betaFeedback

WorksAgricultural and Biological Sciences

Molecular memory reduces timing noise in stochastic gene expression

Abstract

Fluctuations in gene expression affect the timing precision of downstream events triggered when a gene product reaches a critical threshold. The first-passage time (FPT) to this threshold characterizes the timing of such events. In conventional single-step birth–death models, reaction waiting times are exponentially distributed and hence memoryless. In this setting, FPT noise exhibits a U-shaped dependence on the threshold, attaining its minimum near one-half of the steady-state mean. However, mRNA production and degradation often involve multiple sequential steps, leading to hypoexponential distributions of mRNA production times and mRNA lifetimes. Here, we investigate how molecular memory arising from multistep mRNA production and degradation affects FPT statistics in bursty gene expression. We calculate FPT statistics in the small-noise limit. Both multistep mechanisms reduce FPT noise, but through distinct mechanisms. Multistep burst arrival reduces mRNA copy-number fluctuations, thereby suppressing timing noise. Multistep degradation instead reshapes transient accumulation dynamics, reducing relative timing noise even when steady-state mRNA noise increases. The threshold that minimizes FPT noise remains close to one-half of the steady-state mean, except for multistep burst arrival with constant burst sizes, for which it shifts to lower values. Stochastic simulations support the analytical predictions within their regimes of validity and show that these qualitative conclusions persist beyond the small-noise approximation.

Introduction

First-passage processes provide a unifying framework for stochastic phenomena in which a fluctuating variable first reaches a prescribed boundary. Such problems arise throughout statistical physics, from diffusion-controlled reactions and target search to escape processes and neuronal firing [1, 2]. The same framework is particularly natural for intracellular processes, where a cellular event is often initiated when the abundance of a regulatory molecule crosses a critical threshold [5] [3, 4]. When the regulatory molecule is a gene product, fluctuations in its threshold-crossing time arise from the intrinsic stochastic biochemical reactions underlying gene expression [6]. The noise in the corresponding first-passage time (FPT) therefore quantifies the temporal precision of the downstream event. In simple memoryless birth–death models of gene expression, for which reaction waiting times are exponentially distributed, FPT noise depends nonmonotonically on the normalized threshold and is minimized when the threshold is close to one-half of the steady-state mean [7] [4]. This nonmonotonic behavior has also been observed experimentally in bacteriophage λ lysis timing by altering the threshold through sequence mutations in holin [8].

The study of threshold-crossing dynamics has expanded beyond the minimal birth–death model to encompass more complex biological processes and regulatory mechanisms. FPT approaches have been applied to bacterial cell-division control and to gene-expression models incorporating promoter switching, external disturbances, and cell growth [9, 10, 11, 12]. Experimental studies of bacteriophage λ lysis have further demonstrated how the holin threshold and translational efficiency influence lysis-time statistics [8, 13]. The role of regulatory mechanisms involving feedback loops, post-transcriptional regulation, and molecular sequestration in shaping FPT statistics has also been extensively investigated [14, 15, 16, 17, 18, 19]. Collectively, these studies have advanced our understanding of how stochastic gene-expression dynamics and cellular regulation influence the timing of intracellular events.

Despite these advances, most theoretical treatments of FPT statistics still model molecular production and degradation as single-step, memoryless processes with exponentially distributed waiting times. Experimental observations and mechanistic models, however, support multistep descriptions of both transcription and mRNA degradation [20, 21, 22]. When the intermediate states are coarse-grained, the total waiting time for a sequence of exponential steps is generally hypoexponential, introducing effective molecular memory [23, 24]. At the production level, the intermediate states may correspond to transcription initiation, pre-mRNA processing, and nuclear export; at the degradation level, they may correspond to deadenylation, decapping, and exonucleolytic decay [25, 26, 27]. Queueing-theoretic approaches have related general interarrival-time and lifetime distributions to steady-state copy-number moments [28]. However, how memory in burst arrival and mRNA degradation reshapes FPT statistics remains largely unexplored.

Here, we investigate how two sources of molecular memory—multistep burst arrival and multistep mRNA degradation—reshape FPT statistics in stochastic gene expression. We derive transient moments of the mRNA abundance from the chemical master equation and use them to obtain analytical expressions for the mean and noise of the FPT in the small-noise limit. For multistep burst arrival, the analytical calculation of moment dynamics assumes that the mean interburst interval is short relative to the mRNA lifetime, a regime relevant to many mammalian genes [29, 30, 31, 32]. At a fixed mean burst arrival time and mean mRNA lifetime, both multistep burst arrival and multistep degradation reduce FPT noise, but through distinct mechanisms. Multistep burst arrival regularizes production and lowers copy-number noise, thereby directly suppressing timing noise.

