siRNA-Mimetic Ratiometric pH (sMiRpH) Probes for Improving Cell Delivery and mRNA Knockdown

Abstract

Second-generation siRNA-mimetic ratiometric pH probes (sMiRpH-2) were developed by hybridizing a 3′-FAM-labeled 2′-OMe RNA strand with a 3′-Cy5-labeled 25mer RNA strand. These duplexes demonstrated the silencing of cytoplasmic mRNA targets in HeLa cells as measured by RT-qPCR and supported by western blot analysis. Fluorescence intensity and lifetime measurements revealed that a single guanosine (G) positioned adjacent to FAM achieves substantial static quenching at pH 5, with additional collisional quenching rendering the dye almost nonemissive. A FAM-G π–π stacking interaction was evidenced by a red-shifted absorbance spectrum for FAM. Decreased quenching at near-neutral pH enhances the FAM dynamic range in the physiologic pH window and improves the differentiation in cells between endocytic entrapment and cytoplasmic release. Flow cytometric analysis of intracellular pH and uptake using sMiRpH-2 was corroborated by live cell confocal microscopy and found to be predictive of knockdown efficacy. A sMiRpH-2 probe successfully predicted the relative efficacy of two transfection agents in more challenging SK-OV-3 cells, which highlights its use for the rapid assessment of nonviral siRNA delivery vectors.

Introduction

Endosomal release into the cytoplasm remains a consistent obstacle to the successful delivery of RNA vaccines and therapeutics.^1,2 The differential hydronium ion concentration between acidic endosomes and the neutral cytoplasm permits direct, quantitative monitoring of endosomal escape via real-time measurements of intracellular pH.³ Multiple strategies ranging from the use of calcein and acridine orange as cargo surrogates to split-protein complementation assays have been reported.⁴⁻¹⁰ Conjugation of pH-sensitive dyes to oligonucleotide payloads permits facile synthesis while enhancing specificity, sensitivity, and quantitation.¹¹⁻¹⁶

We recently reported the development of a siRNA-mimetic ratiometric pH probe (sMiRpH-1, Scheme 1) to visualize and quantify RNA uptake in cells.¹⁷ Our design featured pH-invariant Cy5 and pH-sensitive FAM conjugated directly to the 3′- and 5′-termini of a nuclease-resistant 2′-OMe RNA sense strand, which was hybridized to a 21mer RNA antisense strand. sMiRpH-1 reported specifically on intracellular pH, which can vary by more than two pH units in the process of endocytosis, endosome maturation/acidification, and endosomal release. Cy5 is an internal standard that controls for variable cellular uptake, and intracellular pH can be quantified from the FAM/Cy5 fluorescence intensity ratio. Although the attenuating effect of proximal guanine on the intensity of FAM fluorescence has been reported previously,^18,19 we discovered that FAM experiences a substantial increase in pH sensitivity (i.e., dynamic range) when positioned near guanine (G) bases,¹⁷ which is further elucidated here. In one example, sMiRpH-1 was capable of silencing CSNK2β, but further investigation revealed that siRNA-mimetic probes require further modifications to achieve knockdown of cytoplasmic mRNA targets (Figure S1A,B). For example, chemical phosphorylation of the 5′-terminus of the antisense strand was investigated;²⁰ however, these double-stranded probes showed no improvement in biological activity by western blot analysis (Figure S1C). This motivated the development of sMiRpH-2 (Scheme 1) for assessing the delivery of viable siRNAs to the cytoplasm.

Scheme 1. sMiRpH-2 Probe Design Introduces siRNA Activity and Enhances pH Sensitivity.

To improve the sMiRpH platform for screening siRNA activity, we considered a design that is more compatible with the knockdown of cytoplasmic gene targets. De Smedt and colleagues previously suggested that activation of the RNA-induced silencing complex (RISC) does not occur when both termini of the sense strand are terminally dye labeled.²¹ According to the classical paradigm, siRNA consists of two complementary

21mer RNA strands each with two 3′-terminal overhanging deoxyribonucleotides (usually thymine) to protect against exonucleases en route to the target. Although most published siRNAs adhere strictly to this format, some notable exceptions have been reported. For instance, Dicer-substrate small interfering RNA (DsiRNA)^23,24 is an asymmetric, longer siRNA variant that is a substrate of both Dicer and Argonaute2 (Ago2) enzymes. In this design, a 5′-phosphorylated 25mer sense strand is hybridized to a 27mer antisense strand, creating a duplex with one overhanging and one blunt end. Dicer cuts this duplex into multiple double-stranded products, which are subsequently funneled to RISC where the target mRNA is degraded by Ago2. DsiRNA often works more effectively than classical design. Furthermore, Schwille et al. demonstrated preservation of gene silencing ability in a prevalidated siRNA against Firefly luciferase in HeLa S3 cells when the 3′-termini were labeled with AlexaFluor 488 and Cy5.²⁵ We applied these lessons to a sMiRpH-2 platform (Scheme 1), in which the 3′-termini of a 25mer 2′-OMe RNA sense strand and 5′-phosphorylated RNA antisense strand (modified with two 3′ DNA nucleotides) were covalently labeled with FAM and Cy5, respectively. 2′-OMe RNA and the 3′-terminal DNA bases are resistant to the activity of RNases including Dicer,^17,26,27 preventing the separation of the FAM label from the sense strand. Additionally, full ribose 2′-O-methylation of the sense strand has been shown to bias RISC toward selection of the antisense strand by Ago2.²⁸ In the present report, we show by both real-time quantitative polymerase chain reaction (RT-qPCR) and western blot analyses that sMiRpH-2 probes silence the cytoplasmic GAPDH, HPRT1, and SOD1 mRNA targets in HeLa cells. Based on fluorescence intensity and lifetime analyses, we attribute the previously reported dynamic range enhancement¹⁷ to a synergistic and distance-dependent combination of contact and collisional quenching interactions between FAM and guanine that is stronger within the acidic endosomal compartment than in the neutral-basic environment of the cytoplasm. We further find that intracellular pH and uptake, as detected by flow cytometry and live cell confocal microscopy, are predictive of silencing efficacy. Finally, we show that the anti-GAPDH sMiRpH-2 construct predicts the relative efficacy of two lipofection reagents in the difficult-to-transfect SK-OV-3 cell line, underscoring our platform’s ability to assess nonviral siRNA delivery agents on a short time scale.

