Axially decoupled photo-stimulation and two photon readout (ADePT) for mapping functional connectivity of neural circuits

Reviewer #1 (Public review):

In this methods paper, the authors introduce a novel and innovative imaging approach for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photo-stimulation in different axial planes. This is a highly exciting technique that enables the axial decoupling of optical imaging of deep neural circuits from surface photo-stimulation of spatially precise (tens of micrometres) brain spots. This method builds on previous developments from the same laboratory, combining DMD-based patterned photo-stimulation with in vivo electrophysiological recordings. To my knowledge, this is the first instance in which patterned photo-stimulation has been combined and axially decoupled from two-photon (2P) imaging.

Beginning with a thorough characterisation of the optical resolution of the photo-stimulation system, the authors applied this method to the olfactory bulb (OB) network, in which sensory inputs are topographically organised at the surface of the OB and thus ideally suited to demonstrate the relevance of this approach. They first showed that this technique can be used to rapidly reveal connectivity patterns of OB output neurons and to identify sister mitral cells. In addition, they manipulated a specific glomerular inhibitory population and demonstrated that these neurons provide spatially heterogeneous long-range inhibition of OB output neurons, with differential effects on mitral and tufted cells (a result previously observed in a paper from the same lab: Banerjee et al., 2015, Neuron). Altogether, the data demonstrate that this technique is well-suited for high-throughput functional mapping of neural circuit properties. The results are compelling and illustrate both the significant advance represented by this method and its feasibility.

Despite my initial enthusiasm, there are several concerns in the present study that must be addressed in order to rule out confounding observations and to resolve remaining uncertainties regarding photo-stimulation resolution. These include the following:

(1) Spatial resolution: Although the authors provide convincing data on spatial resolution in vitro, several observations throughout the paper suggest that the effective photo-stimulation precision may be lower than initially reported. For instance, in Figure 2, the authors observe repeated responses in neighbouring glomeruli (e.g., glomeruli #3 & #5, #4 & #6). To what extent could light scattering along the X/Y/Z-axis above the targeted glomerulus recruit en passage axons, resulting in the inadvertent activation of multiple glomeruli?

A further observation concerns the presence of "inhibited" sister mitral cells (Figure 3). The authors claim this is reminiscent of the differential spike-timing reported between sister cells (Dwawale et al., 2010, Nat Neuro). However, observing both excitatory and inhibitory responses following stimulation of glutamatergic inputs is an altogether different matter, particularly given that sister mitral cells are reciprocally connected via gap junctions. This observation requires further clarification and raises serious questions about the effective resolution of the stimulation. Could the inhibited cell simply correspond to a non-sister mitral cell receiving disynaptic feed-forward inhibition? To verify sister cell identity, the authors could confirm that the predicted sister cells share a similar odour receptive field compared to randomly selected mitral cell pairs. In their previous study employing analogous DMD-based photo-stimulation (Dhawale et al., 2010, Nat. Neurosci.), sister mitral cells did not exhibit such opposite response profiles (firing rate correlation of ∼0.7 between sister cells). Could the authors verify that a comparable activity correlation is also observed among the sister cells identified using ADePT in the present study? In Figure S6, the authors show recordings and stimulation of the same neurons co-expressing GCaMP and ChR2. Applying this experimental design to the mitral/tufted cell population (using a Tbet-Cre mouse transduced in the OB with both GCaMP and Chrimson virus) would constitute a valuable control to clarify the nature of these "inhibited" sister cells.

An additional concern relates to the 21 out of 162 mitral cells that were activated by two distinct glomeruli - a finding that is incompatible with the established OB wiring diagram and that further challenges the claimed stimulation resolution.

A critical control experiment is also absent: in a Thy1-GCaMP6 mouse lacking any light-sensitive opsin, do the authors observe any unintended side effects of photo-stimulation?

Regarding sister cells (Figure 3), tufted cells are not analysed alongside mitral cells in this dataset, whereas this is elegantly performed in Figure 5 using the DAT+ model. Could the authors also demonstrate how the technique can reveal the complete family portrait of sister mitral and tufted cells?