Multistep degradation accelerates the mean mRNA response, increasing its logarithmic sensitivity to time at a fixed threshold and thereby allowing relative timing noise to decrease even when steady-state copy-number noise increases. For constant burst sizes, multistep arrival shifts the optimal normalized threshold below the Markovian value X0≃0.55X_0 \simeq0.55, whereas multistep degradation leaves the optimum close to or above one-half. Stochastic simulations support the analytical predictions within their expected regimes and show that the qualitative trends persist beyond the strict timescale-separation regime. Thus, burst-arrival and degradation memory provide distinct routes for controlling both temporal precision and the optimal triggering threshold.

SymbolDescriptionSymbolDescription
kkmRNA production rateBBBurst size
μ\mumRNA degradation rateNN# sequential transcriptional steps
x(t)x(t)Total mRNA copy number at time ttRR# mRNA degradation steps
⟨x(t)⟩\langle x(t)\rangleMean of mRNA at tt⟨xs⟩\langle x_s\rangleMean steady-state of mRNA
CVx(t)\mathrm{CV}_{x}(t)Coefficient of variation of mRNA at time ttCVxs\mathrm{CV}_{x_s}Steady-state CV of mRNA
σ2\sigma^2VarianceFFFano factor
XXThreshold number of mRNAX0X_0Minimum normalized threshold Xmin⁡/⟨xs⟩X_{\min}/\langle x_s\rangle
TTFirst passage timeCVT\mathrm{CV}_{T}Coefficient of variation of FPT

Table 1. Summary of notation used

First-Passage-Time Framework

We define the first-passage time (FPT) as the minimum time required for the mRNA copy number to reach a critical threshold for the first time. Starting from zero initial conditions, the FPT is

T:=inf⁡{t:x(t)≥X∣x(0)=0},(1)T := \inf\{t : x(t) \ge X|x(0)=0\}, \tag*{(1)}

where XX is the threshold level required to trigger a downstream event. Since gene expression is stochastic, TT varies between realizations. When copy-number fluctuations are small, the mean FPT, ⟨T⟩\langle T\rangle, can be approximated by the time at which the mean mRNA trajectory reaches the threshold,

⟨T⟩≈t∣⟨x(t)⟩=X.(2)\langle T\rangle\approx t|_{\langle x(t)\rangle= X}. \tag*{(2)}

Timing noise is quantified by the squared coefficient of variation of the FPT, CVT2=σT2/⟨T⟩2\mathrm{CV}_{T}^{2} = \sigma_{T}^{2}/\langle T\rangle^{2}, the ratio between variance and mean squared. Using the small-noise approximation [33] the FPT noise can be written as

CVT2≈[(Sx)−1CVx(t)]t=⟨T⟩2,where Sx:=dlog⁡⟨x(t)⟩dlog⁡(t)(3)\mathrm{CV}_{T}^{2} \approx\left[(S_x)^{-1}\mathrm{CV}_{x}(t)\right]_{t=\langle T\rangle}^{2}, \qquad\text{where } S_x := \frac{d\log\langle x(t)\rangle}{d\log(t)} \tag*{(3)}

is the logarithmic sensitivity of the mean mRNA level with respect to time. We emphasize that the formula given by Eq. (3) is valid in the limit of small fluctuations in copy number. Nevertheless, in the small-noise limit, the FPT noise admits a clear physical interpretation through its decomposition into two contributions: the first is determined by the sensitivity of the mean copy-number dynamics to time, while the second is determined by the relative copy-number fluctuations at the mean FPT. While more general approaches have been developed to approximate FPT distributions directly from master equations or Fokker–Planck equations [34], our analysis focuses on the relationship between FPT noise and mRNA moment dynamics. For the multi-state burst-arrival and multistage-degradation processes considered here, we validate the small-noise approximation directly against exact SSA simulation results.

Multistep Burst Arrival Reduces Timing Noise

We first examine how molecular memory in the burst-arrival process affects mRNA fluctuations and, consequently, first-passage-time statistics. A simple way to introduce such memory is to represent the interburst cycle by NN sequential transcriptional states, D0→D1→⋯→DN→D0+mRNAD_0 \to D_1 \to\cdots\to D_N \to D_0 + \text{mRNA}, as shown schematically in Figure 1(a). Here, DiD_i denotes the ithi^{\mathrm{th}} transcriptional state, and the completion of the final transition produces a burst of BB mRNAs, which undergo single-step degradation at rate μ\mu. For NN sequential exponential steps, each with the same rate kk, the interburst time follows an Erlang distribution with mean N/kN/k, and coefficient of variation squared 1/N1/N. The states DiD_i are coarse-grained kinetic states, not necessarily distinct molecular species. Depending on the biological system, they may capture delays associated with promoter remodeling, assembly of the transcriptional machinery, and the transition to productive transcription [22, 24]. The equal-rate irreversible chain is therefore an effective model of the aggregate waiting time and does not imply that the underlying biochemical steps are identical or map one-to-one onto the DiD_i states.