Results

Design of the sMiRpH-2 Platform

To elucidate how proximal G nucleotides contribute to FAM pH sensitivity, we measured the fluorescence intensity for 10 FAM-labeled 2′-OMe RNA 5mer test oligonucleotides at 1.0 μM, in pH 5.0 and 7.5 PBesque, a buffer of high and constant ionic strength (Table 1). The low concentration was necessary to minimize dye self-quenching. We defined the dynamic range (DR) as the ratio of fluorescence intensity at pH 7.5 to that at pH 5.0 (Table S5). We also computed a quenching efficiency (QE) for each sample relative to the FAM-UUUUU oligo at both pH values.²² The concentration of oligonucleotide solutions was estimated using the extinction coefficients determined by IDT.^29,30 At both pH values, the fluorescence intensity (F) rose and QE fell with an increasing FAM-G distance for the singly G-substituted oligos. A similar distance dependence was observed when comparing the FAM-GGUUU, FAM-UGGUU, and FAM-UUUGG oligos, which feature G-clustering at short and longer distances from FAM. These additional proximal G nucleotides provided a modest DR enhancement. A photoinduced electron-transfer (PET) mechanism has previously been reported for FAM and G.^19,31 A new finding reported here is significant static quenching for all but the most distal G-substituted single-stranded oligos at both pH 5.0 and 7.5 (Table 1 and Table S4). We attribute this static quenching to π–π stacking between G and FAM, evidenced by a red-shifted FAM absorbance spectrum,³² which is maximal when G is in position 2 or 3 (and not position 1) relative to FAM. Previous computational studies of aromatic π-stacking interactions with phenol have shown that electron donation can significantly raise the pK_a of the phenolic hydrogen.³³ To examine this effect for the FAM phenol, we measured the pK_a values of free fluorescein and the test oligo series (Table 1). FAM conjugation to the pentauracil oligo raised the pK_a from 6.26 to 6.75 while G-substitution raised the pK_a higher still, reflecting the electron-donating character of G and stronger electrostatic interaction between FAM and the anionic oligonucleotide. We observed that pK_a was directly proportional to the red shift in the FAM absorption maximum, which links π-stacking to these two measurable ground-state phenomena. More significant dynamic quenching was observed at pH 5.0 than at pH 8.0, evidenced by the oligos with a distal G giving the same τ_pH8 value as FAM-UUUUU. In all cases, the single-stranded 5mer test oligos exhibiting the highest fluorescence dynamic range gave the highest QE values at pH 5 and became substantially less quenched at pH 7.5, which we attribute primarily to differences in dynamic quenching. Thus, we designed sMiRpH-2 probes directed against the cytoplasmic mRNA targets GAPDH, HPRT1, and SOD1 (Table 1 and Table S1), by incorporating a single G two nucleotides away from the 3′-FAM. We used 2′-OMe RNA in the sense strand to prevent its degradation by nucleases. Both strands incorporated two overhanging 3′-terminal DNA nucleotides, a feature retained from sMiRpH-1 (Scheme 1). On the sense strand, the included dG base increases the DR of FAM¹⁷ and the overhang generally protects the construct from exonucleases.³⁴⁻³⁶ In buffer, we observed fluorescence DR (between pH 5.0 and 7.5) of 11.7-, 11.3-, and 18.9-fold for the sMiRpH-2 probes directed against GAPDH, HPRT1, and SOD1 (Table 1, Figures S2–S10). At pH 5, the anti-GAPDH, anti-HPRT1, and anti-SOD1 sMiRpH-2 probes are respectively 26%, 41%, and 69% quenched with red shifts of 3, 3, and 4 nm (Table 1, Tables S5 and S6). To gain more insight into the effects of FAM-G π–π stacking, we measured the pK_a values for the sMiRpH-2 constructs directed against GAPDH, HPRT1, and SOD1: pK_a = 6.62, 6.62, and 6.35, respectively. These data reflect the greater acidity of the FAM phenol in the sMiRpH-2 probes compared with all the FAM-labeled 5mer test oligos (Table 1). A recent NMR study of local electrostatic potentials surrounding double-stranded oligos found considerably less negative potentials at the 5′- and 3′-termini, which was attributed to the lower phosphate density compared to positions near the middle of the double-strand.³⁷ The contribution of buffer cations to charge neutralization is supported by earlier experimental work on linear double-stranded DNA utilizing an intercalating electron spin resonance (ESR) probe, which found a surface potential corresponding to just 14% of the canonical phosphate charge density.³⁸ In our system, positioning the FAM-G interaction near one terminus of the siRNA duplex should allow stoichiometric binding between the terminal phosphates and Na⁺, thereby neutralizing charge within the FAM microenvironment and lowering the pK_a. As seen for the SOD1-probe, this can lead to a FAM pK_a (6.35) that is essentially identical with free fluorescein in buffer, and this significantly enhances the probe sensitivity within the desired acidic pH range. Native PAGE analysis of the sMiRpH-2 probes and their component strands confirmed intact duplexes (Figure S11) and thermal denaturation analysis revealed melting temperature (T_m) of 92.1, 92.9, and 69.8 °C for the GAPDH-probe, HPRT1-probe, and SOD1-probe (Table S1, Figures S12–S14). The lower T_m for the SOD1-probe is likely due to its lower GC-content. Notably, for the sMiRpH-2 probes, there was no red-shifted FAM absorbance at pH 7.5, unlike the comparable 5mer test oligos. This suggests that modulation of DR in sMiRpH-2 arises from significant FAM-G π–π stacking and static quenching at pH 5.0 relative to pH 7.5.³⁹ Elucidating the principles governing static vs dynamic G quenching in oligonucleotides with different structural motifs and at varying pH values will be the subject of future studies.

Table 1. Fluorescence Dynamic Range (DR) from pH 5.0 to pH 7.5, Acid Dissociation Constants for FAM (pK_a), Quenching Efficiencies (QE) Relative to FAM-UUUUU,²² Fluorescence Lifetimes (τ), and Red Shift in λ_max of the Absorbance Spectrum (RS) at pH 5.0 and pH 8.0 for Test 5′-FAM-Labeled Oligonucleotides and sMiRpH-2 Probes.

oligonucleotide	DR (fold)	pKa	QE (pH 5.0)	QE (pH 7.5)	τpH5(ns)	τpH8(ns)	RS (pH 5, nm)	RS (pH 8.0, nm)
FAM-UUUUU	8.7	6.75	0	0	3.7	3.9	0	0
FAM-GUUUU	12.4	6.86	80	71	3.2	3.4	1	2
FAM-UGUUU	16.9	6.90	78	57	3.3	3.8	2	2
FAM-UUGUU	13.1	6.96	59	39	3.2	4.0	2	1
FAM-UUUGU	11.0	6.88	52	40	3.3	4.0	1	0
FAM-UUUUG	10.9	6.73	44	30	3.3	4.0	0	1
FAM-GGUUU	21.1	6.96	90	77	2.7	3.5	3	3
FAM-UGGUU	19.9	7.09	90	77	2.6	3.7	6	5
FAM-GUGUG	18.8	6.95	86	70	2.9	3.4	3	3
FAM-UUUGG	11.7	6.74	52	36	3.2	3.9	0	1
anti-GAPDHa	11.7	6.62	26	–24	n.d.	n.d.	3	0
anti-HPRT1b	11.3	6.62	41	–9.2	n.d.	n.d.	3	0
anti-SOD1c	18.9	6.35	69	–24	n.d.	n.d	4	0

^aSense: 5′-UGA CCU CAA CUA CAU GGU UUA CAdG dT-FAM-3′; antisense: 3′-Cy5 dT dAAC UGG AGU UGA UGU ACC AAA UGU-Phos-5′.