(2) The authors have explored only a limited set of photo-stimulation parameters, primarily varying light intensity. They should present additional tests, such as varying the spot size (which appears to be arbitrarily fixed at 30-50 µm) and the z plane of stimulation. The level of activation can vary considerably: for example, in Figure 3a(iii), identical stimulations elicit responses of markedly different amplitudes (see glom#3 and #4). In Figure 2, 5 out of 15 glomeruli failed to respond - could the choice of z-plane account for this variability? The stimulation duration (50-150 ms) also appears somewhat arbitrary: can the authors demonstrate that the technique is compatible with finer temporal patterns (e.g., 10 Hz stimulation for 500 ms using 20 ms light pulses)? What are the spatiotemporal and axial scanning limits of this approach, and can two or three glomeruli be targeted simultaneously with temporally patterned stimulation?

(3) One particularly relevant application of this method would be to guide photo-stimulation based on prior functional measurements - for instance, by generating a photo-stimulation mask specifically targeting odour-responsive glomeruli. In the DAT-Cre × Thy1-GCaMP6 experiment shown in Figure 5e, which glomeruli are activated by a given odour, and how does this odor responsiveness influence the efficiency of DAT+ cell-mediated inhibition?

Reviewer #2 (Public review):

Summary:

In this manuscript, Koh and colleagues describe ADePT (Axially Decoupled Photo-stimulation and Two-photon Readout), a modular approach for combining patterned one-photon optogenetic stimulation with two-photon calcium imaging in independently controlled axial planes. The method relies on a digital micromirror device together with a motorized holographic diffuser to generate spatially confined stimulation patterns while imaging deeper neuronal populations. As proof-of-principle applications, the authors use the system to map excitatory and inhibitory functional connectivity in the mouse olfactory bulb by stimulating superficial glomerular circuits and recording responses from mitral and tufted cells in deeper layers.

This is a well-executed Tools and Resources manuscript. The technical implementation is described in considerable detail, the optical performance is systematically characterized, and the biological experiments provide convincing demonstrations of the types of circuit questions that can be addressed using the method.

Strengths:

The greatest strength of the manuscript is the comprehensive technical characterization of the optical system. The authors carefully benchmark the spatial resolution, axial confinement, registration accuracy, calibration procedure, and practical operating limits of the setup. I found the extensive optical benchmarking particularly helpful, as it gives readers a realistic sense of the operating regime and practical limitations of the approach.

Another strength is the high level of methodological transparency. The optical design, calibration procedures, stimulation strategies, and analysis pipeline are described in sufficient detail that an experienced laboratory could realistically evaluate whether the system is suitable for its own applications. This level of documentation is particularly appropriate for a Tools and Resources article.

A further strength is the clear positioning of ADePT relative to existing approaches. The authors are transparent about the trade-off between spatial resolution and implementation complexity: ADePT does not provide single-cell photostimulation, but offers flexible axial separation, a large stimulation field, and cellular-resolution two-photon readout in deeper planes without requiring a full holographic stimulation system. This defines a credible and potentially useful experimental niche.

The biological applications convincingly demonstrate the utility of ADePT. The experiments identifying sister mitral/tufted cells through selective glomerular stimulation and the mapping of heterogeneous inhibitory influences from DAT-positive interneurons illustrate the types of functional connectivity questions that become experimentally accessible with this approach. Importantly, the authors generally avoid overstating these biological findings and appropriately present them as proof-of-principle demonstrations of the technology.

Weaknesses:

The primary limitation is inherent to the method itself rather than the execution of the study. Because ADePT relies on one-photon patterned illumination, photo-stimulation remains restricted to relatively superficial structures and does not achieve single-cell spatial resolution. The authors appropriately acknowledge these constraints and clearly position the method within this operating regime. Consequently, ADePT occupies a useful niche for interrogating spatially organized functional units such as olfactory glomeruli or cortical barrels, rather than applications requiring single-cell precision or deeper tissue penetration.

Although the manuscript describes the approach as relatively simple and cost-effective, implementation still requires careful optical alignment, registration, calibration, and optimization. This does not diminish the value of the approach, but terms such as modular or accessible may better reflect the practical implementation than simple. Likewise, a brief bill of materials, approximate add-on cost, and indication of which components are essential versus substitutable would help prospective users assess the accessibility of the system.

