Optogenetic manipulation of nuclear Dorsal reveals temporal requirements and consequences for transcription

Optogenetic manipulation of nuclear Dorsal reveals temporal requirements and

consequences for transcription

Virginia L Pimmett

1,

*, James McGehee

2,

*, Antonio Trullo

1

, Maria Douaihy

1,3

, Ovidiu Radulescu

3

,

Angelike Stathopoulos

2,#,

Mounia Lagha

1,#

*These authors contributed equally.

#

Joint corresponding authors:

angelike@caltech.edu

, and mounia.lagha@igmm.cnrs.fr

1

Institut de Génétique Moléculaire de Montpellier, University of Montpellier, CNRS

-

UMR 5535,

1919 Route de Mende, Montpellier, 34293, Cedex 5, France

2

California Institute of Technology, Division of Biology and Biological Engineering, 1200 East

California Boulevard, Pasadena, CA 91125 USA

3

Laboratory of Pathogens and Host Immunity, Univ Montpellier, CNRS, INSERM, Montpellier,

France

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

2

ABSTRACT

Morphogen gradients convey essential spatial information during tissue patterning. While both

concentration and timing of morphogen exposure are crucial, how cells interpret these graded

inputs remains challenging to address. We employed an optogenetic sys

tem to acutely and

reversibly modulate the nuclear concentration of the morphogen Dorsal (DL), homologue of

NF

-

κB

, which orchestrates dorso

-

ventral patterning in the

Drosophila

embryo. By controlling DL

nuclear concentration while simultaneously recording

target gene outputs in real time, we

identified a critical window for DL action that is required to instruct patterning, and characterized

the resulting effect on spatio

-

temporal transcription of target genes in terms of timing,

coordination, and bursting.

We found that a transient decrease in nuclear DL levels at nuclear

cycle 13 leads to reduced expression of the mesoderm

-

associated gene

snail (sna)

and partial

derepression of the neurogenic ectoderm

-

associated target

short gastrulation

(

sog)

in ventral

r

egions. Surprisingly, the mispatterning elicited by this transient change in DL is detectable at the

level of single cell transcriptional bursting kinetics, specifically affecting long inter

-

burst durations.

Our approach of using temporally

-

resolved and re

versible modulation of a morphogen

in vivo

,

combined with mathematical modeling, establishes a framework for understanding the stimulus

-

response relationships that govern embryonic patterning.

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

3

INTRODUCTION

It is appreciated that multiple factors

contribute to spatiotemporal dynamics of

morphogen responses including the temporal

alterations to the morphogen gradient itself,

dynamics relating to signal transduction, as

well as downstream interactions between

target genes

2

. In particular, the duration of

exposure to the morphogen has been

highlighted as a crucial determinant of

patterning responses, as exemplified by

recent studies on Nodal, BMP, and Bicoid

morphogens

3,4

.

However,

how

the

morphogen gradients are sensed, both in

terms of their local concentration and the

critical window they must be interpreted

with

in to drive cell fate decision

-

making,

remains a major question in the field.

Furthermore, gene regulatory networks

(GRNs) act within responding cells to

interpret morphogen signals and perhaps

there is built

-

in robustness to these systems

to

accommodate

varied

morphogen

dynamics

5

.

Whether

morphogen

spatiotemporal dynamics are central to

decoding morphogen input into discrete cell

fates or, instead, downstream transcriptional

responses are the major drivers is a

fundamental, yet unresolved, question.

Distinguishing between these tw

o alternative

and potentially non

-

exclusive modes of

action, involving direct versus indirect

morphogen control, requires approaches in

which the patterning process can be

perturbed and studied in real time.

The dorsoventral axis of the developing

Drosophila

embryo is a well

-

established

example of a morphogen

-

patterned system in

which both the morphogen input and target

gene outputs can be

followed live in time and

space

6

–

8

. In

Drosophila

syncytial embryos,

graded input by DL is important for activating

expression of target genes

snail

(

sna

) and

twist

(

twi

) in ventral regions,

ventral

-

neuroblasts defective

(

vnd

) in ventrolateral

regions, and

short gastrulation

(

sog

) in lateral

regions (Figure 1H)

9

. In addition to the spatial

patterning blueprint encoded by a gradient of

nuclear DL levels along the dorsoventral axis,

other transcription factors (TFs) also ensure

establishment of precise borders and the

adoption of distinct cell fates. For instance,

in

ventral cells, the TFs Twist (Twi) and Snail

(Sna) instruct the mesodermal fate and the

subsequent

epithelial

-

mesenchymal

transition (EMT) program, essential to

gastrulation

10

. Moreover, Sna directly

represses the transcription of the DL target

sog

in ventral cells which form the

presumptive

mesoderm

11,12

.

