Rif1 restrains the rate of replication origin firing in Xenopus laevis

ARTICLE

Rif1 restrains the rate of replication origin

ring in

Xenopus laevis

Olivier Haccard

, Diletta Ciardo

, Hemalatha Narrissamprakash

, Odile Bronchain

, Akiko Kumagai

William G. Dunphy

, Arach Goldar

& Kathrin Marheineke

✉

Metazoan genomes are duplicated by the coordinated activation of clusters of replication

origins at different times during S phase, but the underlying mechanisms of this temporal

program remain unclear during early development. Rif1, a key replication timing factor,

inhibits origin

ring by recruiting protein phosphatase 1 (PP1) to chromatin counteracting S

phase kinases. We have previously described that Rif1 depletion accelerates early

Xenopus

laevis

embryonic cell cycles. Here, we

nd that in the absence of Rif1, patterns of replication

foci change along with the acceleration of replication cluster activation. However, initiations

increase only moderately inside active clusters. Our numerical simulations suggest that the

absence of Rif1 compresses the temporal program towards more homogeneity and increases

the availability of limiting initiation factors. We experimentally demonstrate that Rif1 deple-

tion increases the chromatin-binding of the S phase kinase Cdc7/Drf1, the

ring factors

Treslin, MTBP, Cdc45, RecQL4, and the phosphorylation of both Treslin and MTBP. We show

that Rif1 globally, but not locally, restrains the replication program in early embryos, possibly

by inhibiting or excluding replication factors from chromatin.

https://doi.org/10.1038/s42003-023-05172-8

OPEN

Université Paris-Saclay, CEA, CNRS, Institute for Integrative Biology of the Cell (I2BC), 91198 Gif-sur-Yvette, France.

Institut de Biologie de l

’

Ecole Normale

Supérieure, Ecole Normale Supérieure, CNRS, INSERM, PSL Research University, Paris, France.

Paris-Saclay Institute of Neuroscience, CNRS, Université

Paris-Saclay, CERTO-Retina France, 91400 Saclay, France.

Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA

91125, USA.

✉

email:

kathrin.marheineke@i2bc.paris-saclay.fr

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

1

1234567890():,;

D

NA replication in eukaryotes starts at several hundred to

thousands of sites called replication origins, which are

activated during the S phase according to a regulated

spatio-temporal replication program

–

; the dysregulation of this

program leads to genomic instability. Genome-wide studies have

shown that large adjacent genomic segments, called replication

domains, share a similar replication timing (RT or mean repli-

cation time of a locus in a cell population). Early replication

generally occurs in actively transcribed euchromatin in the A

compartments in the nucleus of differentiated cells. During early

developmental stages, when transcription has not begun and

without distinct eu/heterochromatin, the rapid cell cycles rely

only on maternally supplied factors; whether or how replication

timing during this developmental period is regulated remains

unclear. We and others previously found that in the

Xenopus

vitro system, active replication origins are spaced 5

–

15 kb

apart

, clustered in early- and late-

fi

ring groups of origins

and that an embryonic temporal timing program exists in the

Xenopus

egg extract system and early embryos from

Xenopus

and

Zebra

fi

–

The coordination of more than

fi

fty different protein factors is

necessary to (i) license, (ii) activate (

fi

re) a replication origin, (iii)

