High-throughput evolution of near-infrared serotonin nanosensors

Abstract

Imaging neuromodulation with synthetic probes is an emerging technology for studying neurotransmission. However, most synthetic probes are developed through conjugation of fluorescent signal transducers to preexisting recognition moieties such as antibodies or receptors. We introduce a generic platform to evolve synthetic molecular recognition on the surface of near-infrared fluorescent single-wall carbon nanotube (SWCNT) signal transducers. We demonstrate evolution of molecular recognition toward neuromodulator serotonin generated from large libraries of ~6.9 × 10¹⁰ unique ssDNA sequences conjugated to SWCNTs. This probe is reversible and produces a ~200% fluorescence enhancement upon exposure to serotonin with a K _d = 6.3 μM, and shows selective responsivity over serotonin analogs, metabolites, and receptor-targeting drugs. Furthermore, this probe remains responsive and reversible upon repeat exposure to exogenous serotonin in the extracellular space of acute brain slices. Our results suggest that evolution of nanosensors could be generically implemented to develop other neuromodulator probes with synthetic molecular recognition.

Fig. 1. SELEC evolves ssDNA-SWCNT nanosensors for 5-HT molecular recognition.

(A) Schematic illustration of the SELEC evolution process. SWCNTs are mixed with a >10-fold mass excess of ssDNA with 100 μM 5-HT (experimental library) or alone (control library). (1) The mixture is sonicated to generate the ssDNA-SWCNT complex of either 5-HT–bound ssDNA-SWCNT (blue) or ssDNA-SWCNT (red). (2) Unbound ssDNAs are removed, and (3) bound ssDNAs are isolated from SWCNT by thermal desorption. (4) Isolated ssDNAs are amplified by PCR, and the ssDNA pool is prepared for the next selection round and (5) characterized by high-throughput sequencing. (B) Sequence frequency of the top 50 unique sequences in the sixth round of the experimental library, compared to their sequence prevalence in rounds 2 to 5. (C and D) Probability for each nucleotide at each position in the experimental library calculated from the top 200 sequences from (C) round 2 and (D) round 6. (E) MDS plot of trimer frequency table for the top 200 sequences in the experimental and control SELEC libraries at each round. Data are plotted on the first and second coordinates of MDS analysis. Contour lines are included for better visualization of mutual divergence between experimental and control SELEC libraries for the same round. Divergence at round 2 may be too small to be distinguished. (F and G) Palindromic trimer incidences of (GTG)/(TGT) at each nucleotide position for (F) experimental and (G) control libraries.

Fig. 2. Evolution of ssDNA-SWCNT demonstrates increased fluorescence sensitivity toward 5-HT.

(A) ΔF/F₀ upon addition of 100 μM 5-HT to top 10 ssDNA sequences from rounds 3 to 6 (R3 to R6) in experiment (Exp; blue circle) and control (Ctrl; red circle) SELEC groups (n = 3 trials). The most sensitive 5-HT nanosensor, E6#9, is indicated by a black dashed circle. ΔF/F₀ from initial (round “zero”) ssDNA-SWCNT library is represented by a dashed line. (B) Data from (A) are represented as a histogram. (C) Fluorescence spectra of E6#9 ssDNA-SWCNT before (black) and after (red) addition of 100 μM 5-HT. Inset: Schematic illustration of recognition of 5-HT by nIRHT nanosensor. a.u., arbitrary units. (D) ΔF/F₀ 5-HT concentration dependence of E6#9-SWCNT (blue) and C6#8-SWCNT (red). Error bars denote SD from n = 3 independent trials and may be too small to be distinguished in the graph. Experimental data are fitted with the Hill equation (solid trace).

Fig. 3. Validation and use of nIRHT 5-HT nanosensors under neurologically relevant conditions.

(A) 5-HT concentration–dependent ΔF/F₀ of nIRHT nanosensor in PBS (blue squares) and aCSF (green circles) fit with the Hill equation (solid trace). (B) ΔF/F₀ of nIRHT nanosensor upon exposure to 100 μM neurotransmitters and 5-HT metabolites [dopamine (DA), norepinephrine (NE), acetylcholine (ACh), γ-aminobutyric acid (GABA), tyrosine (Tyr), glutamate (Glu), and octopamine (Oct)]. Error bars are SD from n = 3 independent trials. (C) ΔF/F₀ of nIRHT nanosensor upon exposure to 100 μM 5-HT with and without 1 μM 5-HT receptor drugs fluoxetine (nonselective agonist), 3,4-MDMA (nonselective agonist), 25I-NMOMe (5-HT₂ agonist), and quetiapine (Que) (5-HT_1A agonist). Error bars are SD from n = 3 independent trials. ****P < 0.0001. n.s., nonsignificant differences in one-way analysis of variance (ANOVA). (D) Reversibility of immobilized nIRHT nanosensors on glass substrate upon exposure to 100 μM 5-HT. (E and F) nIR fluorescence images of the same field of view (E) before and (F) after addition of 100 μM 5-HT. (G) ΔF/F₀ of immobilized nanosensors upon exposure to 100 μM 5-HT (blue bar) and rinsed with PBS (black bar) in a flow chamber. (H) nIR imaging of dorsomedial striatum from mouse acute brain slice after nIRHT nanosensor loading into brain tissue. (I to K) ΔF/F₀ images of the same field of view of an acute striatal brain slice (I) before, (J) 3 min after exogenous addition of 100 μM 5-HT, and (K) 3 min after rinsed with continuous aCSF flow. Bright ΔF/F₀ hotspots following 5-HT bath application results from heterogeneous nIRHT slice labeling but does not correlate with endogenous neuromodulator responses observed in other studies (26).