This

combinatorial action of DL, Sna, and other

factors such as Zelda increasingly restricts

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

4

sog

expression to lateral regions prior to

gastrulation

13,14

.

To orchestrate this complex gene regulatory

network, the action of DL must be tightly

regulated in both space and time. The DL

gradient has been shown to be dynamic, as

levels were observed to change over time

15

–

17

. Although these changes in the gradient

over time have been detected, the functional

relevance during dorsoventral patterning

remains unresolved. Recently, optogenetic

tools have been used to manipulate DL levels

using an opto

-

degron

1,18

. However, the

irreversible nature of this perturbation inhibits

identification of transient critical windows of

DL availability

18

. By establishing a reversible

optogenetic manipulation paradigm for DL

nuclear levels, this study has identified gene

-

specific temporal requirements for DL across

the dorsoventral axis. Timing is a key concept

in gene regulation, particularly during

devel

opment, where the dynamics of

morphogens play a pivotal role in shaping

gene expression that ultimately leads to

patterning. Here, we have identified specific

temporal requirements, showing that DL

nuclear reduction in nc13

decreases

sna

expression and leads to a change in

sog

stochastic transcription properties in nc14

like decreasing the duration of a long

transcriptionally inactive (OFF) state (long

inter

-

burst periods). Our analysis further

reveals changes of these transcriptional

bursting properties ac

ross space, suggesting

a mispatterning at the kinetic level. This

analysis was only possible through a novel

combination of approaches, including

mathematical modeling.

RESULTS

We used DL

LEXY

, which supports inducible

DL nuclear export and is reversible (

Figure

S1 A

-

E

)

1,19,20

, together with the MS2/MCP

imaging system to monitor spatiotemporal

expression of target genes

in vivo

21,22

. In the

presence of blue light, DL

LEXY

is translocated

to the cytoplasm, and when the blue light is

removed DL

mCh

-

LEXY

reenters the nucleus at

levels similar to before the light was applied

(

Figure S2 A

-

D

)

1

, without significant

photobleaching (

Figure S2E

). To monitor

transcriptional dynamics of DL targets, we

focused on four DL targets

expressed in

ventral, ventrolateral, or lateral regions

spanning the dorsoventral axis (

Figure 1H

).

We assayed gene expression of these four

targets when blue light was applied during

specific embryonic stages to define critical

windows of DL action.

The critical window for DL activity is nc11

-

13 for

twi

and nc13 for

sna

We first examined the window of time DL

must act to support target gene expression in

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

5

ventral regions of the embryo, within the

presumptive mesoderm. We used a

published

sna

-

BAC

23

transgene and created

a new

twi

-

MS2

allele by inserting MS2 into

the

endogenous

twi

locus

using

CRISPR/Cas9

(

Figure 1I,K

; see Methods

)

.

In the dark,

twi

-

MS2

expression is reliably

detected from nc12 up through mid

-

nc14 (i.e.

nc14b) but is diminished in nc14 (

Figure 1A,

Movie S1

). When blue light was applied at

nc11

-

12,

twi

-

MS2

signal was greatly reduced

at nc12 and 13, but only moderately reduced

at nc14 (

Figure 1B

). Alternatively, if embryos

were exposed to light from nc11

-

13,

twi

-

MS2

signal was reduced at nc12, 13, and 14,

including

when returned to the dark during

nc14 (

Figure 1C, Movie S1

). When

illuminated at nc13,

twi

-

MS2

signal is

reduced at nc13 and at nc14, but to a lesser

degree than when illuminated from nc11

-

13

(

Figure 1D, J, Movie S1

)

. This indicates the

DL

-

critical window for

twi

activation appears

to be nc11

-

13, with nc13 appearing to be

more important for the initial peak in

twi

expression than nc11

-

12. Importantly, the

decrease in

twi

transcription under blue light

is

not

due

to

photobleaching,

as

demonstrated by our bleaching analysis

(

Figure S3A

). Taking a similar approach, we

found that another ventral target gene,

sna,

exhibits a narrower critical window for DL

action. When blue light is shone at nc13,

sna

-

MS2

signal is greatly reduced at nc13 and

nc14

(

Figure 1E

-

G, Movie S2

)

. Since this

reduction in the number of

sna

transcription

sites (TS) was so great (

Figure 1L

),

illuminating embryos from nc11

-

13 was not

necessary for

sna

. Thus, reduction of DL

levels in this short window of ~15

-

20 minutes

(nc13) is sufficient to silence

sna

transcription

in nc13 and drastically reduce expression at

following stages. We also observed that the

occurrence of gastrulation is reduced in

embryos illuminated at nc13 (

Figure 2A

-

D

).