Multistep burst arrival suppresses both mRNA abundance noise and first-passage-time noise

Figure 1. Multistep burst arrival suppresses both mRNA abundance noise and first-passage-time noise. (a) A schematic of NN-state transcription process. The nodes DiD_i, i=0,…,N−1i = 0, \ldots, N - 1, denote successive transcriptional states, and transitions occur at rate kk. Completion of the final transition produces an mRNA burst of size BB and resets the system to D0D_0. Individual mRNAs undergo single-step degradation at rate μ\mu. (b,c) Representative stochastic mRNA trajectories and corresponding first-passage-time distributions for (b) N=1N = 1 and (c) N=10N = 10, evaluated at the same event threshold. Increasing NN reduces mRNA noise and narrows the timing distribution. (d–h) Results for constant burst. (d) Mean mRNA abundance and (e) squared coefficient of variation, CVx2(t)\mathrm{CV}^{2}_{x}(t), as functions of dimensionless time μt\mu t. (f) Dimensionless mean FPT, μ⟨T⟩\mu\langle T\rangle, and (g) FPT noise, CVT2\mathrm{CV}^{2}_{T}, as functions of the normalized threshold X/⟨xs⟩X/\langle x_s\rangle. (h) Optimal normalized threshold, X0X_0, where FPT noise is minimum, as a function of NN for ⟨B⟩=1\langle B\rangle= 1 and 33. (i–m) Corresponding results for zero-inclusive geometrically distributed bursts. (i) Mean mRNA abundance, (j) mRNA noise, (k) dimensionless mean FPT, (l) FPT noise, and (m) optimal normalized threshold. Solid curves show the analytical approximations in Eqs. (3), (4), (5), and (7), and symbols show stochastic simulation results [39]. Parameter values for these plots: ⟨B⟩=1\langle B\rangle= 1, ⟨xs⟩=50\langle x_s\rangle= 50 and μ=0.1\mu= 0.1, and k=⟨xs⟩μN/⟨B⟩k = \langle x_s\rangle\mu N/\langle B\rangle to keep ⟨xs⟩\langle x_s\rangle fixed.

To obtain the transient abundance statistics, we derive the moment equations from the joint chemical master equation for the transcriptional state and the mRNA copy number. The detailed calculation is provided in Appendix Sec. A. The resulting mean and noise (coefficient of variation squared) dynamics are given as follows:

⟨x(t)⟩=⟨xs⟩(1−e−μt),⟨xs⟩=k⟨B⟩μN(4)\langle x(t) \rangle= \langle x_s \rangle\left(1-e^{-\mu t}\right), \qquad\langle x_s \rangle= \frac{k\langle B\rangle}{\mu N} \tag*{(4)}
CVx2(t)=12⟨xs⟩[1+⟨B2⟩⟨B⟩+2⟨B⟩(1(1+μ/k)N−1−⟨xs⟩⟨B⟩)](1+e−μt1−e−μt)−e−μt⟨x(t)⟩.(5)CV_x^2(t) = \frac{1}{2\langle x_s\rangle}\left[1+\frac{\langle B^2\rangle}{\langle B\rangle}+2\langle B\rangle\left(\frac{1}{(1+\mu/k)^N-1}-\frac{\langle x_s\rangle}{\langle B\rangle}\right)\right]\left(\frac{1+e^{-\mu t}}{1-e^{-\mu t}}\right)-\frac{e^{-\mu t}}{\langle x(t)\rangle}. \tag*{(5)}

In the steady-state limit (t→∞t\to\infty), our noise expression CVx2CV_x^2 becomes