^bSense: 5′-AGA GCU AUU GUA AUG ACC AGU CAdG dT-FAM 3′; antisense: 3-Cy5 dC dTUC UCG AUA ACA UUA CUG GUC AGU-Phos-5′.

^cSense: 5′-GUA UUA AAC UUG UCA GAA UUU CUdG dT-FAM-3′; antisense: 3′-Cy5-dC dCCA UAA UUU GAA CAG UCU UAA AGA-Phos-5′. Underlining indicates 2′-OMe RNA bases.

Quantification of Gene Expression Following Transfection with sMiRpH-2 Probes

siRNA transfections with Lipofectamine 2000 (L2000) or poly-l-lysine (PLL) model very efficient and very poor delivery scenarios, respectively. Using both vectors, sMiRpH-2 probes were transfected into HeLa cells alongside positive control siRNA and nontargeting control siRNA, all in biological triplicate. At 24 h post-transfection, the cells were harvested via trypsinization, lysed in TRIzol, and frozen at −80 °C prior to automated RNA extraction and relative gene expression analysis via the TaqMan RT-qPCR technique. Using the Livak method for ΔΔC_t calculation,⁴⁰ we quantified the gene expression for the experimental and positive controls relative to the negative control for each vector and gene target (Figure 1). Target silencing at the protein level was demonstrated via western blot analysis (Figures S15 and S16). As expected, the positive control siRNAs resulted in superb knockdown using L2000, with 18.1 ± 0.3% and 11.5 ± 0.2% relative expression of GAPDH and HPRT1, respectively. Less efficient knockdown of SOD1 (75.2 ± 1.1% relative expression) was achieved with the positive control siRNA and L2000, most likely due to the inadvertent selection of a suboptimal siRNA. The sMiRpH-2 probes achieved good knockdown with L2000, with expression values for GAPDH, HPRT1, and SOD1 of 45 ± 1%, 59 ± 1%, and 78 ± 1%, relative to negative control. While knockdown is less pronounced with the sMiRpH-2 probes, these findings unequivocally demonstrate biological activity despite ribose 2′-O-methylation of the sense strands and the 3′-dyes attached to both strands. Using PLL, transfection with the positive controls resulted in 90 ± 1%, 107 ± 4%, and 102 ± 2% relative expression of GAPDH, HPRT1, and SOD1, respectively. Numbers greater than 100% indicate conditions where the expression was higher than the negative scramble controls. PLL-mediated transfection of the sMiRpH-2 probes gave relative expression values of 75 ± 1%, 64 ± 2%, and 100 ± 2%, respectively for GAPDH, HPRT1, and SOD1. When using PLL, superior knockdown with the sMiRpH-2 probes relative to that with the positive controls is not wholly unexpected. Due to its structure,^41,42 the FAM label is more likely to trigger the so-called “proton sponge” effect than the unmodified positive control siRNAs.⁴³⁻⁴⁶ Under these experimental conditions, sMiRpH-2 probes behave similarly to natural siRNA and therefore have the potential to both predict and explain the results of knockdown experiments in HeLa cells.

Figure 1.

RT-qPCR-derived expression ratios (n = 3 biological replicates each with n = 4 technical replicates) of GAPDH, HPRT1, and SOD1 mRNA in HeLa cells transfected with either targeted sMiRpH-2 (Probe) or siRNA positive control (PC), reported relative to negative control (NC). Error bars represent ±1 SE.

Intracellular pH and Uptake Are Predictive of Knockdown Success in HeLa Cells

Flow Cytometric Quantification of Intracellular pH and Uptake

For pH measurements, HeLa cells were grown in two 6-well plates. Nine wells were transfected with the sMiRpH-2 probe, while three were left untreated as blanks. The cells were harvested and pelleted 4 h post-transfection. Cells pelleted from six wells were resuspended in intracellular pH clamping buffer, while the remaining three pellets and the blanks were resuspended in PBS before analysis by flow cytometry. When transfected using L2000, the sMiRpH-2 probes targeted against GAPDH, HPRT1, and SOD1 reported intracellular pH values of 7.37 ± 0.03, 7.41 ± 0.01, and 7.27 ± 0.02, respectively (Figure 2D), consistent with efficient endosomal escape and high knockdown. Conversely, the PLL transfections yielded pH values of 5.44 ± 0.14, 5.16 ± 0.14, and 5.39 ± 0.11, which are consistent with endosomal entrapment and low knockdown (Figure 2D). Percent uptake was also derived from the flow cytometric data by using a gate derived from the FAM and Cy5 signals measured for the blanks (Figure 2E). L2000 transfection resulted in 93 ± 1%, 85 ± 2%, and 93 ± 1% uptakes of the sMiRpH-2 probe targeted against GAPDH, HPRT1, and SOD1, respectively. The analogous PLL transfections yielded respective uptake percentages of only 35 ± 2%, 39 ± 1%, and 48 ± 2%. The high and low uptake percentages are consistent with the observed intracellular pH and relative knockdown values.

Figure 2.

Workflow for the flow cytometric determination of intracellular pH and uptake. Transfected cells are harvested and transferred to intracellular pH clamping buffer (see Materials and Methods) prior to analysis, locking the intracellular pH to known values. The geometric mean values of FAM and Cy5 fluorescence are extracted from the dot plots (A) and used to construct a calibration curve (B) in which the FAM:Cy5 ratio is plotted as a function of known pH. The ratios determined from the experimental samples in PBS (C) are used in the fit equation generated from (B) to solve for pH. The gating strategy for percent uptake determination of the experimental samples is described in Materials and Methods. Intracellular pH (D) and uptake (E) of sMiRpH-2 probes in HeLa cells as a function of vector (n = 3, *p < 0.05, one-way ANOVA). Error bars represent ±1 SD. Dot plots (A and C) and calibration curve (B) are taken from the GAPDH, Lipofectamine 2000 condition.