We believe that these gastrulation defects

are due to the absence

of Sna protein,

elicited by

DL

nuclear reduction in nc13. We

confirmed the absence of Sna protein with a

more sustained light exposure of DL

LEXY

embryos with a blue

-

light box set

-

up (

Figure

S1F

; see Materials and methods). Therefore,

while both

twi

and

sna

target genes are

expressed in ventral regions where high

levels of nuclear DL are present, these genes

exhibit different temporal dependencies, with

DL needing to act earlier to support

twi

and

only later to support

sna

.

Next, we asked whether the time of DL action

or the duration of this input signal is the

critical factor driving

sna

expression. To

discriminate between these two alternatives,

DL was exported with blue light in nc12 as

well as in early nc14 within the same embryo,

for a total duration of 26.3 ± 1.0 min (mean ±

SEM), ~6 min longer than the length of time

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

6

blue light was given during a complete nc13

cycle duration, which was 20.5 ± 0.6 min

(mean ± SEM). When we illuminated

embryos with blue light during nc12 and a

second time at early nc14,

sna

expression

was able to recover during nc13 and late

nc14 (

Figure 1G,L; Movie S2

). A similar

experiment could not be performed for

twi

, as

twi

expression is difficult to detect at mid to

late nc14 and recovery of

twi

expression

would not be observable after removal of blue

light. These results support the hypothesis

tha

t the timing of DL input in nc13 is critical

for

sna

expression.

vnd

and

sog

persist during nc14 despite

reduced nuclear DL levels at nc13

Having defined the critical temporal window

required to activate ventral targets, we next

questioned whether DL’s action at nc13 was

also important for ventrolateral and lateral

targets. To this end, we imaged

vnd

, a DL

target expressed in the ventrolateral domain,

and

sog,

which is expressed in the lateral

domain (

Figure 1H

). We employed a reporter

transgene to follow

vnd

expression in early

embryos based on output of a single

enhancer (vndNEE

-

MS2)

24

(

Figure 2E

)

and

used published fly stocks with MS2

engineered into the

sog

locus

(sog

-

MS2)

1,25

(

Figure 2J

)

.

When

assayed using DL

LEXY

with illumination

on the ventral side

(Figure 2F,K)

, we

observed that

vnd

and

sog

responded

differently to removal of Dorsal.

vndEEE

expression was acutely decreased in the

presence of blue light when it was applied at

either nc13 or parts of nc14, but there was no

lasting effect as observed for

sna

(Figures

2G,H,I, and S4),

regardless of whether

vndEEE

was imaged ventrolaterally or

ventrally

(Figures 2F and S4B; Movie S3)

.

We quantified the number of active

vndEEE

nuclei at nc14 and found it appears largely

unaffected by removal of DL at nc13 (

Figure

2H

). A previous study showed that no matter

how long the blue light is shone on DL

LEXY

embryos,

sog

expression is retained (e.g. for

the entirety of nc13

-

14)

1

. Even though DL

LEXY

is predominantly cytoplasmic under blue

light, there is some residual, low

concentration DL in the nucleus

1

(

Figure S2

).

This small amount of transient DL may be

sufficient to support

sog

transcriptional

activation

.

When embryos are illuminated

with blue light on the ventral side, some

sog

was detected in regions where it should be

repressed

(

Figure 2K,L,N; Movie S3

)

. The

number of

sog

active TS was higher with light

at nc13 when compared to the dark condition

(

Figure 2M

).

Since we previously did not

observe a change in

sog

expression with

removal of DL

1

and the number of

sog

TS

increased instead of decreasing, this

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

7

suggests a loss of repression under

blue light

in ventral regions.

These experiments were able to define

distinct temporal requirements for DL in

supporting the activation of two target genes.

DL action is required early at nc11

-

13 for

twi

,

and at nc13 for

sna

. If DL levels are reduced

during these time windows, then the

respective genes expression is reduced at

nc14. In contrast, nc13 was not critical for

activation of the ventrolateral and lateral DL

target genes

vnd

and

sog

. For nc13 and 14,

DL input to

vndEEE

is acutely required, but

there is no long

-

term memory effect

and gene

expression recovers once DL is available

again. The number of nuclei exhibiting an

active

sog

TS was apparently refractory to a

reduction of DL levels, even during blue light

illumination. In fact, reducing DL levels on the

ventral side led to an increase in

sog

TS,

likely due to a loss of Sna (

Figure S1F

).