Finally, the manuscript provides an impressive level of technical characterization, but much of the practical guidance for adopting the system is distributed across the Results and Discussion. Bringing together the principal limitations, recommended operating regime, expected calibration workflow, evidence for long-term alignment stability, and the circumstances in which ADePT is preferable to alternative approaches would further strengthen the manuscript as a community resource.

Author response:

eLife Assessment

This paper introduces a valuable optical method for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photostimulation in different axial planes. The evidence for effective optical separation of excitation and imaging is convincing, although the in vivo data in the olfactory bulb suggest potential confounding factors arising if stimulation not only affects cell bodies but also neuronal processes. The work will be of broad interest to neurobiologists working in circuit and systems neuroscience, as well as to specialists in optical microscopy.

We thank the reviewers for their constructive feedback. We are planning to address their concerns as detailed below. Specifically, in the revised manuscript, we will streamline and consolidate the text to include:

(1) An in-depth discussion of the awake recording results in the main text.

(2) Further discussion of light scattering, photo-stimulation specificity of targeting individual glomeruli and resolution.

(3) A summary (including also a table) of the operating regime and comparisons with alternative techniques for patterned photo-stimulation and imaging of the ensuing responses. We will highlight the advantages and constraints of the current implementation of ADePT. In particular, here we explored a small set of spatiotemporal parameters to understand the limits of our technique and provide a proof of principle of the strategy. These parameters can be varied further depending on the exact research question.

(4) Implementation considerations and technical guidelines for calibration and long-term stability of the rig for ADePT (i.e. ‘a how-to guide’).

(5) A bill of materials and estimated hardware costs.

Furthermore, we will provide additional controls and rephrase some of the statements in the text as suggested (e.g. replace ‘accessible’ with ‘simple’, etc.). We will correct the unfortunate grammatical errors, improve clarity of text, and update the references accordingly.

Public Reviews:

Reviewer #1 (Public review):

In this methods paper, the authors introduce a novel and innovative imaging approach for simultaneous in vivo multiphoton imaging of the mouse brain combined with DMD-based one-photon patterned photo-stimulation in different axial planes. This is a highly exciting technique that enables the axial decoupling of optical imaging of deep neural circuits from surface photo-stimulation of spatially precise (tens of micrometres) brain spots. This method builds on previous developments from the same laboratory, combining DMD-based patterned photo-stimulation with in vivo electrophysiological recordings. To my knowledge, this is the first instance in which patterned photo-stimulation has been combined and axially decoupled from two-photon (2P) imaging.

Beginning with a thorough characterisation of the optical resolution of the photo-stimulation system, the authors applied this method to the olfactory bulb (OB) network, in which sensory inputs are topographically organised at the surface of the OB and thus ideally suited to demonstrate the relevance of this approach. They first showed that this technique can be used to rapidly reveal connectivity patterns of OB output neurons and to identify sister mitral cells. In addition, they manipulated a specific glomerular inhibitory population and demonstrated that these neurons provide spatially heterogeneous long-range inhibition of OB output neurons, with differential effects on mitral and tufted cells (a result previously observed in a paper from the same lab: Banerjee et al., 2015, Neuron). Altogether, the data demonstrate that this technique is well-suited for high-throughput functional mapping of neural circuit properties. The results are compelling and illustrate both the significant advance represented by this method and its feasibility.

We thank the Reviewer for their constructive input.

Despite my initial enthusiasm, there are several concerns in the present study that must be addressed in order to rule out confounding observations and to resolve remaining uncertainties regarding photo-stimulation resolution. These include the following:

(1) Spatial resolution: Although the authors provide convincing data on spatial resolution in vitro, several observations throughout the paper suggest that the effective photo-stimulation precision may be lower than initially reported. For instance, in Figure 2, the authors observe repeated responses in neighbouring glomeruli (e.g., glomeruli #3 & #5, #4 & #6). To what extent could light scattering along the X/Y/Z-axis above the targeted glomerulus recruit en passage axons, resulting in the inadvertent activation of multiple glomeruli?

