Discussion
During evolution, organisms have accumulated paralogues. As this is a
widespread phenomenon, a question was raised regarding the functions of
paralogues, do they serve as backups, or each paralogue has acquired a
unique function? Here we focus on two paralogues, Rat1 functions mainly,
if not exclusively, in the nucleus, whereas Xrn1 shuttles between the
nucleus and cytoplasm and accumulates in the cytoplasm. As these
paralogues are active in different compartments, the issue of their
potential capacity is especially interesting. It was proposed that the
two paralogues are fully replaceable, and their distinct functions are
attributed to their cellular localization (Johnson, 1997). Xrn1 is a
highly conserved and exceptionally large protein (175 kDa in yeast) with
multiple biological functions in many eukaryotes. Many of these
functions can be attributed to its enzymatic 5’→3’ exonuclease activity
which acts on several types of RNAs. Rat1 is smaller (116 kDa in yeast)
and also functions in several nuclear RNAs decay pathways, including
those linked to RNA pol II-coupled mRNA transcriptional termination (Kim
et al., 2004). Both Xrn1 and Rat1 substrates are single-stranded
unstructured 5’Phosphate end with lengths of at least four nucleotides
(Nagarajan et al., 2013; Basu et al., 2021). Many of Xrn1 functions
require only two-thirds of the molecule, which is homologous to Rat1.
The last Xrn1 C-terminal third (CTD) is apparently unstructured, less
conserved, and not important for some functions (Bashkirov et al.,
1995). CTD is necessary for Xrn1 interaction with eIF4G to regulate the
translation of certain membrane-protein-encoding mRNAs (Blasco-Moreno et
al., 2019). It also serves as an interaction platform for other proteins
in higher eukaryotes (Braun et al., 2012). Being unstructured, it is
likely involved in Xrn1 capacity to reside in liquid-liquid
phase-separated droplets (Currie et al., 2020). To perform nuclear
functions, Xrn1 shuttles between the cytoplasm and nucleus (Haimovich et
al., 2013; Medina et al., 2014; Pérez-Ortín & Chavez, 2022). Nuclear
import depends on two NLSs whereas export to the cytoplasm is proposed
to be mediated by its ability to bind or “imprint” mRNAs
(Chattopadhyay et al., 2022). Shuttling and mRNA imprinting have been
shown to affect both mRNAs synthesis and decay (Chattopadhyay et al.,
2022). Thus, although these two paralogues are very similar and both
have highly conserved domains that carry 5’→3’ exonuclease activity,
their cellular localization is different. This raised the question
whether this difference in localization is the main reason for their
disparate functions. This question was previously addressed by A.
Johnson (1997), who conclusdedthat they are functionally
interchangeable. Because we have gained more information about these
paralogues since 1997, we decided to reevaluate this conclusion. The
above question is important for understanding the functions of these two
paralogues. Moreover, addressing this issue can also serve as a paradigm
for other paralogues with identical enzymatic activity.
To this end, we investigated whether the cytoplasmic function of Xrn1
could be performed by its nuclear counterpart Rat1. We found that a
cytoplasmic version of Rat1 (cRat1) can only partially complement the
5’→3’ exonuclease activity of the missing Xrn1 in the cytoplasm. Even
fusing the Xrn1 CTD to the C-terminus of cRat1 (cRat1-Cterm), did not
improve cRat1 performance. Specifically, although cRat1 could mostly
recover cell volume, growth rate and global mRNA decay, it fails to
fully recover co-translational mRNA decay. Previously, we found that
deletion of XRN1 resulted in defective transcription, which was
attributed to both direct and indirect effects (Haimovich et al., 2013;
Medina et al., 2014; Chattopadhyay et al., 2022). Direct effects should
be the result of the loss of the direct function of Xrn1 in
transcription activation (Xrn1 is a component of the pre-initiation
complex), whereas indirect effects are probably the consequence of slow
growth of the strain. Given the observation that Rat1 recovers the
growth defect of xrn1 cell, the cRat1-mediated increase in
transcription of these xrn1Δ cells might be due to the growth
rate effect. Interestingly, cRat1 could slightly, but significantly,
increase transcription in wild-type cells. cRat1 might compete with Xrn1
on binding the substrates. This possible competition could provide a
plausible explanation for this transcriptional increase in wild-type
cells, as cRat1 could increase the amount of free Xrn1, capable of
importing to the nucleus and activating transcription. In addition, it
is possible that mRNA decay per se enhances transcription (e.g.,
by supplying more nucleotides).