CVxs2=12⟨xs⟩[1+⟨B2⟩⟨B⟩+2⟨B⟩(1(1+μ/k)N−1−⟨xs⟩⟨B⟩)],(6)CV_{x_s}^2 = \frac{1}{2\langle x_s\rangle}\left[1+\frac{\langle B^2\rangle}{\langle B\rangle}+2\langle B\rangle\left(\frac{1}{(1+\mu/k)^N-1}-\frac{\langle x_s\rangle}{\langle B\rangle}\right)\right], \tag*{(6)}

which agrees with the previous queueing-theoretic results for bursty gene expression with Erlang-distributed interburst times [28, 35, 36, 37]. For a fixed mean interburst time, hence for a fixed steady-state value of the mean mRNA number (xsx_s), the mean dynamics in Eq. (4) do not depend on the number of burst arrival steps NN (see Figure 1 d, i, respectively, for constant and geometric burst distributions). However, the noise in the mRNA level in Eq. (6) decreases with NN for any burst distribution. To compare the mean and noise dynamics of mRNA levels with stochastic simulation, we consider constant and a geometric burst size distribution (see Figure 1 d, e, i, and j). For constant burst size, each production event generates a fixed number of mRNAs, B=⟨B⟩B = \langle B\rangle, such that ⟨B2⟩=⟨B⟩2\langle B^2\rangle= \langle B\rangle^2. In particular, B=1B = 1 corresponds to non-bursty production, where each event produces a single mRNA. For geometrically distributed burst sizes, we consider the zero-inclusive geometric distribution, Pr⁡(B=b)=p(1−p)b\Pr(B = b) = p(1-p)^b, for b=0,1,2,…b = 0,1,2,\ldots, where p=1/(1+⟨B⟩)p = 1/(1+\langle B\rangle) and ⟨B2⟩=⟨B⟩(1+2⟨B⟩)\langle B^2\rangle= \langle B\rangle(1+2\langle B\rangle).

Using the mean mRNA dynamics in Eq. (4), the mean FPT, the average time to reach a threshold mRNA level XX, can be obtained from Eq. 2 as

⟨T⟩≈−1μln⁡(1−X⟨xs⟩).(7)\langle T\rangle\approx-\frac{1}{\mu}\ln\left(1-\frac{X}{\langle x_s\rangle}\right). \tag*{(7)}

The mean FPT for a given XX is independent of burst arrival steps NN, which agrees with exact stochastic simulation as well for both constant and geometric burst distributions (see Figure 1 f, k).

Since the FPT noise in Eq. (3) depends directly on the mRNA noise, the reduction in copy-number fluctuations is expected to improve the precision of threshold-crossing times. This effect is illustrated in Figure 1(b,Figure 1c), where sample trajectories cross the same threshold with substantially less temporal dispersion for N=10N = 10 than for N=1N = 1. The corresponding timing histogram is narrower for larger NN, indicating reduced FPT variability. The quantitative dependence of FPT noise on the number of burst-arrival steps is shown in Figure 1 for deterministic and geometric bursts, in panels (g) and (l), respectively. For both cases, increasing NN lowers CVT2CV_T^2. However, the dependence of the optimal threshold on NN is strongly influenced by the burst-size statistics. Figure 1(h) and (Figure 1m) show the optimal threshold X0X_0, defined as the value of the normalized threshold X/⟨xs⟩X/\langle x_s\rangle at which CVT2CV_T^2 is minimized, as a function of NN for constant and geometric bursts, respectively. For constant burst, increasing the number of transcription steps NN shifts X0X_0 below the classical value near 0.5. In contrast, for geometrically distributed bursts, increasing NN lowers the overall magnitude of the FPT noise while leaving the optimal threshold close to one half of the steady-state mean.

So far, we have considered a fixed mean burst size, ⟨B⟩=1\langle B\rangle= 1. To examine how burst magnitude modifies FPT statistics, we vary both ⟨B⟩\langle B\rangle and NN for constant and geometric burst-size distributions in Appendix Fig. 4. For both distributions, increasing ⟨B⟩\langle B\rangle increases FPT noise, min⁡(CVT2)\min(CV_T^2), which is also consistent with previous FPT studies [38]. At a given NN and ⟨B⟩\langle B\rangle, geometric bursts produce larger FPT noise than constant burst because of the additional fluctuations in burst size. The optimal threshold X0X_0 also depends on mean burst size, ⟨B⟩\langle B\rangle, for both constant and geometric burst distributions. In particular, X0X_0 shifts toward higher values as ⟨B⟩\langle B\rangle is increased. Thus, the effects of NN and ⟨B⟩\langle B\rangle compete: increasing the number of burst-arrival steps reduces FPT noise and tends to shift the optimal threshold downward, whereas increasing the burst size enhances FPT noise and shifts the optimum toward higher thresholds. These results extend the fixed-⟨B⟩\langle B\rangle findings and show that the precision of threshold-crossing times and the location of the optimal threshold are jointly controlled by burst-arrival regularity and burst magnitude.