Indeed, we cannot rule out this possibility. Fibers of passage are a potential concern. This is why we systematically sample different light intensities and assess their impact on specificity of dendritic mitral and tufted cell responses within the glomerular layer (same axial-plane optical stimulation and imaging experiments, Fig. 2). We identify a range of intensities that on average result mostly in activation of the targeted glomeruli. Within the range of intensities used for identifying sister cells, >90% of responses were on the diagonal (targeted glomeruli) and ~5% pixels that cleared the signal significance criterion used were in off-target glomeruli, as stated in the text and quantified in Fig. 2f. In the revised manuscript, we will further clarify and expand on these points.

A further observation concerns the presence of "inhibited" sister mitral cells (Figure 3). The authors claim this is reminiscent of the differential spike-timing reported between sister cells (Dwawale et al., 2010, Nat Neuro). However, observing both excitatory and inhibitory responses following stimulation of glutamatergic inputs is an altogether different matter, particularly given that sister mitral cells are reciprocally connected via gap junctions. This observation requires further clarification and raises serious questions about the effective resolution of the stimulation. Could the inhibited cell simply correspond to a non-sister mitral cell receiving disynaptic feed-forward inhibition?

This is indeed what we think it is happening (i.e. disynaptic feed-forward inhibition as the Reviewer points out). We observe inhibition in some of the mitral cells in the field of imaging when we stimulate not their parent glomerulus, but other glomeruli in the neighbourhood. As the Reviewer points out, sister cells are connected via gap junctions, but they also receive inhibitory chemical synaptic inputs via their secondary (and primary dendrites) from other (not-their-parent) glomeruli mediated by numerous types of interneurons including the DAT+/GABAergic (a.k.a. superficial short axon cells) and granule cells. Our data is consistent with differential inhibitory input from other glomeruli on sister cells getting input from the same parent glomerulus. As it appears that we failed to present this point clearly in the initial submission, we will further expand along these lines in the revised manuscript. Briefly:

First, we identify a photo-stimulation regime that results mostly in the activation of a given targeted glomerulus (and not of other glomeruli in the field of stimulation). To this end, we photo-stimulate and image ensuing neuronal responses in the same axial optical plane. We strobe (alternate) between monitoring dendritic mitral and tufted cell (enhanced) GCaMP responses within the targeted glomerulus and other glomeruli in the field of imaging, while varying systematically the light intensity (Figs. 2,3a; Suppl. Fig. 4b, Suppl. Fig, 5b-d;h-j). We use as criterion for specificity a condition when >95% of significantly responding pixels (above a statistically defined signal response threshold) lie within the anatomical boundaries of the targeted glomerulus. For each glomerulus (or pixel within a glomerulus) we compared the average light response across trials with the baseline reference distribution in the absence of light stimulation. If this value crossed the 99th percentile of the baseline distribution, the glomerulus/pixel within glomerulus was classified as responsive to the photo-stimulation.

Second, using the minimal light intensity regime experimentally identified as ‘specific’ for targeting individual glomeruli in the field of photo-stimulation (< 5% significant activation of off-target pixels), we decouple photo-stimulation in the glomerular layer from monitoring responses of mitral and tufted cells in the deeper layers of the olfactory bulb (100-250 µm axial displacement). This approach enables us to map cohorts of sister (daughter) cells associated with any specific target glomerulus in the field of view (1,2,3…n) by monitoring excitatory (enhanced) responses of mitral and tufted cell bodies. A cohort of sister cells associated with glomerulus x_i (daughters of glomerulus x_i) is defined by those cells which show statistically significant excitatory responses (4 SD - standard deviations - above their baseline fluctuations) specifically in response to photo-stimulation of glomerulus x_i.

Third, in the process, as we photo-stimulate different glomeruli in the field of stimulation, we also observe at times suppressed (inhibitory) responses in a subset of the mitral and tufted cells (exceeding 3 SD in the negative direction their baseline fluctuations). These suppressed responses occur in response to photo-stimulating not the parent glomerulus of a given cell, but other glomeruli in the field. These experiments revealed that within a cohort of sister cells (daughters of glomerulus x_i), only a subset of cells are suppressed by activation of glomerulus x_j, and, in a few example cases, different cells are suppressed by activation of different glomeruli (e.g. x_j vs. x_k), presumably through disynaptic feed-forward inhibition (Fig. 3b iii; 3d; Suppl. Figs. 5f,g; l,m). These preliminary observations suggest that sister cells receive differential inhibitory inputs from glomeruli in the neighborhood. In the revised manuscript, we will expand to further clarify these points.