In the absence of XRN1 , cRat1 can provide some mRNA buffering
activity (Figure 1A). Hence, it seems that a partial mRNA buffering can
be accomplished without Xrn1 and without its specific shuttling feature.
Interestingly, we found that compensation introduced by cRat1 is
proportional to the defect caused by the absence of Xrn1 (Figure S2).
Maybe, the observation that decapping rate is one of the limiting
activities in mRNA decay (Parker 2012) is relevant for this correlation.
We propose that mRNA degradation per se can provide some
buffering, probably together with other pathways and partners that act
in concert or in parallel. In particular, our results suggest that the
transcription activation capacity of Xrn1 is not absolutely required for
buffering (assuming that Rat1 does not have this function) However,
since this buffering is partial, it is clear that mRNA decay is not
sufficient to provide full buffering. Maybe the transcription activation
and the specific shuttling features of Xrn1 are required for efficient
buffering.
The effect of cRat1 expression on the start of 5’→3’ mRNA decay after
decapping (HT-5Pseq profiles and the 3’/5’ index) of xrn1Δ cells
is partial. Using heatmaps of the HT-5Pseq data, we noticed that in
wild-type cells, the peak at the TSS was clearly less strong than in thexrn1Δ cells (Figures 2F and S4), which suggests that Xrn1 is
located in the vicinity of the RNA 5’ end, and is ready to act as soon
as the cap is removed. cRat1 did not significantly improve thexrn1Δ pattern. The 3 nt degradation pattern inside the ORFs in axnr1Δ is lost. Taking the relative FPI value of this mutant
lacking any cytoplasmic 5’→3’ exoribonuclease as a bottom reference
(0.79) and the wild-type value as 1, the cRat1 strain only increases the
FPI value by 14% (up to 0.82) indicating that cRat1 is not a good
substitute for Xrn1 closely following the ribosome position during
co-translational mRNA decay. Given that Cterm of Xrn1 does not improve
the function of cRat1 we conclude that the N-terminal two-thirds of Xrn1
contain all the determinants needed for 5’→3’co-translational decay.
This idea is supported by the fact that the Xrn1 interacts with
ribosomes via the 1-772 amino acids region and no interaction is
detected with its Cterm domain (Tesina et al., 2019). The reason why
cRat1 can hardly replace Xrn1 in co-translational decay remains to be
determined. It is possible that the imprinting capacity of Xrn1
(Chattopadhyay et al., 2022) plays some role. The association of the
imprinted Xrn1 with mRNA also in the cytoplasm may localize Xrn1 near
the 5’ end of the mRNA, and upon decapping it is available immediately
for degradation. We note that the expression level of cRat1 is probably
lower than the cytoplasmic level of Xrn1 in the wild type. However,
given that it has been shown that the cytoplasmic level of Xrn1 is not
limiting for 5’→3’ mRNA decay (Chattopadhyay et al., 2022) we consider
that this is not the reason of the differences found between cRat1 and
wild type cells.
XRN1 and RAT1 are paralogous genes that are not derived
from the whole-genome duplication (Wolfe, 2015) but from an older
small-scale duplication (Fares et al., 2013). They show intermediate
homology divergence and are present in most eukaryotes from yeast to
humans (Parker & Song, 2004), suggesting a long time for sequence
divergence. However, they can reciprocally rescue each other’s lethal
phenotypes in yeast (Johnson, 1997). Here we confirm that their
exonuclease activities are interchangeable. Nevertheless, it seems that
the specialization of the two 5’→3’ exonucleases makes them different in
terms of the interactive partners they bind to, perhaps affected by the
presence or absence of Xrn1 Cterm. However, the addition of Xrn1-Cterm
to Rat1 was insufficient to make it fully interchangeable with Xrn1.
Therefore, it is important to note that, their subcellular distribution
and dynamics are critical for their differential activities. Rat1 is
predominantly nuclear whereas Xrn1 is largely cytoplasmic (Sharma et
al., 2022; Johnson, 1997; Haimovich et al., 2013). Rat1 can substitute
for Xrn1 when its NLS is deleted and becomes cytoplasmic (Sharma et al.,
2022, this paper) whereas Xrn1 can only complement the rat1mutant when it overexpressed from a multicopy plasmid and is fused to a
strong SV40 large-T-antigen NLS (Johnson, 1997). Xrn1, moreover, has the
capacity to shuttle. Thus, the number of nuclear and cytoplasmic 5’→3’
exonuclease molecules and the capacity to shuttle determine the
functional difference between these two exonucleases. These features
should have appeared early during eukaryotic evolution and made it
impossible to completely substitute one protein for the other, despite
their similar and partially interchangeable exonuclease activities.