Native PAGE Analysis of sMiRpH-2 Following Transfection into HeLa Cells

To verify the stability of the sMiRpH-2 probes under our experimental conditions, we transfected HeLa cells with the anti-GAPDH construct with L2000 and PLL in triplicate followed by native PAGE analysis of whole-cell lysates at 1, 2, and 4 h post-transfection (Figures S17–S19). Composite images of the FAM and Cy5 channels at each time point demonstrate that when delivered by L2000, the probe remains intact. There are no visible bands in either channel for the PLL conditions at any time point, reflecting the poor uptake observed by flow cytometric analysis.

Live Cell Imaging of Transfected HeLa Cells

The transfections described above were reproduced using HeLa cells in an 8-well chambered coverslip format before live cell imaging via confocal laser scanning microscopy. The micrographs obtained from cells transfected with the three sMiRpH-2 probes using L2000 (Figure 3) show Pearson coefficients for the GAPDH, HPRT1, and SOD1 conditions of 0.806, 0.719, and 0.726, respectively, indicating that FAM and Cy5 are not appreciably sequestered from each other. There are also signs of nuclear exclusion in the Cy5 channel with corresponding nuclear infiltration in the FAM channel. This follows the expectation that the RNA antisense strands hybridize to target mRNA in the cytoplasm. The propensity of 2′-OMe RNA oligos to enter the nucleus has been described in the literature.^25,47 The analogous PLL transfections (Figure S20) confirmed poor uptake and release of cargo, explaining the observations from the RT-qPCR, flow cytometry, and native PAGE experiments.

Figure 3.

Representative live cell confocal micrographs of HeLa cells transfected with sMiRpH-2 probes targeted against GAPDH, HPRT1, and SOD1 mRNA using Lipofectamine 2000 and incubated for 4 h at 37 °C and 5% CO₂. Images are representative examples shown as separate green (FAM) and red (Cy5) channels. White arrows indicate cells showing signs of nuclear exclusion and inclusion in the Cy5 and FAM channels, respectively.

Successful Ordering of Lipofection Conditions in SK-OV-3 Cells

We next sought to ascertain the platform’s ability to predict the outcomes of transfections in a more challenging epithelial cell line, SK-OV-3, which is an immortalized lineage of ovarian cancer cells used in oncologic research. Specialized reagents are required to achieve high transfection efficiencies in SK-OV-3 cells. Using L2000 and a SK-OV-3-specific liposomal vector called Avalanche, we delivered the sMiRpH-2 probe targeted against GAPDH to SK-OV-3 cells for RT-qPCR analysis at 24 h post-transfection as described, along with positive and negative control siRNA all in biological triplicate (Figure 4). For cells transfected with the positive control siRNA, the Avalanche and L2000 conditions resulted in 5.7 ± 0.1% and 40 ± 1% expression, respectively, relative to GAPDH. The cells receiving the sMiRpH-2 probe displayed 69 ± 1% and 92 ± 3% relative GAPDH expression for the Avalanche and L2000 conditions, mirroring the control siRNA results. Flow cytometric analysis following L2000 transfection with the sMiRpH-2 probe targeting GAPDH and 4 h incubation reported intracellular pH and uptake values of 7.20 ± 0.01 and 89 ± 1%; with Avalanche, the values were 7.43 ± 0.04 and 99 ± 1% (Figure 5). As seen with HeLa cells, the SK-OV-3 live cell imaging corroborates the quantitation of the relative gene expression, intracellular pH, and uptake (Figure 6). The Avalanche reagent results in robust and diffuse release of the payload while L2000 results in less apparent release and more punctate signal. The Pearson correlation coefficients for the FAM and Cy5 channels in the Avalanche and L2000 conditions respectively were 0.751 and 0.727, indicating colocalized signal as in the HeLa experiments. We conclude that variable GAPDH expression arises from differences in the cargo release rather than uptake.

Figure 4.

RT-qPCR-derived expression ratios (n = 3 biological replicates each with n = 4 technical replicates) of GAPDH mRNA in SK-OV-3 cells transfected with either targeted sMiRpH-2 (Probe) or siRNA positive control (PC), reported relative to negative control (NC). Error bars represent ±1 SE.

Figure 5.

Intracellular pH (A) and uptake (B) of sMiRpH-2 duplexes in SK-OV-3 cells as a function of vector (n = 3, p < 0.05, one-way ANOVA). Error bars represent ±1 SD.

Figure 6.

Representative live cell confocal micrographs of SK-OV-3 cells transfected with the sMiRpH-2 probe targeted against GAPDH mRNA using Lipofectamine 2000 and Avalanche, both incubated for 4 h at 37 °C and 5% CO₂. Images are representative micrographs shown as separate green (FAM) and red (Cy5) channels as well as a composite merge.