To verify sister cell identity, the authors could confirm that the predicted sister cells share a similar odour receptive field compared to randomly selected mitral cell pairs. In their previous study employing analogous DMD-based photo-stimulation (Dhawale et al., 2010, Nat. Neurosci.), sister mitral cells did not exhibit such opposite response profiles (firing rate correlation of ∼0.7 between sister cells). Could the authors verify that a comparable activity correlation is also observed among the sister cells identified using ADePT in the present study? In Figure S6, the authors show recordings and stimulation of the same neurons co-expressing GCaMP and ChR2. Applying this experimental design to the mitral/tufted cell population (using a Tbet-Cre mouse transduced in the OB with both GCaMP and Chrimson virus) would constitute a valuable control to clarify the nature of these "inhibited" sister cells.

We thank the Reviewer for the suggestion. We consider that the experiments shown here are proof-of-principle in nature, highlighting the potential of ADePT for mapping functional neural circuit connectivity. In our opinion, further investigating the logic of similarities and differences in the odor responses of sister mitral cells and the nature of inhibitory glomerular interactions forms the focus of future studies. We also note that firing rate correlations can be notoriously difficult to compare and interpret across experimental regimes (spikes vs. calcium imaging).

An additional concern relates to the 21 out of 162 mitral cells that were activated by two distinct glomeruli - a finding that is incompatible with the established OB wiring diagram and that further challenges the claimed stimulation resolution.

Indeed, this reflects some degree of non-specific activation of the targeted glomeruli as discussed above. In the revised manuscript, we will further highlight this issue.

A critical control experiment is also absent: in a Thy1-GCaMP6 mouse lacking any light-sensitive opsin, do the authors observe any unintended side effects of photo-stimulation?

In the revised manuscript, we will include an additional control as suggested by the reviewer. Within the range of intensities used, we did not observe significant modulation of GCaMP6s activity in mitral and tufted cells in mice lacking light-sensitive opsins.

Regarding sister cells (Figure 3), tufted cells are not analysed alongside mitral cells in this dataset, whereas this is elegantly performed in Figure 5 using the DAT+ model. Could the authors also demonstrate how the technique can reveal the complete family portrait of sister mitral and tufted cells?

We thank the Reviewer for the suggestion. We think that the differences between mitral and tufted cells are indeed very interesting to investigate, but in our opinion form the subject of future studies.

(2) The authors have explored only a limited set of photo-stimulation parameters, primarily varying light intensity. They should present additional tests, such as varying the spot size (which appears to be arbitrarily fixed at 30-50 µm) and the z plane of stimulation. The level of activation can vary considerably: for example, in Figure 3a(iii), identical stimulations elicit responses of markedly different amplitudes (see glom#3 and #4). In Figure 2, 5 out of 15 glomeruli failed to respond - could the choice of z-plane account for this variability? The stimulation duration (50-150 ms) also appears somewhat arbitrary: can the authors demonstrate that the technique is compatible with finer temporal patterns (e.g., 10 Hz stimulation for 500 ms using 20 ms light pulses)? What are the spatiotemporal and axial scanning limits of this approach, and can two or three glomeruli be targeted simultaneously with temporally patterned stimulation?

Indeed, here we explored a limited set of spatiotemporal parameters to understand the limits of our technique and provide proof of principle. These can be varied depending on the exact question. In the revised manuscript, we will clearly state what the constraints of the current implementation are and provide context for further optimizations. Briefly, in the current version, individual as well as multiple glomeruli can be photo-stimulated together and 20 ms per pulse regime in trains of pulses is feasible.

(3) One particularly relevant application of this method would be to guide photo-stimulation based on prior functional measurements - for instance, by generating a photo-stimulation mask specifically targeting odour-responsive glomeruli. In the DAT-Cre × Thy1-GCaMP6 experiment shown in Figure 5e, which glomeruli are activated by a given odour, and how does this odor responsiveness influence the efficiency of DAT+ cell-mediated inhibition?

