatin formation [47], and finally impairs RNA polymerase from initiating transcription. Several online tools are currently available to assist in identifying gene promoters [48]. To improve the CRISPRi efficacy for hcn transcript knock-down, one could re-examine the current TSS annotation with independent prediction algorithms to uncover additional and potentially favorable sgRNA target sites [48]. Alternatively, the TSS could be experimentally determined by molecular biological tools, like 5 primer-extension or 5 rapid extension of cDNA ends (RACE) [49,50]. However, these experiments were beyond the scope of the current investigation. Nevertheless, the CRISPRi approach demonstrated specific and reasonable hcn transcript knock-down. In comparison to conventional CRISPR/Cas experiments, CRISPRi might overcome potential complaints concerning off-target effects, due to enzymatically inactive dSaCas9 [39].
In comparison to the results that were obtained with CRISPRi, RNAi approaches resulted in robust knock-down of hcn transcript levels in recombinant cell lines, PHNs, as well as OHSCs. Interfering with transcript levels by RNAi strategies has the advantage that the target site(s) to which shRNA molecules can bind are less restricted. In principle, the entire primary transcript of a gene might serve as a template for shRNA binding [30]. For each hcn gene, we designed at least four individual shRNA molecules (Supplementary  Table S1) and examined their efficacy and specificity in recombinant cell lines expressing homomeric HCN channels. For each hcn gene, we identified at least one shRNA molecule (see Table 3) that effectively reduced the expression of its target gene and did not cross-react with the other HCN channel-encoding genes. In transgenic cell-lines, transcript and protein levels were specifically reduced when compared to control values. With approximately 80% reduction, an impairment of hcn2 and hcn4 transcript levels in rAAV-transduced PHNs reached similar values to those that were determined in transgenic cell lines.
The main physiological question in this study was the dissection of the specific contribution of a single HCN isoform to the neuronal I h current expressed in hippocampal pyramidal neurons. HCN channels play an important role in neuronal signaling and, here, we specifically examined the effects of HCN4 channel loss on basic electrophysiological properties of hippocampal neurons. As the basal transcript level of hcn4 in hippocampal neurons is low when compared to hcn1 and, especially, hcn2 transcript levels, the HCN4 isoform might serve to diversify native I h currents in heterotetrameric channels [9]. Indeed, electrophysiological recordings of PHNs showed that HCN4 has a substantial influence on I h current activation potentials and kinetics, but not on basic neuronal properties, like resting membrane potential or input resistance, which is in contrast to the knock-out and knock-down of HCN1 (see Figures S5 and S6). Therefore, we suggest that the level of hcn4 expression might diversify I h current kinetics and activation potentials. By incorporating the subunit into functional heterotetrameric ion channels, the pronounced sensitivity of the HCN4 subunit to cyclic nucleotides might additionally modulate channel properties [6]. Notably, in a previous study, we have shown that a knock-down of HCN4 channel expression in the hippocampus of adult mice lead to a pronounced anxiogenic effect, which had not been observed, e.g., by HCN1 channel down-regulation [50]. In summary, our study has shown that RNAi-mediated knock-down of HCN channel expression in recombinant cell lines, PHNs, and OHSCs is efficient and specific. In comparison to transgenic approaches, like gene knock-out, conditional gene knock-out, or gene-editing by CRISPR/Cas9, the current experimental design is less time consuming as well as less prone to potential offtarget effects or developmental failure. Furthermore, using rAAVs for cargo delivery to post-mitotic PHNs supports current ideas [43,44,51] to use viral transductions in basic and translational approaches.

Supplementary Materials:
The following are available online at https://www.mdpi.com/2073-4 409/10/2/324/s1, Figure S1: rAAV-mediated gene knock-down by CRISPRi and RNAi, Figure S2: sgRNAs guide dSaCas9 to the promoter of target genes, Figure S3: sgRNAs guide dSaCas9 to the promoter of target genes, Figure S4: HCN4 channel knock-down has no effect on basic parameters of PHNs, Figure S5: Immunohistochemistry and I h -current comparison of HCN1 knock-out vs knock-down PHNs, Figure S6: Effects of HCN-channel knock-out and knock-down on I h -current activation and sag potential, Table S1: Sequences and positions of all tested sgRNAs and shRNAs, Table S2: Electrophysiological properties of hippocampal neurons and HEK293 cells. Funding: Work in the laboratory of C.R.R. was funded by the DFG (SPP1757: Ro2327/9-2 and FOR2795: Ro2327/13-1).

Institutional Review Board Statement:
Not applicable because no experiments were performed on live animals in this study.

Informed Consent Statement: Not applicable.
Data Availability Statement: The data and materials presented in this study are available on request from the corresponding author.