Reduction of DL levels in nc13 does not

affect the kinetics of

sog

transcription at

nc13

While this quantification suggests that the

probability to activate the transcription of the

sog

promoter is insensitive to blue light

-

induced DL nuclear export, it cannot

determine if a more subtle effect may operate

at the level

of transcriptional kinetics

. For

instance, it is well appreciated that most

genes are transcribed in a discontinuous

manner with bursts occurring at various

timescales

26

. Like other developmental

genes, endogenous

sog

also exhibits a

bursty transcription in the early embryo

(

Movie S3; Figure 3F,H

)

25,27

. Analysis of live

transcription signals shows that transcription

activity is intermittent, alternating between

active and several inactive periods.

Mathematical models can capture this

process, termed ‘bursting’, and characterize

both the probability and

durations of the

active and inactive transcriptional periods. In

addition, as

sog

is the only target gene that

remains on during blue light, it was the best

candidate for this analysis. Therefore, we

further investigated the bursting behavior of

sog

when nuclear DL

LEXY

levels are perturbed

by blue

-

light at nc13, with the understanding

that this would lead to a substantial reduction

in

sna

expression but only a partial reduction

in

twi

. To examine

sog

bursting kinetics, we

imaged its transcription with high

temporal

resolution. Such a fast imaging set

-

up

inevitably leads to bleaching when using a

red fluorescent detector, such as the MCP

-

RFPT employed here. However, our

bleaching study demonstrates that prior to 20

min into nc14, less than 20% of bleaching is

observed, allowing adequate detection

(

Figure

S3B

).

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

8

Since

sog

is expressed in nc13 throughout

ventral and lateral regions (the presumptive

mesoderm and neurogenic ectoderm) with no

respect to the future differentiation of these

two presumptive domains, we pooled

sog

+

nuclei (with an active TS) for kinetic profiling

from the entire imaging window spanning

ventral to ventrolateral regions (

Figure 3A,B,

dashed box). Owing to the absence of

fluorescently labeled nuclear markers, we

were unable to orient live transcription traces

relative to mitosis and thus trac

es were

tracked relative to the first detected

transcriptional activation event in each

nuclear cycle.

As discussed above, the number of active

nuclei expressing

sog

increased in nc13

when nuclear DL levels are perturbed with

blue light (

Figure 2M

). We also quantified the

integral amplitude in nc13 for each trace,

which is a proxy for total mRNA output, and

noted no difference in total

sog

mRNA output

(

Figure 3C

). This was confirmed by

quantifying nascent

sog

transcription using

single molecule FISH (

Figure S5

), where

extended (2h) DL export did not result in

substantial perturbations to

sog

nascent

transcri

ption compared to the control. Thus,

the embryo

-

level transcription of

sog

appears

to be unaffected by optogenetic removal of

DL.

However,

sog

transcription is bursty,

meaning that the fluctuations, not just the

mean mRNA production, may be biologically

relevant. Therefore, we next turned to a

previously

established

mathematical

modeling approach to examine the

underlying kinetic parameters of

sog

transcriptional bursting at the single nucleus

level when the DL gradient was perturbed.

During a given nuclear cycle, the MS2 signal

that carries information on the bursting

kinetics, progresses through several stages.

Initially, following

mitotic exit, no signal is

detected and this period corresponds to time

for DNA replication and resumption of

transcription. This period is typically modeled

by considering the distribution of the time to

reactivation

28

. Following this initial period,

transcription builds up at each transcription

foci and can reach a stationary phase, during

which the signal remains stable. Bursting

during this stationary phase is analyzed using

BurstDECONV

29

.

We used BurstDECONV to deconvolve

sog

-

MS2

signal, i.e., reconstruct the sequential

polymerase initiation events contributing to

this signal

(

Figure 3D

)

29

. The deconvolution

relies on a model of the contribution of a

single polymerase to the MS2 signal, which

depends, among other factors, on the length

of the transcribed MS2 and post

-

MS2

sequences. Since the MS2 loops were

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

9

inserted in

sog’s

first intron, an additional

point of consideration for the modeling was to

understand the gene’s splicing, recently

observed to be largely recursive in human

cells

30

. We ensured

sog

splicing was not

recursive at this stage by consulting NET

-

seq

data from

Drosophila

embryos

31

. Instead,

sog

splicing likely occurs at intron

-

exon junctions

and is co

-

transcriptional as supported by

FISH data

11

. We therefore considered that

the MS2 sequence was entirely transcribed

into mature RNA.