Discussion

To empower the sMiRpH platform for guiding functional siRNA delivery, we studied the pH-dependent fluorescence behavior of an expanded panel of single-stranded 5′-FAM-labeled 5mer test oligos (Table 1), recording and calculating the fluorescence dynamic range, quenching efficiency, fluorescence lifetime, and red-shifting of the UV/vis spectra as a function of G position under acidic and basic conditions. We attributed the static quenching to a π–π stacking interaction between FAM and G. We learned that placing a single G just two bases away from FAM led to a large increase in fluorescence dynamic range, and this feature was confirmed in all sMiRpH-2 probes by placing the quenching G nucleotide within a 3′-terminal overhang. The sMiRpH-2 probes exhibited no stacking interactions under neutral/basic conditions and significant stacking in acid, as evidenced by quenching efficiency and red-shifting. The placement of FAM and the quenching G within an overhang lowered the pK_a values of the sMiRpH-2 series, with the anti-SOD1 probe registering a low pK_a of 6.35, which enhanced its sensitivity for physiological pH measurements. This is in agreement with previous experimental data demonstrating lower surface charge density and less negative electrostatic potential at the termini of double-stranded oligonucleotides.^37,38 Our findings provide additional experimental support for Manning’s theory of ion condensation, which predicts that the charge density of highly charged polyelectrolytes such as nucleic acids is reduced to approximately 20% of its theoretical maximum.^48,49 Once the counterion concentration exceeds the charge on the polyelectrolyte, the magnitude of the surface potential reduction is independent of the ionic strength. RT-qPCR analysis (Figure 1) shows that the sMiRpH-2 probes targeted against GAPDH, HPRT1, and SOD1 demonstrate biological activity in HeLa cells that is commensurate with the chosen delivery method, so they are viable surrogates for their natural siRNA analogs and are stable in cells under the conditions tested. Comparison of gene knockdown results with flow cytometry (Figure 2) data demonstrated successful ordering of transfection outcomes based on intracellular pH and uptake, which was corroborated by live cell imaging (Figure 3). Following endosomal escape, the strands comprising the sMiRpH-2 probes are separated during RISC engagement. This separation does not appear to affect the accuracy of the intracellular pH measurement under the tested conditions. As a control, we delivered sMiRpH-1 (Scheme 1),¹⁷ which has both dyes on the sense strand, to HeLa cells using Lipofectamine 2000 and PLL and quantified pH by flow cytometry after 4 h incubation at 37 °C (Figure S21). The L2000 and PLL conditions exhibited intracellular pH values of 7.12 ± 0.30 and 5.74 ± 0.30, respectively, which are congruent with those observed for the sMiRpH-2 probe series (Figure 2D). We note that the stability of the sense strand and its linkage to FAM under our experimental conditions do not preclude the premature cleavage of Cy5 from the antisense strand. That said, the fluorescence properties of Cy5 are unlikely to change appreciably in this event. In fact, a recent report of fluorescently labeled DNA nanostructures demonstrated that the intracellular behavior and signal of Cy5 was so blind to degradation as to be misleading in the absence of other controls.⁵⁰ These observations strongly suggest that the sMiRpH-2 probes are siRNA-mimetic with respect to their intracellular trafficking as well as their structure and ability to fruitfully engage with the RISC, further validating their use as predictive tools. In SK-OV-3 cells, we demonstrated a small but statistically significant difference in GAPDH knockdown efficacy when using two similar commercial lipofection reagents (Figure 4). Analogous flow cytometry (Figure 5) and live cell imaging experiments (Figure 6) likewise demonstrated the successful relative ordering of transfection outcomes based on the same parameters used for the HeLa experiments despite the use of similar delivery methods in the SK-OV-3 experiments. Taken together, these findings underscore the potential to use sMiRpH-2 probes to predict and explain the relative success of siRNA knockdown experiments based on straightforward, discrete parameters after significantly less incubation time post-transfection than is required to measure gene knockdown. Our results highlight the utility of sMiRpH-2 probes in the screening of nucleic acid delivery vectors. More broadly, we provide evidence of FAM-G ground-state π–π stacking interactions and excited-state collisional quenching. The relative contributions are challenging to quantify and vary considerably between different oligonucleotide sequences and secondary structures. These processes contribute to the variable FAM fluorescence intensity that has been observed for many FAM-labeled oligonucleotides in the literature.

Materials and Methods

Sequences and Sources of Oligonucleotides

The name, sequence(s), and source of each oligonucleotide referenced herein are provided as a table in the SI (Table S4).

Acquisition of Test Constructs, RNA Antisense Oligonucleotides, and Control siRNAs

Ribose 2′-O-methylated test oligonucleotides (Tables 1, S1) were purchased from Integrated DNA Technologies (IDT) (Coralville, IA) with the RP-HPLC option for purification. GAPDH-antisense, HPRT1-antisense, and SOD1-antisense RNA oligonucleotides were also purchased from IDT, but with the RNase-free HPLC option. Positive control siRNAs against GAPDH, HPRT1, and SOD1 were purchased from IDT and used without further purification. Negative control siRNA (Qiagen #1027310) was used as provided.

Synthesis of FAM-Labeled 2′-OMe RNA Sense Strands

Synthesis of GAPDH-sense, HPRT1-sense, and SOD1-sense 2′-OMe RNA oligomers proceeded as follows. All phosphoramidites and supports were purchased from Glen Research (Sterling, VA). The oligonucleotides were synthesized by solid-phase phosphoramidite chemistry on an ABI 394 synthesizer utilizing UltraMild 2′-OMe RNA phosphoramidites and a 3′-(6-fluorescein) CPG (20–2964–41) on a 1.0 μmol scale as previously described¹⁷ and purified via reverse-phase (RP) HPLC (Agilent 1260 Infinity II) on a C18 column at 1.0 mL/min and 40 °C. A gradient of increasing acetonitrile (B) in 0.1 M TEAA (A) was employed for purification (B starting at 10%, reaching 20% at 10 min, 17.5% at 25 min, 80% for 28–32 min, and 10% for 35–45 min). The product eluted at approximately 22 min based on monitoring absorbance of nucleobases (254 nm) and FAM (505 nm). Eluted fractions were combined and concentrated via vacuum centrifugation and buffer-exchanged with RNase-free water. Using the same method, the concentrate was subjected to analytical RP-HPLC to confirm purity (≥95%). Absorbances at 260 nm were measured using an Agilent Cary UV/vis spectrophotometer, and concentrations were determined using the Beer-Lambert law and extinction coefficients estimated using IDT’s free OligoAnalyzer tool. Yields and recovery percentages were calculated based on concentration and a starting scale of 1.0 μmol. Oligonucleotide mass was determined by Novatia, LLC using an LCMS system with electrospray ionization (Oligo HTCS). Analytical HPLC traces, purity estimates, yields, recoveries, and mass determinations are reported in the SI (Figures S22–S27, Table S5).

Formation of sMiRpH-2 Probes

Sense and antisense strands were either reconstituted or diluted to a concentration of 200 μM in 1X STE buffer, pH 8.0 (Fisher Scientific). To form duplexes, the sense and corresponding antisense strands were mixed in equimolar ratios. The resulting mixtures were heated in a 95 °C water bath for 5 min, cooled to RT on the bench for 45 min, and held on ice for 15 min before storing at −20 °C.

Gel-Based Duplex Integrity Assay

To separate dilutions of 10X BlueJuice loading buffer (Invitrogen) (1.0 μL) in RNase-free water (9.0 μL) was added 1.0 μL of GADPH-duplex, GAPDH-sense, GAPDH-antisense, HPRT1-duplex, HPRT1-sense, HPRT1-antisense, SOD1-duplex, SOD1-sense, and SOD1-antisense (each 10 μM in STE buffer). The diluted oligos were thoroughly mixed and subjected to native PAGE (15% acrylamide) for 60 min on ice at 150 V. The gel was imaged (Figure S11) for FAM and Cy5 fluorescence on a Typhoon FLA 7000 laser scanner (General Electric).