We thank the Reviewer for the suggestion. We are thinking along exactly the same lines. In particular, we would like to investigate the relationship between the degree of overlap in odor responses of individual glomeruli and the strength and specificity of their inhibitory interactions mediated by DAT+ interneurons. In the revised manuscript, we will further expand on discussing this venue of study. We feel however that this investigation is beyond the scope of this technical report.

Reviewer #2 (Public review):

Summary:

In this manuscript, Koh and colleagues describe ADePT (Axially Decoupled Photo-stimulation and Two-photon Readout), a modular approach for combining patterned one-photon optogenetic stimulation with two-photon calcium imaging in independently controlled axial planes. The method relies on a digital micromirror device together with a motorized holographic diffuser to generate spatially confined stimulation patterns while imaging deeper neuronal populations. As proof-of-principle applications, the authors use the system to map excitatory and inhibitory functional connectivity in the mouse olfactory bulb by stimulating superficial glomerular circuits and recording responses from mitral and tufted cells in deeper layers.

This is a well-executed Tools and Resources manuscript. The technical implementation is described in considerable detail, the optical performance is systematically characterized, and the biological experiments provide convincing demonstrations of the types of circuit questions that can be addressed using the method.

Strengths:

The greatest strength of the manuscript is the comprehensive technical characterization of the optical system. The authors carefully benchmark the spatial resolution, axial confinement, registration accuracy, calibration procedure, and practical operating limits of the setup. I found the extensive optical benchmarking particularly helpful, as it gives readers a realistic sense of the operating regime and practical limitations of the approach.

Another strength is the high level of methodological transparency. The optical design, calibration procedures, stimulation strategies, and analysis pipeline are described in sufficient detail that an experienced laboratory could realistically evaluate whether the system is suitable for its own applications. This level of documentation is particularly appropriate for a Tools and Resources article.

A further strength is the clear positioning of ADePT relative to existing approaches. The authors are transparent about the trade-off between spatial resolution and implementation complexity: ADePT does not provide single-cell photostimulation, but offers flexible axial separation, a large stimulation field, and cellular-resolution two-photon readout in deeper planes without requiring a full holographic stimulation system. This defines a credible and potentially useful experimental niche.

The biological applications convincingly demonstrate the utility of ADePT. The experiments identifying sister mitral/tufted cells through selective glomerular stimulation and the mapping of heterogeneous inhibitory influences from DAT-positive interneurons illustrate the types of functional connectivity questions that become experimentally accessible with this approach. Importantly, the authors generally avoid overstating these biological findings and appropriately present them as proof-of-principle demonstrations of the technology.

We thank the Reviewer for their constructive input.

Weaknesses:

The primary limitation is inherent to the method itself rather than the execution of the study. Because ADePT relies on one-photon patterned illumination, photo-stimulation remains restricted to relatively superficial structures and does not achieve single-cell spatial resolution. The authors appropriately acknowledge these constraints and clearly position the method within this operating regime. Consequently, ADePT occupies a useful niche for interrogating spatially organized functional units such as olfactory glomeruli or cortical barrels, rather than applications requiring single-cell precision or deeper tissue penetration.

We agree. In the revised manuscript, we will further expand on these points, highlighting the limitations and advantages of ADePT compared to other techniques in a table discussing various operating regimes.

Although the manuscript describes the approach as relatively simple and cost-effective, implementation still requires careful optical alignment, registration, calibration, and optimization. This does not diminish the value of the approach, but terms such as modular or accessible may better reflect the practical implementation than simple. Likewise, a brief bill of materials, approximate add-on cost, and indication of which components are essential versus substitutable would help prospective users assess the accessibility of the system.

We agree. We will change the text accordingly and provide the additional information as suggested.

Finally, the manuscript provides an impressive level of technical characterization, but much of the practical guidance for adopting the system is distributed across the Results and Discussion. Bringing together the principal limitations, recommended operating regime, expected calibration workflow, evidence for long-term alignment stability, and the circumstances in which ADePT is preferable to alternative approaches would further strengthen the manuscript as a community resource.

We agree. We will proceed accordingly in the revised manuscript.