Furthermore, the deconvolution method

implemented by BurstDECONV assumes

that the signal is stationary. Therefore, we

first determined whether

sog

transcription in

nc13

(

Figure 3E

-

I

)

reached

stationary

dynamics by plotting the mean waiting time

between polymerase initiation events for a

short window (81 sec;

Figure 3J

). This

represents the mean transcription rate for

each window, expressed as the product of

the Pol II initiation rate (k

INI

) in the active state

and the probability of a nucleus to be active

(P

ON

). Transcription is considered stationary

if this value remains stable across sequential

time windows. Indeed, this demonstrated that

in both dark (nuclear DL, black) and DL

-

exported (blu

e) conditions, the underlying

kinetic dynamics reached stationarity.

Having established the specific window

exhibiting stationarity of the

sog

signal, we

next sought to extract

sog

promoter switching

rates. To access transcription kinetics, we

examined the distribution of waiting times

between Pol II initiation events during the first

15 minutes of nc13 and fitted it with a multi

-

exponential function (

Figure 3K,L

). The

multi

-

exponential fitting gives access to the

kinetic parameters driving

sog

transcription

such as the duration of the productive (ON)

and non

-

productive (OFF

) states as well as

the Pol II firing rate in the ON state. Critically,

this approach does not presume a specific

number of states

a priori

, but instead

discovers it as an emergent property of the

data itself, resulting from the number of

exponentials needed to fit the data. In both

dark (nuclear DL) and light (DL export), this

distribution could not be fitted by a bi

-

exponential distribution

, indicating that the

classical

two

-

state

random

telegraph

promoter model is insufficient to fit the data.

A three

-

ex

ponential fitting, corresponding to

a three

-

state model, was sufficient, however,

for both the nuclear and exported DL cases

(

Figure 3K,L

, compare blue and orange

curves). This three

-

state promoter scheme

consists of a competent ON state and two

distinct OFF states, each characterized by a

different mean duration: OFF1 (long

-

lived,

295

-

303 sec) and OFF2 (short

-

lived, 13

-

16

sec). In terms of

probability, the prolonged

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

10

inactive state (OFF1) is the most probable

state, dominant in both dark and light

conditions. When nuclear DL levels were

reduced with blue light in nc13,

sog

promoter

bursting dynamics also were captured by a

three state model with similar kinetics to

those observed in the dark (

Figure 3M

). This

analysis of

sog

transcriptional dynamics

suggests that upon optogenetic perturbation

of DL with the LEXY system in nc13,

sog

transcriptional kinetics are not acutely

affected.

DL export in nc13 leads to mispatterning

in nc14, accompanied by a change in

bursting kinetics

While in nc13,

sog

is expressed throughout

the presumptive mesoderm and neurogenic

ectoderm, it is progressively repressed in the

nc14 mesoderm via increasing levels of Sna

but maintained in the neurogenic ectoderm

32

.

Given the specific requirement of DL in nc13

to promote the expression of

sna

, and by

extension the repression of

sog

in nc14, we

examined the transcription of

sog

in nc14

when nuclear DL levels are perturbed in nc13

(

Figure 4A

).

Because

sog

is known to have variable

expression dynamics based on the tissue in

which it is expressed

25,32,33

, we began by

examining

sog

repression exclusively in the

presumptive mesoderm. In dark embryos, we

identified the mesoderm/neuroectoderm

boundary based on repression of

sog

by Sna

in the mesoderm and spatially selected only

those nuclei located in the presumptive

mesoderm (see

Figure 4B

). In illuminated

embryos, the mesoderm/neuroectoderm

boundary driven by Sna protein is perturbed

due at least in part to changes in

sna

expression, and thus cannot be used to

differentiate the tissues. To circumvent this,

embryos were imaged with

the presumptive

ventral furrow oriented to one edge of the

imaging window, and the mesoderm was

selected as the 30 μm to either side of the

failed furrow at the end of nc14. As expected,

given the reduction in

sna

transcription, we

observed an extended maintenance of

sog

transcription in the presumptive mesoderm

under illumination compared to control dark

embryos (

Figure 4C

).

sog

derepression in

the ventral part of the embryos was also

observed by an alternative approach not

relying on MS2/MCP imaging. We p

erformed

single molecule RNA FISH in DL

LEXY

embryos with a more sustained light

exposure using an optobox (

Figure S5A

-

C

).