establish two replication forks, and (iv) achieve the faithful

duplication of the genetic material. In late mitosis and G1 phase,

origins are licensed for replication by loading onto chromatin the

pre-replicative complex (pre-RC, for review, see

), composed of

the six ORC (origin recognition complex) subunits, the Cdc6

(cell-division-cycle 6) and the MCM (mini-chromosome main-

tenance) 2

–

7 helicase complex. Cyclin- and Dbf4/Drf1-dependent

kinases (CDKs and DDKs) activate the pre-RC at the start and

during S phase. In budding yeast, Sld7/Sld3, Dbp11 and Sld2

regulate the maturation of the pre-RC into the functional helicase

complex Cdc45-MCM-GINS with pol

(CMGE) in different

DDK-CDK-dependent steps (for review, see

). Their respective

counterparts in metazoans, MTBP (Mdm2-binding protein)/

Treslin, TopBP1 and RecQL4 are loaded onto the pre-RC to build

the pre-initiation (pre-IC) complex

–

. The activation of the

replicative helicase is the crucial step during origin

fi

ring. The

competition between origins for limiting replication factors such

as Sld3, Sld2, Dbp11, and Dbf4 contributes to setting their

fi

ring

time in budding yeast

. Treslin, RecQL4, TopBP1, and the

embryonic DDK regulatory subunit Drf1 levels become limiting

for replication following the increase in the nuclear to cytosolic

ratios in vitro and in vivo after 12 cell cycles during

Xenopus

development

Rif1 (Rap1-interacting factor 1) is a key regulator of the

replication timing program

and modulates late replication in

yeast

, mice

, and human cell culture lines

. In the absence

of Rif1, genome-wide studies (Repli-Seq) revealed substantial

switches of late RT domains becoming early and early RT

domains becoming late

–

. Rif1 binds protein phosphatase 1

(PP1) and recruits it to phospho-sites on the pre-RC complexes,

acting both negatively on origin activation by counteracting the

phosphorylation of the replicative helicase complex MCM2

–

7by

DDK

–

and positively on the licensing step

. However, single-

molecule (SM) analyses by DNA

fi

ber stretching techniques led to

contradictory conclusions on more local effects, such as origin

activation and fork speed, in mammalian cells after Rif1

depletion

. Rif1 mostly coats broad late regions (Rif1-asso-

ciated domains, RADs) in multicellular organisms

, consistent

with reduced helicase activation in late heterochromatin due to

locally high phosphatase activity in

Drosophila

. It also binds G-

quadruplex-like sequences

and is attached to the nuclear

lamina

. Rif1 regulates replication

fi

ring time as a possible

organizer of the chromatin architecture, and it has been proposed

that Rif1 is a molecular hub to co-regulate RT and nuclear

architecture

. After Rif1 KO in human mESC cells, RT changes

lead to chromatin modi

fi

cations and genome compartment

alterations, suggesting that Rif1 contributes to maintaining a

global epigenetic state

. But how these different roles of Rif1 can

be integrated with its central role in the eukaryotic replication

program regulation has not been explored so far. In

Xenopus

egg

extracts, Rif1 depletion led to damage-resistant DNA synthesis

and increased bulk DNA synthesis during normal S phase

Recently, we have shown that Rif1 depletion in early embryos

accelerates cell divisions and substantially increases the number

of active forks during the early S phase in

Xenopus

egg extracts

measured on single DNA

fi

bers by DNA combing

. However, we

did not investigate how Rif1 regulates the coordinated

fi

ring of

replication origins.

Stochasticity is commonly accepted to play an essential role in

the replication process of all eukaryotic organisms, from yeasts to

humans

. Early origins

fi

re, on average with a higher prob-

ability than late ones. To picture the complex process of DNA

replication, numerical models represent the replication as the sum

of spatio-temporal dynamics of limiting initiation factors and

initiation probabilities

–

. Using this representation, we mod-

eled the dynamics of the replication process in the unchallenged,

checkpoint-inhibited, or Plk1-depleted

Xenopus

in vitro

system

. This led us to propose that in this system, as in other

eukaryotes, the genome would be segmented into regions of low

and high probabilities of origin

fi

ring. These regions would

therefore be reminiscent of RT domains characterized in

differentiated cells.

Our model

’

s generality motivated us to use it as a framework to

analyze Rif1

’

s role in DNA replication. We performed an in-

depth analysis of our SM-data set from DNA

fi

bers with

numerical simulations and experimentally tested our model

predictions after Rif1 depletion in the

Xenopus

in vitro system.

We describe the dynamics of the Rif1 protein during early

embryonic cell divisions and the S phase and found that after Rif1

depletion, more replication clusters are activated on combed

DNA

fi

bers. In contrast, origin distances are only moderately

reduced within clusters, and the apparent fork speed is increased.

The analysis of the experimental SM-data by our previously

established model suggests that Rif1 depletion accelerates the

temporal replication program. The model also suggests that Rif1

depletion increases the number of limiting initiation factors and

regions with a high probability of origin

fi

ring; experiments

provided us with further support for these predictions. Therefore,

the dynamic and integrated picture of DNA replication we pro-

pose is in adequation with the Rif1-dependent control of the

replication program during rapid embryonic S phases.