Melting Point Determination

Samples of the GAPDH-duplex, HPRT1-duplex, and SOD1-duplex were prepared at 4.0 μM in 1X STE buffer (100 μL) and transferred to quartz cuvettes with a 1.0 cm path length. Mineral oil (50 μL) was floated on top of each sample to prevent solvent evaporation during the analysis. The thermal stability of the duplexes was assessed by using an Agilent Cary UV/vis spectrophotometer with a multicell Peltier temperature controller. Samples were heated at 0.5 °C/min from 75 to 105 °C (GAPDH-duplex and HPRT1-duplex) or 60 to 80 °C (SOD1-duplex), held at 105 or 80 °C for 2 min, and then cooled at 0.5 °C/min back to 75 or 60 °C. Absorbance at 260 nm was measured every 0.2 °C for all samples. Average T_m values were assigned to each transition using numerical first-derivative analysis (Figures S12–S14).

Measurement of Fluorescence Response as a Function of pH

An aqueous buffer (PBesque) consisting of NaCl (138 mM), NaOAc (10 mM), MES (10 mM), Na₂HPO₄ (10 mM), TRIS (10 mM), and NaHCO₃ (10 mM) was prepared according to the method of Lavis et al.⁵¹ Small aliquots of PBesque were removed and adjusted with HCl(aq) or NaOH(aq) to make a series of buffers at pH 4.5, 4.9, 5.4, 5.9, 6.5, 6.9, 7.4, 7.7, and 8.0. Aliquots (2.0 μL) of oligo or duplex (50 μM) were pipetted rowwise in quadruplicate for each pH value into a black flat clear bottom 96-well plate (Corning #3615). A 100 μL volume of buffer per well (including one blank well per row) was aliquoted using a multichannel pipet. Fluorescence intensity spectra were collected on a Tecan Infinite M1000 Pro microplate reader as follows. For oligos containing only FAM and for the free dye, the samples were excited at 488 nm detecting emission from 500 to 600 nm in 1.0 nm increments. For oligos containing both FAM and Cy5, the samples were excited at 488 and 633 nm detecting emission from 500–700 nm and 650–800 nm, respectively, in 1.0 nm increments. For all experiments, excitation and emission bandwidths were set to 5.0 nm, PMT voltage was set using the highest-pH sample as a reference (or kept constant when direct comparisons were required), fluorescence was measured from the bottom of the plate, and flash frequency was set to 400 Hz with 100 flashes per well.

Analysis of Fluorescence Data

The quadruplicate emission spectra for each pH condition were averaged pointwise into a single spectrum with a standard deviation at each emission wavelength (Figures S2–S3, S5–S6, and S8–S9). To obtain integrated intensities, FAM and direct Cy5 spectra were integrated from 500 to 600 and 650 to 800 nm, respectively, using the trapezoidal approximation. Integrated fluorescence intensity was plotted as a function of pH for pK_a calculations (Figures S64–S77), which were computed by nonlinear generalized reduced gradient (GRG) fitting as previously described using F_total = F_4.5 + (F_8.0 – F_4.5)/(1 + 10^{pK_a – pH}).⁵¹ Relative fold change (RFC) plots (Figures S4, S7, and S10) were constructed by plotting the integrated FAM and/or Cy5 signal at a given pH relative to that at pH 4.5. For all G-substituted 5mer test oligos, we computed a quenching efficiency (QE) by calculating the quantity 100 – (100 × F/F₀) at pH 5 and 7.5, where F₀ is the integrated fluorescence intensity recorded for the FAM-UUUUU oligo.²²

Western Blot Analysis of HeLa Cells Transfected with sMiRpH-2 Probes

HeLa cells were seeded at 1 × 10⁵ cells per well in a six-well plate 24 h prior to the experiment. Using Lipofectamine 2000 (Invitrogen) according to manufacturer’s instructions, the cells were transfected with 75 pmol probe (in triplicate), unlabeled positive control siRNA (duplicate), and a scramble siRNA sequence (Qiagen #1027310). After incubating for 72 h at 37 °C and 5% CO₂, the cells were lysed with 1× RIPA buffer (10× concentrate from Cell Signaling Technology) and treated with one cOmplete Mini EDTA-free protease inhibitor cocktail tablet (Millipore Sigma). The lysates were sonicated on ice three times each for 15 s before centrifugation at 16000g for 30 min at 4 °C. The total protein concentration of each lysate was quantified by DC Assay (Bio-Rad) before SDS-PAGE. Along with a MagicMark XP Western Protein Standard chemiluminescent ladder (Life Technologies), 30 μg worth of total protein from each sample was denatured, loaded into a 4–15% TGX gel (Bio-Rad), and subjected to electrophoresis at 85 V for 15 min to stack followed by 130 V for 50 min on ice to resolve. The gel was transferred to a PVDF membrane using the wet transfer technique (35 V at 4 °C overnight). To detect GAPDH and β-actin, the membrane was blocked for 1 h at RT in 15 mL 5% BSA in 1× tris-buffered saline with 0.1% Tween 20 (TBST) with gentle agitation. Then, mouse anti-GAPDH (Abcam #ab8245) and mouse anti-β-actin (Abcam #ab8226) were added 1:1000 directly into the blocking buffer and gently rocked overnight at 4 °C. To detect HPRT1 and β-actin, the membrane was Ponceau stained and then cut to separate the rows containing the two targets. The piece containing HPRT1 was blocked for 1 h at RT in 15 mL TBSTM [1× TBST containing 5% w/v nonfat dry milk (Lab Scientific bioKEMIX, Inc.)] while the piece containing β-actin was blocked in 5% BSA, both with gentle agitation. Then, rabbit anti-HPRT1 (Abcam #ab133242) and mouse anti-β-actin (Abcam #ab8226) primary antibodies were added to the respective blocking buffers to concentrations of 1:1000 and incubated overnight at 4 °C with gentle agitation. The following day, the blots were thoroughly rinsed and washed with 1× TBST before incubation with HRP-tagged goat antirabbit IgG (Abcam #ab6721) or goat antimouse (Abcam #ab205719) secondary antibody 1:3000 in the same blocking buffer as the primary for 2 h at RT with gentle agitation. Following a final thorough rinse and wash with 1× TBST, the blot was developed using the Super Signal ECL Kit (ThermoFisher #34075) before imaging on the ChemiDoc system (Bio-Rad). The blots are shown in Figures S1, S15, and S16.