Interestingly, in these conditions the

quantification of TS intensities suggests that

in the light, transcription of

sog

in ventral

nuclei is not different from more laterally

located nuclei in the dark (

Figure S5E

).

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

11

To understand how reducing DL levels in

nc13, and the subsequent reduction of nc14

Sna protein levels (

Figure S1F

), affects

sog

transcription

dynamics

in

nc14,

we

characterized two metrics from the movies for

each nucleus: the time to reactivation (

Figure

4D

, double

-

arrow domain) and the repression

breakpoint (

Figure 4D

, switch from green to

red trace), corresponding to the post

-

mitotic

transcription reactivation and to the onset of

repression, respectively.

W

e first considered

the timing required to reach

the first

transcriptional activation after mitosis, which

we call the time to reactivation

34

(

Figure 4E

).

This time was not significantly different

between the control (dark) and DL

-

exported

(light) nuclei. As discussed in Dufourt*,

Trullo*

et al,

2018, the reactivation time has

two components, a deterministic component

common to all nuclei, and a stochastic

component. Although both components

contribute to the reactivation period duration,

only the stochastic component depends on

the number and du

ration of rate

-

limiting steps

in

reassembling

the

transcriptional

machinery

35,36

. The reactivation metric used

here accounts only for the stochastic part of

sog

expression in early nc14 when nuclei

reinitiate

sog

expression following mitosis.

Indeed, time zero is defined here as the

moment when the first nucleus resumes

transcription after mitosis, rather than the

moment of mitosis itself. By making this

choice, we exclude the deterministic

component of the reactivation time, which is

consistent across all nuclei. As mentioned

earlier, this approach does not hinder our

ability to detect changes in the mechanism of

reactivation.

We then turned to Bayesian Change Point

Detection (BCPD) to determine the onset of

repression, when each individual nucleus

underwent a transition from building up

repression to stable, full repression

37,38

. For

further modeling purposes, the signal is

considered stationary after the repression

onset. In the DL

-

exported (light) nuclei, we

observed that repression onset for

sog

was

significantly delayed (

Figure 4F

, dash lines).

Moreover, upon DL

-

export in 13,

sog

repression within the ventral region appeared

much less coordinated (i.e. the repression

onset shows greater internuclear variability)

in nc14 than in dark conditions (

Figure 4F

).

Thus, acute reduction of DL levels in the short

temporal window spanning nc13 has a

substantial, persistent impact on the

dynamics of

sog

transcription later in nc14.

Because DL export in nc13 affects

sna

expression (

Figures 1F and S1F

) and

because Sna is an important driver of the

mesodermal fate

39,40

, we expect that the DL

-

depleted ventral cells may have either ‘lost’

or failed to acquire a mesodermal fate and

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

12

may have adopted instead a more lateral

-

like

fate (

Figure 4G

). To test this hypothesis at

the kinetic level, we sought to extract the

transcriptional bursting kinetics for

sog

in

various spatial regions of the embryo at

stationarity in nc14 when kept in the dark or

exposed to light at nc13. Ventral region nc14

sog

data in embryos previously exposed to

light in nc13 (Ventral

LIGHT

) or continuously

kept in the dark (Ventral

DARK

) were compared

with lateral regions (i.e. Lateral

LIGHT

and

Lateral

DARK

). Because em

bryos fail to

gastrulate when exposed to light (i.e. there is

limited ventral furrowing), the lateral region

was confidently identifiable only in embryos

kept in the

dark (Lateral

DARK

).

We used BurstDECONV to convert individual

fluorescent traces into polymerase initiation

events (

Figure S6D

-

F

) from which kinetic

parameters can be modeled

29

for the three

conditions: Ventral

DARK

, Ventral

LIGHT

, and

Lateral

DARK

at stationarity in nc14 (see

Methods and

Figure 3D

). After model fitting

(

Figure S6J

-

L

), both the control and DL

-

exported nuclei could not be fit with a two

-

state dynamic and instead required a three

-

state fitting with a competent ON state as well

as both long (minutes) and short (seconds)

OFF states. In the dark, the

sog

promoter

seems to transition between three promoter

states in both nc13 and nc14, however the

duration and probabilities of the dominant

long OFF state

are different between these

two cycles (

Figures 4H

-

M and S6M

-

O

). After

DL export in nc13, we observed that the

duration of the long non

-

productive (OFF1)

state in nc 14 was reduced by 25% and its

probability was decreased by 12% in the

ventral domain. Furthermore, the duration of

the short non

-

productive (OFF2) state

in

nc14 was slightly increased, and its

probability was significantly increased by

42% in the ventral domain (

Figure 4H,I

,

compare Ventral

DARK

with Ventral

LIGHT

).