Results and discussion

Rif1 dynamics during early development and the S phase in

Xenopus

. Studies in

Drosophila

embryos suggested that Rif1 is

required for the late replication of satellite sequences after the

mid-blastula transition (MBT)

, when a different replication

timing program is established at the onset of zygotic transcrip-

tion. We

fi

rst monitored the changes in Rif1 abundance in whole

Xenopus

embryos before and after MBT, which occurs between

cell cycles 12 and 13 in this organism

. We found that Rif1

protein levels were high from fertilization and during the early

cell cycles, decreased after the mid-blastula transition and

remained low until the late larval stages (Fig.

a, b). This intri-

guing observation suggested that the Rif1 abundance could play

an important role during these early embryonic cycles when cells

highly proliferate. We followed Rif1 recruitment to chromatin in

the ef

fi

cient, well-characterized in vitro replication system,

mimicking the

fi

rst S phase after fertilization (Fig.

c). Upon

ARTICLE

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

2

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

addition of sperm DNA to egg extracts, chromatin is assembled,

replication proteins are imported and recruited on chromatin,

and nuclei synchronously start DNA replication. We found that

Rif1 accumulation to chromatin paralleled that of the ORC-

complex, exhibiting a continuous increase until reaching a pla-

teau; this observation further corroborates a role for Rif1

throughout the S phase in this model system. Next, we

investigated the nuclear localization of Rif1 by immuno-

fl

uorescence during early and late S phases in the in vitro system.

In parallel, we pulse-labeled early and late replicating forks with

rhodamine-dUTP to follow whether Rif1 could co-localize with

ongoing replication (Fig.

d, e). We found that Rif1 staining was

not homogenous, co-localized with DNA, but not with early

replication forks, as shown in mouse cells

. However, during the

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

ARTICLE

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

3

late S phase, Rif1 staining at least partially overlaps with large

replicating chromatin regions. This raises the question of whether

Rif1 could participate in the temporal regulation of DNA repli-

cation in the

Xenopus

in vitro system, as shown in other

organisms.

Rif1 depletion increases initiation frequency and apparent fork

speed during the early S phase

. To better understand how Rif1

could regulate the replication program in the in vitro system, we

analyzed replication origin activation by DNA combing after

immunodepletion of Rif1 from egg extracts (Fig.

a). We pre-

viously focussed on comparing the role of Rif1 in reducing fork

density

. Here, we enrich the DNA combing experiments with

new data and a detailed, quantitative analysis in the absence of

Rif1. We found that Rif1 depletion boosted mean DNA synthesis

–

8-fold in the early S phase (50

–

60 min) versus a two-fold

increase mid-S phase (90

–

105 min) (Supplementary Fig. 1). This

may result from increased initiation events, fork speed, or both.

We calculated initiation frequencies (number of initiations per

time unit per unit length of unreplicated DNA),

I(t)

, and found

that initiations were about three-fold higher during the early S

phase after Rif1 depletion compared to the control versus a two-

fold increase at mid-S phase (Fig.

b). Thus, Rif1 depletion led to

a substantial increase in initiations, especially during the early S

phase. We, therefore, expected to observe a sharp decrease of

origin distances. However, we found that measured origin dis-

tances (eye-to-eye distances, ETEDs) remained unchanged during

the early S phase and only moderately decreased in mid S phase

(Fig.

c, median decrease 1.1

–

1.4 fold) in the absence of Rif1. The

apparent discrepancy between the global initiation frequency and

ETEDs measured on single

fi

bers can be explained by the limit in

fi

ber length set by DNA breaks or microscope

fi

eld and the

organization of 2

–

8 origins per cluster

. To study differences in

cluster activation in both conditions, we analyzed the distribution

of all replication tracks per

fi

ber as described

. Both distributions

contained an excess of

fi

bers with either no eye or multiple eyes

compared to a random distribution (Supplementary Fig. 2a, b),

consistent with the fact that origins are not activated indepen-

dently of each other but in clusters. Moreover, the distributions

from Mock and Rif1 depleted extracts differed signi

fi

cantly

(Fig.

value

=

4.77 10

−

, Mann

–

Whitney test). After Rif1

depletion, we observed that the percentage of

fi

bers without

replication tracks was reduced during the early S phase. Con-

versely, the percentage of

fi

bers containing more than 2 labeled

replication tracks per

fi

ber sharply increased. Importantly,

fi

bers

showing more than 5 tracks were 5

–

10-fold more frequent, sug-

gesting either larger clusters or more activated clusters in the

absence of Rif1 (representative

fi

bers in Supplementary Fig. 3a).