General Cell Culture Methods

HeLa (CCL-2) and SK-OV-3 (HTB-77) adherent cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA). HeLa was cultured and maintained in a growth medium consisting of Dulbecco’s modified Eagle medium (DMEM) (Gibco) containing 10% v/v fetal bovine serum (FBS) (Gibco) and 1% v/v PenStrep (Gibco). SK-OV-3 cells were cultured in McCoy’s Modified 5A Medium (Gibco) supplemented with 10% v/v FBS, 1% v/v MEM nonessential amino acids (NEAA) (Gibco), 1% v/v l-glutamine (Gibco), and 1% v/v PenStrep. Both cell types were incubated at 37 °C and 5% CO₂, passaged with 0.25% Trypsin-EDTA (Gibco) upon reaching 90–95% confluence, and maintained for a maximum of 20 passages.

General Parameters for Transfection of HeLa and SK-OV-3 Cells

Cells were grown to 85–90% confluence for all experiments. Prior to transfection, cells were washed twice with warm PBS (without Mg²⁺ or Ca²⁺) before immersion in warm serum- and antibiotic-free DMEM (HeLa) or McCoy’s (SK-OV-3). HeLa cells were transfected using Lipofectamine 2000 (Invitrogen) and poly-l-lysine (PLL) at final oligo concentrations of 33 and 150 nM, respectively. SK-OV-3 cells were transfected using Avalanche (EZ Biosystems, #EZT-SKOV-1) and Lipofectamine 2000 at final oligo concentrations of 37.5 and 33 nM, respectively. OptiMEM Reduced Serum Media (Gibco) was used as the diluent for oligo and vector in all transfections involving Avalanche and Lipofectamine 2000, while sterile HEPES (10 mM) was used for those involving PLL. HeLa cells were transfected with GAPDH-duplex, HPRT1-duplex, and SOD1-duplex, while SK-OV-3 cells were only transfected with GAPDH-duplex.

Transfection of HeLa and SK-OV-3 Cells for Flow Cytometry and Native PAGE

For each experiment, two six-well plates were seeded with HeLa and SK-OV-3 cells 24 and 48 h prior to transfection, respectively, to contain 2 × 10⁶ cells in 2 mL of complete growth medium. For HeLa, 7.5 μL of Lipofectamine was paired with 75 pmol of oligo per well, and 5 μg of PLL was mixed with 345 pmol of oligo per well. For SK-OV-3, 1.7 μL of Avalanche was paired with 75 pmol of oligo per well, and 7.5 μL of Lipofectamine 2000 was paired with 75 pmol of oligo per well. Transfected cells incubated for 4 h at 37 °C and 5% CO₂. In all experiments, nine wells were transfected with oligo, and three were transfected with a vehicle-only control. The SK-OV-3 cells were centrifuged for 5 min at 300g immediately following transfection.

For native PAGE analysis, only HeLa cells were transfected with the sMiRpH-2 probe directed against GAPDH, and L2000 and PLL transfections were performed in triplicate. Transfected cells were incubated at 37 °C and 5% CO₂ for 1, 2, and 4 h.

Determination of sMiRpH-2 Probe Integrity in HeLa Cells by Native PAGE

Following incubation and aspiration of transfection media, the cells were washed with ice-cold PBS, gently scraped into 1 mL of PBS on ice, and carefully transferred to prechilled 1.5 mL microcentrifuge tubes. The cells were pelleted via centrifugation at 400g and 4 °C, the supernatant was aspirated, and the cells were gently lysed via trituration in 80 μL ice-cold 0.1% NP-40 in PBS treated with one cOmplete Mini protease inhibitor cocktail tablet containing EDTA. Lysed cell suspensions (20 μL) and 10 μL control samples of sMiRpH-2 duplex, sense, and antisense (all 1 μM) were mixed with equal volumes of Gel Loading Buffer II (Invitrogen). All samples were subjected to native PAGE (15% acrylamide) for 60 min on ice at 150 V. The gels were imaged (Figures S17 and S19) for FAM and Cy5 fluorescence on a Typhoon FLA 7000 laser scanner (General Electric).

Flow Cytometric Analysis of HeLa and SK-OV-3 Cells Transfected with sMiRpH-2 and sMiRpH-1 Probes

At 4 h post-transfection, cells were harvested from 6-well plates using 0.25% trypsin-EDTA (200 μL/well) and then pelleted in six separate sterile microcentrifuge tubes. The blanks and three wells’ worth of transfected cells were resuspended in 1.0 mL of ice-cold PBS while the transfected cells in the remaining six vials were resuspended in 1.0 mL each of six ice-cold intracellular pH clamping buffers (pH = 4.54, 5.31, 5.95, 6.61, 6.90, 7.45). These buffers were prepared by mixing 50 mM HEPES (pH 7.5) and 50 mM MES (pH 5.0), each also containing 50 mM NaCl, 30 mM ammonium acetate, and 40 mM sodium azide.⁵² The cells in each vial were washed by pelleting and resuspending them once more in the same ice-cold buffers. The contents of each vial were transferred to labeled 5 mL polystyrene round-bottom tubes (Falcon) before transporting them on ice to the flow cytometry lab for analysis as previously described.¹⁷ Dot plots and calibration curves for all experiments are shown in the SI (Figures S28–S35). For cell uptake measurements, another gate was constructed by using the blank and applied to the populations previously gated for live cells and singlets. The lower bounds were determined from the sum of the median fluorescence intensity and the robust standard deviation from each channel. The calculation for robust standard deviation uses 68.26% of events around the median and sets an upper and lower range. The reported value is less affected by outliers and is equal to (upper range + lower range)/2. The uptake gate had no upper bound.

The experiments with sMiRpH-1¹⁷ (Figure S21) only used HeLa cells and featured experimental conditions in duplicate, four calibration conditions (pH = 4.97, 5.82, 6.58, and 7.47), and no uptake gates.

Transfection of HeLa and SK-OV-3 Cells for Live Cell Imaging

For each experiment, one collagen I-coated eight-well chambered coverslip (Ibidi no. 80809) was seeded with 5 × 10⁵ HeLa and SK-OV-3 cells 24 and 48 h prior to transfection, respectively, in 200 μL of complete growth medium. For HeLa, 0.23 μL Lipofectamine was paired with 6.6 pmol oligo per well, and 60 ng of PLL was mixed with 30 pmol of oligo per well. For SK-OV-3, 0.41 μL of Avalanche (1:5 in sterile H₂O) was paired with 7.5 pmol of oligo per well and 0.23 μL of Lipofectamine 2000 was paired with 6.6 pmol of oligo per well. Transfected cells were incubated for 4 h at 37 °C and 5% CO₂. In all experiments, six wells were transfected with oligo, and two were transfected with a vehicle-only control. The SK-OV-3 cells were centrifuged for 5 min at 300g immediately following transfection.