Notably, these trends associat

ed with the

ventral domain after light exposure closely

matched the dynamics of the lateral domain

control (

Figure 4H

-

I

, compare Ventral

LIGHT

with Lateral

DARK

). Collectively, these results

suggest

sog

transcription in the ventral region

may convert to a more lateral

-

like

transcriptional profile in nc14 following

transient depletion of DL in nc13.

DISCUSSION

Using an optogenetic approach to control DL

localization with LEXY

in vivo

(

Figure 5A

),

we identified specific time windows in which

DL input is essential for the expression of

select DL targets expressed in ventral

regions (

sna

and

twi

;

Figures 1

and

5B

). In

contrast, we found that lateral genes like

vnd

and

sog

do not respond in the same way as

ventral genes to the narrowest critical window

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

13

of DL activity (i.e. nc13) (

Figures 2

and

5B

).

Additionally,

we

characterized

the

spatiotemporal bursting kinetics of the

endogenous gene

sog

gene across different

stages of the patterning process (

Figure 5C

).

Notably, when DL is displaced from the

nucleus early on,

sog

is derepressed in

ventral regions at later stages, resulting in

mispatterning. This mispatterning causes

ventral cells, the presumptive mesoderm, to

instead exhibit properties characteristic of

neurogenic ectoderm. These shifts in identity

w

ere quantifiable by changes in bursting

kinetic properties, with ventral cells adopting

a kinetic profile usually associated with lateral

cells. These findings were supported by the

observation that light exposure at nc13 leads

to gastrulation defects, lik

ely because ventral

cells can no longer undergo the cell shape

changes necessary to support invagination

(

Figure 2A,C

).

These results echo, but are distinct from, the

temporal requirement identified for the Bicoid

morphogen that controls patterning along the

posterior

axis

of

Drosophila

embryos. Duration of Bicoid input is important

for targets expressed at the anterior pole,

presumed to be high threshold targets

4

.

However, we found that

sna

, a presumed

high

-

threshold DL target gene, is not

sensitive to duration of DL input before mid

nc14; as ~25

-

30 min blue light exposure

during nc12 + nc14 early did not turn off

sna

,

whereas a less than 20 min exposure during

nc13 did.

Instead, our data support the view

that

sna

and

twi

, both expressed in ventral

regions, require DL input at a critical time (i.e.

nc13 for

sna

and nc11

-

13 for

twi

) when DL

activity is necessary so that

sna

and

twi

can

be expressed later at nc14. It is notable that

sna

and

twi

exhibit different critical windows

of DL input despite being both high

-

threshold

DL targets. We propose this relates to

combinatorial control by other TFs in addition

to DL, including opposite autoregulatory

activities.

twi

is known to exhibit positive

autoregulatory feedback whereas

sna

is

known to exhibit negative autoregulatory

feedback

41

. We suggest that if a little Twi or

Sna is made at or following the time that DL

is removed at nc13, the respective

autoregulatory feedbacks might lead to

retention of

twi

and loss of

sna

transcripts

.

Therefore, the simplistic model of threshold

responses to DL is not sufficient to predict

whether or not a gene will be retained when

DL is perturbed.

Previously, we showed that

sna

also exhibits DL dependence at early

nc14 using an optogenetic approach based

on

degradation

18

. Therefore, while DL activity

at nc13 is required, it is likely not sufficient for

correctly activating

sna.

There are a number of possible explanations

for why DL is critical at nc13. One is that DL

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

14

may support an epigenetic change or perform

some other pioneering activity at the

sna

locus only possible at nc13 that is required for

later expression of

sna

at nc14. While it

remains unclear why the DL level of mother

nuclei at nc13 has an effect in nc14, we can

envisage

several

potential

molecular

mechanisms including mitotic bookmarking

(e.g. GAF)

42

, mitotic

-

assisted repression

12

,

‘memory’ transcription hubs

43

or post

-

transcriptional modulation of mRNA or

protein half

-

life. Our results regarding the

criticality of nc13 also agree with a previous

study that used an optogenetic approach to

activate extracellular signal

-

regulated kinase

(ERK)

44

. Specifically, they found that when

light was applied in the trunk at nc13, ectopic

expression of

hückebein (hkb)

, a repressor of

sna

, leads to reduction in

sna

transcription

and defects in gastrulation.