Given that incomplete eyes or gaps are excluded from ETED

measured on

fi

bers (or intra-cluster ETED), it follows that large

inter-cluster ETEDs have a higher probability of being excluded.

We thus also analyzed these excluded ETED (or inter-cluster

ETED) for both conditions (Supplementary Fig. 3b, Supplemen-

tary Table 1), using a method described

. We observed that these

distances were 5.3

–

1.2-fold larger from the very early to the mid S

phase, respectively, in controls compared to Rif1-depleted

extracts. This re

fl

ects that, depending on S phase progression,

5.3

–

1.2 more replication clusters were activated in the absence of

Rif1. The observed changes in the replication track patterns, the

inter-cluster distances (excluded distances), together with the

moderate changes in the intra-cluster ETED, suggest that

the initiation increase after Rif1 depletion was mainly caused by

the activation of whole replication clusters. In mammalian cells,

DNA

fi

ber analysis led to somehow contradictory observations in

the absence of Rif1. Whereas no change in origin distances was

observed in mouse cells

, an increase was detected in human cell

lines, attributed to larger chromatin loop sizes

; another study

described a decrease of origin distances due to the degradation of

Orc1 in the absence of Rif1

. These different observations across

species suggest that the effect of Rif1 at the level of single origins

appears minor compared to its strong global impact at the level of

chromatin domains.

Next, we noticed that replication eye length (EL) distributions

were signi

fi

cantly shifted towards larger sizes at the early S phase

after Rif1 depletion compared to the control (Fig.

e, 1.4

–

2-fold

median increase), but not during mid-S phase, pointing to an

increase in the apparent fork speed during early S phase. This

increase could be either due to a substantial increase in fork speed

or the consequence of new initiations close to active forks, which

led to replication eye merging. Some studies in mammalian cells

and

Drosophila

showed an increase in fork speed in the absence

of Rif1, measured by increased eye length or copy number

analysis

. However, other mammalian cells

’

DNA

fi

ber

studies revealed little or no changes in fork speed

.We

conclude that Rif1 depletion strongly accelerates DNA synthesis

in egg extracts by advancing the activation of replication clusters

at the beginning of the S phase, whereas it leads to only a

moderate increase in origin activation inside clusters and an

increase in the apparent fork speed.

Rif1 depletion globally accelerates DNA replication kinetics

and increases early replication foci

. To provide a more com-

prehensive picture of the role of Rif1 in DNA replication, we

fi

tted our recent minimal dynamic model of DNA replication,

which had proven its robustness to simulate the replication

process under different experimental conditions

, to our Mock

and Rif1 depletion combing data. This numerical model assumes

that origin

fi

ring is stochastic and is driven by the initial number

of the limiting replication factors

, interacting with potential

origins and the import rate

of these limiting replication factors

(Fig.

a). The genome needs to be divided into regions with a

high probability of origin

fi

ring,

, in the fraction

θ

of the

genome, and regions with a lower probability of

fi

ring,

out

located in the remaining fraction 1-

θ

. The fraction

θ

of the

genome corresponds to regions with ef

fi

cient and synchronous

origins, mainly early

fi

ring. In contrast, fraction 1-

θ

corresponds

to regions with inef

fi

cient, mostly dispersed, essentially late-

fi

ring

origins. Finally, replication forks promote other origin

fi

ring with

Fig. 1 Rif1 dynamics during early

Xenopus

development and in the S phase of the in vitro system. a

Time course analysis of Rif1 expression throughout

development; whole embryo protein extracts were analyzed by western blotting against indicated proteins, tubulin was used as loading control; MBT (

mid-

blastula transition), UF (unfertilized egg).

Quanti

cation of Rif1 abundance in three biological replicates and two technical replicates of western blot series

of embryonic whole cell extract, plotted as mean OD normalized to Tubulin, scaled using min and max: xscaled

=

−

xmin)/(xmax

−

xmin), Mean with

SED,

=

5 (data points).

