Multidimensional quantitative analysis of mRNA expression within intact vertebrate embryos

Abstract

For decades, in situ hybridization methods have been essential tools for studies of vertebrate development and disease, as they enable qualitative analyses of mRNA expression in an anatomical context. Quantitative mRNA analyses typically sacrifice the anatomy, relying on embryo microdissection, dissociation, cell sorting and/or homogenization. Here, we eliminate the trade-off between quantitation and anatomical context, using quantitative in situ hybridization chain reaction (qHCR) to perform accurate and precise relative quantitation of mRNA expression with subcellular resolution within whole-mount vertebrate embryos. Gene expression can be queried in two directions: read-out from anatomical space to expression space reveals co-expression relationships in selected regions of the specimen; conversely, read-in from multidimensional expression space to anatomical space reveals those anatomical locations in which selected gene co-expression relationships occur. As we demonstrate by examining gene circuits underlying somitogenesis, quantitative read-out and read-in analyses provide the strengths of flow cytometry expression analyses, but by preserving subcellular anatomical context, they enable bi-directional queries that open a new era for in situ hybridization.

Keywords: Multiplexed in situ hybridization; Quantitative in situ hybridization; Read-in; Read-out.

Fig. 1.

Quantitative in situ hybridization chain reaction (qHCR). (A) Two-stage protocol independent of the number of target mRNA species (Choi et al., 2014, 2016). Detection stage: DNA probes carrying DNA HCR initiators (I1 and I2) hybridize to target mRNAs and unused probes are washed from the sample. Amplification stage: metastable fluorophore-labeled DNA HCR hairpins (H1 and H2; green stars denote fluorophores) penetrate the sample without interacting; initiators trigger chain reactions in which H1 and H2 hairpins sequentially nucleate and open to assemble into tethered fluorescent amplification polymers; unused hairpins are washed from the sample. (B) Conceptual schematic: for subcellular voxels within whole-mount vertebrate embryos, HCR signal scales approximately linearly with mRNA abundance, enabling quantitative analysis of mRNA expression in an anatomical context.

Fig. 2.

Accuracy and precision assessed by redundant detection. (A) Each target mRNA is detected using two probe sets, each initiating an orthogonal and spectrally distinct HCR amplifier (red channel, Alexa 647; green channel, Alexa 546). (B) Two-channel redundant detection of four target mRNAs: desma, Gt(desma-citrine) and elavl3 in whole-mount zebrafish embryos (fixed 26 hpf); and Acta2 in a whole-mount mouse embryo (fixed E9.5). Confocal microscopy: 0.7×0.7 µm pixels (desma, citrine and elavl3) or 0.07×0.07 µm pixels (Acta2). (C) Highly correlated normalized signal (Pearson correlation coefficient, r) for 2×2×2 µm voxels in the selected regions of B. Accuracy: linear with zero intercept. Precision: scatter around the line. (D) Scatter as a function of voxel size for desma. See section S2.1 in the supplementary material for control experiments testing for potential systematic penetration and crowding effects; see section S2.2 in the supplementary material for an examination of the effect of voxel size and probe set size on quantitative precision; and see section S2.3 in the supplementary material for additional data.

Fig. 3.

Measuring a twofold difference in mRNA levels. (A) Two-channel imaging of citrine (red channel, Alexa 647) and desma (green channel, Alexa 546) target mRNAs in homozygous Gt(desma-citrine)^{ct122a/ct122a} embryos and heterozygous Gt(desma-citrine)^ct122a/+ embryos. Confocal microscopy: 0.7×0.7 µm pixels. Whole-mount zebrafish embryos fixed at 26 hpf. Depicted regions are analyzed in B-D. (B) Normalized signal for citrine (red) and desma (green) targets in homozygous and heterozygous embryos (mean±s.d. via uncertainty propagation, n=3 embryos). (C) Ratio of citrine target in homozygous versus heterozygous embryos (mean±s.d. via uncertainty propagation, n=3 embryos). (D) Normalized signal for 2×2×2 µm voxels within the selected regions of A. See section S2.4 in the supplementary material for additional data.

Fig. 4.

Quantitative read-out and read-in. (A) Four-channel quantitative image for four target mRNAs in a whole-mount zebrafish embryo. Confocal microscopy: 0.7×0.7 µm pixels, mean intensity over five focal planes. Embryo is fixed at 10 hpf. (B) Normalized expression profiles for four target mRNAs along a strip of interest (see A) crossing four somites (S7, S8, S9 and S10) and the presomitic mesoderm (PSM). (C) Read-out from a region of interest (see A) within a four-channel image (left) to pairwise expression scatter plots (right), revealing distinct expression clusters with different expression characteristics. Each point within an expression scatter plot represents normalized voxel intensities for a pair of target mRNAs. Voxel size: 2×2×6 µm. (D) Read-in from pairwise expression scatter plots (left) to a four-channel image (right), revealing the anatomical locations corresponding to four expression clusters of interest. Expression clusters selected in the her7-myod1 quadrant; cluster shading (lilac, yellow, orange, cyan) propagated to the other three quadrants. See section S2.5 in the supplementary material for additional data.

Fig. 5.

Quantitative snapshots of gene co-expression changes during somite formation and maturation. (A) Anatomical regions of interest within somites S7, S8, S9 and S10, and the presomitic mesoderm (PSM). (B) Expression scatter plots for four target mRNAs shaded by anatomical regions in A. (C) Subcircuit expression scatter plots. Amplitude of her1-her7 subcircuit versus amplitude of myod1-tpm3 subcircuit for the anatomical regions of A. x denotes normalized signal for each target mRNA. Confocal microscopy: mean intensity over five focal planes, 2×2×6 µm voxels. Embryo is fixed at 10 hpf. See section S2.6 in the supplementary material for additional data.

Fig. 6.

Work flow for quantitative read-out and read-in analyses using qHCR imaging. Step 0: Acquire and normalize data. Step 1: Read-out from anatomical space to expression space. Step 2: Read-in from expression space to anatomical space. If desired, steps 1 and 2 can be performed iteratively, moving back and forth between regions of interest in anatomical space and clusters of interest in expression space.