Live Cell Imaging by Confocal Laser Scanning Microscopy

Transfected cells in live cell imaging buffer (DMEM + 20 mM HEPES phenol red) were imaged on an Olympus FV1000 laser scanning confocal microscope equipped with a UPLFLN 40X oil immersion objective (NA = 1.30) and a 1 Airy disk pinhole. For each condition, the FAM (500–600 nm bandpass filter) and Cy5 (650 nm long-pass filter) channels were acquired simultaneously with a differential interference contrast (DIC) image, and two fields of view were acquired for each well. FAM and Cy5 were excited by a 488 and 633 nm laser set to 20% power. Photomultiplier tube (PMT) voltage also remained constant for all of the conditions. Each micrograph was acquired at 512 × 512 pixels with a scan rate of 12.5 μs/pixel. All images shown were created using the image processing package Fiji.⁵³ Pearson correlation coefficients for the FAM and Cy5 channels were determined using JaCoP (Just Another Colocalization Plugin)⁵⁴ in Fiji.

Transfection of HeLa and SK-OV-3 Cells for RT-qPCR

For each experiment, nine 6 cm dishes were seeded with HeLa and SK-OV-3 cells 24 and 48 h prior to transfection, respectively, to contain 3 × 10⁶ cells in 5 mL of complete growth medium. For HeLa, 16.7 μL of Lipofectamine was paired with 167 pmol oligo per dish, and 12.2 μg PLL of was mixed with 844.4 pmol oligo per dish. For SK-OV-3, 4.8 μL of Avalanche was paired with 220 pmol oligo per dish and 16.7 μL of Lipofectamine 2000 was paired with 167 pmol oligo per dish. Transfected cells were incubated for 24 h at 37 °C and 5% CO₂. In all experiments, cells were transfected with the sMiRpH-2 duplex, positive control siRNA, and negative control siRNA in biological triplicate. For the Avalanche condition, the medium was replaced with complete growth medium at 5 h post-transfection.

RT-qPCR Analysis of Transfected HeLa and SK-OV-3 Cells

At the conclusion of the incubation period, the cells were washed with PBS, harvested with the aid of 600 μL of trypsin-EDTA (0.25%), and pelleted in sterile 1.5 mL microcentrifuge tubes (5 min at 400g). The cells were washed in PBS and pelleted again before the supernatant was aspirated and replaced with ice-cold TRIzol (700 μL, Zymo Research). Immediately after addition of TRIzol, the cells were lysed via vortex (1 min) before prompt storage at −80 °C, where they remained until transport to the Molecular Profiling Laboratory on dry ice. RNA extraction was performed using RNeasy Mini Kits on a Qiacube (both Qiagen). All samples were subjected to DNase treatment to remove residual gDNA. RNA concentration and quality were checked using Nanodrop (concentration, A₂₆₀/A₂₈₀, and A₂₆₀/A₂₃₀) and Bioanalyzer (electrophoresis) instruments. cDNA synthesis was carried out with 1.5 μg of isolated total RNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems #4368814) according to manufacturer’s instructions with thermocycler settings of 25 °C for 10 min, 37 °C for 2 h, 85 °C for 5 min, and 4 °C indefinitely. qPCR of cDNA was performed in quadruplicate on a 394-well plate using a QuantStudio 12K Flex system (Applied Biosystems #285880688), gene-specific TaqMan Gene Expression Assay, VIC kits (Applied Biosystems #4448491; GAPDH Hs02786624_g1, HPRT1 Hs02800695_m1, SOD1 Hs00533490_m1) and the TaqMan Fast Advanced Master Mix for qPCR (Applied Biosystems #4444964) according to manufacturer’s instructions. The presence of FAM labels on the sMiRpH-2 probes necessitated the use of VIC-based TaqMan kits. Thermocycler parameters were 95 °C for 20 s followed by 40 cycles of 95 °C for 1 s and 60 °C for 20 s. Gene expression of experimental and positive controls was calculated relative to negative controls using the ΔΔC_T method.⁴⁰ For experiments targeting GAPDH and SOD1, HPRT1 was used as the housekeeping gene control. When targeting HPRT1, GAPDH was the reference gene. RNA quality control data, amplification traces, and Real-Time PCR Data Essential Spreadsheet (RDES) files⁵⁵ for each experiment are available in the SI.

Measurement and Calculation of Absorbance Red Shifts

Triplicate samples of free fluorescein, ribose 2′-O-methylated test oligos, and sMiRpH-2 probes were prepared at 2.0 and 1.0 μM in PBesque buffer (pH 5.0 and 8.0, respectively) in disposable plastic cuvettes (Fisher Scientific) with a 1 cm path length after obtaining baseline spectra. The pH 5.0 value was selected as the minimum to avoid contributions from the nonfluorescent neutral and cationic forms of FAM to the absorbance spectra, while pH 8.0 was selected to isolate the phenolate form.⁵⁶ Experiments with sMiRpH-2 probes were done at 4 and 25 °C. λ_max values, corresponding A₄₈₈ values, and red shifts relative to FAM-UUUUU are reported in the SI (Tables S4, S6). Raw absorbance spectra from these experiments are included as Figures S36–S41.

Fluorescence Lifetime Measurements

Time correlated single photon counting (TCSPC) measurements of fluorescence lifetime decays for 1 μM samples of free fluorescein and ribose 2′-O-methylated test oligos in PBesque buffer (pH 5.0 and pH 8.0) were collected on a PTI Quantamaster 40 instrument using a pulsed LED with a maximum emission at 486 nm. The minimum pH value of 5.0 was chosen to lessen contributions from the FAM dark states to the observed lifetime, while pH 8.0 was chosen to isolate the dianion.⁵⁶ Fluorescence emission (30,000 counts) was collected at 520 nm for each sample with 20 nm slit widths. The instrument response function (IRF) (30,000 counts) was collected from a sample of colloidal silica (30% in H₂O) diluted 1:100 in H₂O at a wavelength of 486 nm with 20 nm slit widths. Data were analyzed using PTI’s FelixGX software (ver. 4.2.2) using a single exponential decay model. Decays, fits, reduced chi-squared values, and residuals are provided in the SI (Figures S42–S63).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.4c00545.

• Additional tables, figures, materials, and experimental methods; supplementary confocal micrographs and RT-qPCR amplification traces are included as appendices; RDES files for all RT-qPCR experiments performed for this study; RNA quality control results from samples analyzed by RT-qPCR (PDF)

Author Contributions

M.R.H. performed all experiments and analyzed all data except for Figures S17–S19, which were performed by L.L.V. M.R.H. and I.J.D. wrote the manuscript.

We thank the National Institutes of Health (grant R35-GM-131907 to IJD) for funding.

The authors declare no competing financial interest.