Time course analysis during the S phase; sperm nuclei were incubated in

Xenopus

egg extracts and chromatin was isolated for

immunoblotting at indicated time points before and during DNA replication.

Rif1 nuclear localization during S phase; sperm nuclei were incubated in egg

extracts, rhodamine-dUTP was added at the beginning of the incubation and stopped during the early S phase (30 min), left panel, or for 2 min at the end of

S phase (60

–

62 min), right panel, reactions were stopped and nuclei were isolated and processed for immuno

uorescence.

Fluorescence intensity pro

plots along the indicated yellow line, to visualize colocalization between rhodamine and Rif1 for two example nuclei from (

ARTICLE

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

4

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

a probability

loc

within a distance

from the fork. For control

and Rif1 depleted conditions, the model well reproduced the

replication parameters (Supplementary Fig. 4a

–

c). We found that

all replication parameters changed when Mock and Rif1 deple-

tions were compared at the same time point (90 min), demon-

strating the dramatic effect of Rif1 depletion seen by DNA

combing experiments: on one hand,

θ

out

, and to a lesser extent,

and

signi

fi

cantly increased (Fig.

b). On the other hand,

, and

loc

decreased. Altogether, this suggests that Rif1 depletion

may lead to (i) an increase in the fraction of regions with highly

fi

cient, early origins (

θ

), (ii) an increase in origin

fi

ring for late

dispersed origins (

Pout

) and (iii) a small increase of origin

fi

ring

inside early regions (

). We decided to test experimentally

whether Rif1 depletion increases

θ

, meaning advances replication

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

ARTICLE

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

5

at the domain level. Replication foci probably correspond to the

genomic replication domains in metazoans

. In the absence of

Rif1, the replication foci pattern changed in mice and human

cells

. Temporally de

fi

ned, albeit more rudimentary, replica-

tion foci patterns can also be distinguished in the

Xenopus

in vitro

system

. We, therefore, performed a quantitative analysis of

replication foci after short pulse labeling with rhodamine-dUTP,

feasible at the very early S phase when foci can be quanti

fi

ed in

this experimental system. This analysis showed a signi

fi

cant, two-

fold increase of the mean replication foci number after Rif1

depletion compared to the control (Fig.

c, d, Supplementary

Fig. 5). It suggests that in the absence of Rif1, more chromatin

domains are activated early, in agreement with the effect in our

numerical model.

However, we found in our previous work that

θ

Pout

, and

model parameters also vary in the same range and dynamics

during normal S phase progression

. This suggests that Rif1

depletion would accelerate the S phase by compressing the

replication program without changing many replication para-

meters per se. In other words, Rif1 depletion would lead to a more

homogenous temporal program in the

Xenopus

system.

If this is true, local parameters should not differ after Rif1

depletion if they are compared at equivalent replication extents.

We, therefore, pooled DNA

fi

bers from the two replicates and all

time points, classi

fi

ed them according to their replication extent

(replicated fraction

(Fig.

a), and then compared different

replication parameters in the presence and absence of Rif1

(Fig.

b). No or only little signi

fi

cant differences in eye-to-eye

distances, eye lengths and fork density were observed between

Mock and Rif1 depleted conditions at the level of single

fi

bers in

this analysis. The initiation frequency increased slightly slower

after Rif1 depletion, which could illustrate a more homogenous

origin activation inside clusters. Altogether, these results strongly

suggest that Rif1 depletion affects the replication program by a

global acceleration at the level of chromatin domains but with

only few changes of replication parameters locally.

Rif1 depletion leads to an increase of DDK and origin

fi

ring

factors on chromatin

. Next, we asked whether in silico para-

meters would change after Rif1 depletion when compared at an

equivalent replication extent (Fig.

a, Supplementary Fig. 4a, c).

We found no signi

fi

cant increase for

Pout

and

θ

, again

consistent with a global acceleration after Rif1 depletion. How-

ever, the number of limiting initiation factors,

, still increased

in the absence of Rif1, suggesting that Rif1 lowers the number of

initial initiation factors independent of S phase progression.