Our inducible and reversible manipulation of

a morphogen TF while recording target gene

responses allowed us to dissect time

-

dependent transcriptional dependencies

within a complex gene regulatory network

(GRN). Indeed, target gene sensitivity to a

dedicat

ed temporal window could be a direct

effect or the result of complex feed

-

forward/combinatorial regulation of other

genes by the morphogen input, a common

feature of GRNs

45,46

. Gene

-

gene interactions

can influence both the timing and duration of

critical temporal windows. Using DL

LEXY

to

decrease

DL

levels

and

via

our

characterizations

of

critical

temporal

windows, we can study how the GRN

responds to reducing the levels of a specific,

lineage

-

driving TF in a system in which DL

levels have recovered to their previous level.

For instanc

e,

sna

is a critical component of

the gene regulatory network acting to support

mesoderm formation and a regulator of

neurogenic ectoderm and

mesectoderm

patterning. Sna protein is known to act as a

transcriptional repressor, and without

sna

transcription at nc13 (due to DL export in

nc13), Sna protein levels are diminished in

nc14, leading to patterning failure. In this

manner, we could distinguish both direct, fast

effects of DL perturbation (i.e. reduced

expression of targets such as

sna

,

twi

, and

vnd

) as well as phenotypes that present later,

after blue light is removed and DL nuclear

levels are recovered (i.e. ventral expansion of

vnd

and

so

g

)

.

This study also highlights the value of

combining transient perturbations with

quantitative approaches to explore more

subtle effects on single

-

cell state changes.

We used this insight of the critical time

window for

sna

in nc13 to study the

transcriptional dynamics of other target

genes. Hence, we could decode how DL

levels affect bursting across time (nc13

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

15

versus nc14) but also across space (ventral

versus lateral). Indeed, by nc14

sog

is

repressed in the mesoderm (by Sna) while

kept active in lateral regions, thus offering the

possibility to compare bursting kinetics in

various spatial domains. In both nc13 and

nc14 and regardless of DL levels,

sog

promoter dynamics can be recapitulated by a

three

-

state model, comprising a competent

ON state, a short

-

lived OFF state (seconds)

and a longer

-

lived (minutes) highly probable

second inactive state (e.g.

Figures 3K

,L,

orange line,

and S6J,K,L

). While our data

are unable to demonstrate the biochemical

nature of these promoter states, their

timescale and their sensitivity to DL levels

and spatial region provide clues to discuss

what they could represent. The competent

ON state, from which polyme

rase initiates, is

generally interpreted as the preinitiation

complex (PIC)

-

assembled promoter state.

Importantly, while other studies propose that

the probability to be ON stands as the main

parameter

underlying

transcriptional

dynamics

8,47,48

,

our

study

uncovered

changes in the duration of an inactive state.

The only promoter state that shows a clear

change following transient DL manipulation is

the longer lived OFF1 state. There are

several possible interpretations of this result.

In the blastoderm,

sog

expression is

orchestrated by two enhancers, an intronic

and a distal enhancer, both regulated by DL.

Interestingly,

deletion

studies

have

suggested that instead of being redundant,

these two enhancers integrate activation and

repression

signals

differently

25,33

.

For

example, the intronic enhancer appears to be

the principal enhancer in ventral and

ventrolateral regions at nc13. Since the long

OFF state changes between nc13 and nc14

(see

Figure 4K

-

M

), we speculate that this

state may correspond to enhancer

-

promoter

interactions: a long OFF state in nc13 when

expression primarily relies on one enhancer,

that could be reduced in nc14 thanks to the

combinatorial action of two enhancers

modulated by the

presence or absence of a

repressor

25

. The kinetics of these long

inactive states (minute

-

range) can possibly

provide a clue on the timescale and

probabilities

of

enhancer

-

promoter

encounters

49

. An

alternative, possibly simpler

explanation, interprets the long OFF1 state

as DNA bound repressor molecules. The

lifetime of this state in the case of multiple

binding sites, representing the time required

to clear all the sites, may depend on the

number o

f occupied sites. The concentration

of the repressor is thus reflected in this

duration, as higher concentrations lead to a

longer lifetime.

Taken together our results demonstrate that

optogenetic

perturbations

with

high

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint

16

resolution imaging and quantitative modeling

can dissect how a TF affects target gene

transcription dynamics in space and time.

When employed in the context of a pivotal

morphogen TF, patterning can be disrupted

and lead to mispatterning: here, presumptive

mesoderm (ventral) to neurogenic ectoderm

(lateral), detectable at the level of the kinetics

of transcriptional bursting.

.

CC-BY-NC 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted November 28, 2024.

;

https://doi.org/10.1101/2024.11.28.623729

doi:

bioRxiv preprint