It was uncertain whether Rif1 could regulate limiting initiation

factors, essential for maturing the pre-replication complexes into

the pre-initiation complexes in vertebrates

. To experimen-

tally test our unexpected in silico prediction, we analyzed by

western blotting the effect of Rif1 depletion on the binding to

chromatin of several origin

fi

ring factors (Treslin, MTBP,

TopBP1, RecQL4) and the S phase kinase Cdc7/Drf1 (DDK).

After Rif1 depletion, the total amount of chromatin-bound

Treslin, MTBP, Cdc7, Drf1, RecQL4, and Cdc45 signi

fi

cantly

increased in three replicates, whereas TopBP1 slightly decreased

throughout the S phase (Fig.

b, c, Supplementary Fig. 6a). The

increase of the pre-IC proteins Treslin/MTBP, RecQL4 and

Cdc45 is consistent with the increased origin activation we

observed after Rif1 depletion. The increased chromatin binding of

Cdc7/Drf1 after Rif1 depletion could be an indirect effect

related to Rif1

’

s role in regulating nuclear architecture, as it was

suggested in mammalian cells, where DDK accumulation was

observed to a lesser extent

than we show here in

Xenopus

Alternatively, it could also be due to a direct inhibition of DDK by

Rif1/PP1; in budding yeast, an interaction between Rif1 and Dbf4

has been observed, but the removal of Rif1 did not alter DDK

activity

. The moderate decrease of TopBP1, also observed in a

former study

, could be linked to its replication-independent

roles

. In the absence of Rif1 and PP1, slower migrating forms of

hyperphosphorylated MCM4 were detected, as demonstrated in a

former study in

Xenopus

. We have previously shown that

Treslin and MTBP are phosphorylated in a DDK-dependent

manner

. If Rif1 would recruit PP1 also to pre-initiation

complexes, we expected that MTBP/Treslin would remain

phosphorylated in the absence of Rif1/PP1. We found that the

slower migrating forms of Treslin and MTBP were enriched on

chromatin compared to the quicker migrating forms in the

absence of Rif1 in Fig.

b and in Fig.

d, respectively.

Dephosphorylation by

-phosphatase treatment con

fi

rmed that

the slower migrating bands after Rif1 depletion represents a

phosphorylated form of MTBP or Treslin (Supplementary

Fig. 6b). These results show that the phosphorylation state of

Treslin and MTBP depends on the presence of Rif1 in

Xenopus

similar effect was described in budding yeast for the Treslin

homolog Sld3

. This observation could be explained by Rif1

recruiting PP1 or another phosphatase on the pre-initiation

complex. Alternatively, maybe not exclusive, this could be the

consequence of a Rif1-dependent inhibition of DDK recruitment

(or activity), since an enrichment of DDK was observed after Rif1

depletion. The signi

fi

cance of the DDK-dependent phosphoryla-

tion and the Rif1-dependent dephosphorylation on Treslin/

MTBP activity remains to be explored in future studies; MTBP

phosphorylation by CDK or checkpoint kinases regulates origin

fi

ring, possibly by controlling the stability of the MTBP-Treslin

complex

. Altogether, our observation of proteins enriched

on chromatin after Rif1 depletion provides strong support for the

prediction of the numerical model that Rif1 limits initiation

factors. We were not able to generate a recombinant form of Rif1

that could rescue the depleted extracts. However, vertebrate Rif1

is a very large protein and is dif

fi

cult to produce in an active form

in recombinant systems

. Immunodepletion of Rif1 does not

noticeably decrease the abundance of TopBP1 and other known

checkpoint regulators in egg extracts

, and two different

proteomics approaches in

Xenopus

and mouse have not revealed

Rif1 interactions with known negative replication regulators other

Fig. 2 Rif1 depletion strongly increases origin activation in the early-S phase at the level of replication clusters. a

Principle of DNA combing experiment

with a

ber example and measured parameters. Sperm is replicated in control (

Mock) or Rif1 immunodepleted (

Rif1) egg extract in the presence of

biotin dUTP, DNA was isolated at the indicated times, and then DNA combing was performed in two independent experiments.

Mean initiation

frequencies (

I(t)

) with standard error of mean with SEM,

=

2 (data points) and ratio

Rif1/

Mock (

I(t)

) were calculated.

Scatter dot plot of eye-to-eye

distances (ETED) at different time points from replicate 1, median with interquartile range, Mann

–

Whitney test.

Percentage of unreplicated

bers and

bers with increasing number of replication tracks per

ber from early S phase from both independent experiments, with an equivalent

ber length

distribution in mock and Rif1 depleted condition; percentage ratios indicated below each class. (

=

2519

bers for each distribution, Mann

–

Whitney test on

distribution,

value 4 × 10

−

Scatter dot plot of eye lengths (EL), replicate 1, median with interquartile range, Mann

–

Whitney test; n indicated below,

ns: non-signi

cant, * indicates signi

cant difference (Mann

–

Whitney U test, two-sided,

< 0.05:

values: * 0.01

–

0.05; ** 0.001

–

0.01; *** 0.0001

–

0.001;

**** <0.0001).

ARTICLE

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

6

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

Fig. 3 Rif1 depletion changes all in silico replication parameters in numerical simulations and increases the number of early replication foci. a

Diagram

of the numerical model

with parameters described in the text.

Inferred model parameters by

tting

Mock and

Rif1 depleted combing data of two

replicates for the same time interval (90 min). Circles indicate the mean value of the parameters over 100 different runs of the genetic algorithm; the

error

bars are standard deviations, (*) indicates a signi

cant difference between the

Mock and

Rif1 samples,

χ

test.

Nuclei were incubated in Mock or Rif1

depleted extracts in the presence of rhodamine-dUTP and stopped early in S phase, representative images of nuclei by

uorescence microscopy.

Scatter

dot plot of early replication foci number per nucleus, mean with SD, two-tailed Mann

–

Whitney test.

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

ARTICLE

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio

7

than PP1

. Finally, we analyzed MTBP protein levels in pre-

and post-MBT embryos. We found that MTBP abundance is

constant but declines after the MBT (Supplementary Fig. 7a, b),

as shown in a proteomic study

, when the S phase lengthens and

the replication program slows down. This result suggests that

MTBP could be, like its partner Treslin and other initiation

factors

, concentration limiting for the replication program

during Xenopus development.

The results of our study have several important implications for

the replication program and the role of Rif1 in early embryos. First,

the early embryonic S phase is not running at its fastest biochemical

rate and can be accelerated by Rif1 depletion in vitro and in vivo

(this study and

). Second, origin

fi

ring factors such as MTBP/

Treslin, Drf1, and RecQL4 do not seem to be concentration

limiting per se in

Xenopus

egg extracts since, in the absence of Rif1,

fi

ring factors and Drf1 are recruited to chromatin, and many

more origins can be activated. Rather than being concentration

limited, we propose that DDK and

fi

ring factors might be excluded

from chromatin by Rif1-compacted chromatin domains, or their

activity might be regulated at some origins or regions by Rif1-

dependent PP1 targeting. Another possibility could be that self-

organizing Rif1-PP1 hubs break down when DDK/CDK levels

reach a threshold level leading to a more or less synchronized origin

fi

ring in later replicating sequences as proposed for satellite

sequences in

Drosophila

. It has been suggested that one conserved

role of the RT program in eukaryotes is to adapt the optimal

number of replication forks to the number of available initiation

factors

. In the absence of Rif1 in yeast and mammalian cells, some

early origins or regions

fi

re late, consistent with the hypothesis

of the exhaustion of initiation factors as an indirect result

of unscheduled origin

fi

ring

. In egg extracts, the stockpile of

maternal proteins may be suf

fi

ciently enriched in initiation factors

Fig. 4 Rif1 depletion does not intrinsically change origin distances or eye lengths. a

Principle of analysis: All DNA

bers were grouped into six classes

according to their replicated fraction f

-f

Box and whisker plots for initiation frequency (

I(f)

), fork density, ETED and EL after Mock and Rif1 depletion for

each replicated fraction class, with medians (black lines) and means (black points), upper and lower quartiles, min and max, n indicated below, * indi

cates

signi

cant difference, Mann

–

Whitney U test, two-sided.

ARTICLE

COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-023-05172-8

8

COMMUNICATIONS BIOLOGY

| (2023) 6:788 | https://doi.org/10.1038/s42003-023-05172-8 | www.nature.com/commsbio