Multistep Degradation Reduces Timing Noise

Having considered molecular memory in burst arrival, we next investigate memory generated by multistep mRNA degradation, as illustrated in Figure 2(a). Standard coarse-grained models represent mRNA degradation as a single first-order reaction, resulting in an exponentially distributed and memoryless transcript lifetime. In cells, however, mRNA degradation can involve several successive biochemical processes, including deadenylation, decapping, and exonucleolytic decay. Such pathways can produce nonexponential, age-dependent lifetime distributions [21]. The resulting molecular memory can substantially modify gene-expression fluctuations, as demonstrated in theoretical and experimental studies of general lifetime distributions and multistage gene-expression models [20, 22, 40].

Multistep mRNA degradation model and results for constant and geometric burst size distributions

Figure 2. For bursty production, multistep mRNA degradation increases copy-number noise but reduces FPT noise. (a) Schematic of the multistep-degradation model. Bursts arrive with rate kk and average size ⟨B⟩\langle B\rangle, and each mRNA passes through RR sequential degradation steps, each with transition rate μ\mu, before removal. The total mRNA abundance, x(t)x(t), is the sum over all degradation states. (b,c) Mean mRNA dynamics and (d,e) corresponding copy-number noise, CVx2(t)\mathrm{CV}_x^2(t), for constant and geometric burst size distributions, respectively, with ⟨B⟩=1\langle B\rangle= 1. Solid curves show the exact analytical results in Eqs. (8) and (10), and symbols denote stochastic simulation algorithm (SSA) [39] results for R=1R = 1, 5, and 10. (f,h) Mean FPT as a function of the normalized threshold X/⟨xs⟩X/\langle x_s\rangle. Solid curves show the threshold-crossing estimates obtained from the exact mean dynamics using Eq. (2), while symbols denote SSA results. (g,i) FPT noise, CVT2\mathrm{CV}_T^2, obtained from SSA matches well with Eq. (3) under small noise approximation. We keep the mean mRNA lifetime (R/μR/\mu) fixed with RR by changing μ\mu. The parameter values are ⟨xs⟩=50\langle x_s\rangle= 50 and μ=R/10\mu= R/10 to keep the mean mRNA lifetime fixed at 10 minutes. The burst-arrival rate is chosen as k=⟨xs⟩μ/(R⟨B⟩)k = \langle x_s\rangle\mu/(R\langle B\rangle) to keep the steady-state mean fixed.

In our model, transcriptional bursts arrive according to a Poisson process with rate kk and burst size BB. Newly produced transcripts enter the first degradation state, M0M_0, and subsequently pass through RR sequential states, M0,…,MR−1M_0,\ldots,M_{R-1}, before removal. All transitions between degradation states, including removal from the final state, occur with rate μ\mu. The total mRNA abundance xx includes transcripts in all degradation states. For equal transition rates, the mRNA lifetime follows an Erlang distribution with mean mRNA lifetime R/μR/\mu, and coefficient of variation squared 1/R1/R. The case R=1R=1 recovers the conventional exponential lifetime, whereas R>1R>1 produces a nonexponential lifetime with increasing temporal regularity. When all degradation steps have identical rate μ\mu, the exact mean dynamics of the total mRNA abundance is given as follows:

⟨x(t)⟩=k⟨B⟩μ[R−e−μt∑j=0R−1(R−j)(μt)jj!].(8)\langle x(t)\rangle= \frac{k\langle B\rangle}{\mu}\left[R-e^{-\mu t}\sum_{j=0}^{R-1}(R-j)\frac{(\mu t)^j}{j!}\right]. \tag*{(8)}

The first term represents the steady-state mean mRNA abundance, determined by the mean production rate and the average mRNA lifetime. The second term accounts for the transient buildup from the initially empty system, with its time dependence reflecting the sequential degradation steps. As this transient contribution vanishes at long times, the mean mRNA abundance approaches its steady-state value,

⟨xs⟩=k⟨B⟩Rμ.(9)\langle x_s\rangle= \frac{k\langle B\rangle R}{\mu}. \tag*{(9)}

The detailed derivation is provided in Appendix Sec. B 1. The mean mRNA dynamics for constant and geometric burst distributions are shown, respectively, in Figure 2(b) and (c) for ⟨B⟩=1\langle B\rangle=1. For both burst distributions, Eq. (8) agrees well with the stochastic simulations for R=1R=1, 55, and 1010 degradation steps. When the mean mRNA lifetime R/μR/\mu is held fixed, increasing RR makes the dynamics faster. Consequently, the mean FPT defined in Eq. (2) reduces with RR, as shown in Figure 2(f,Figure 2h). The analytical predictions for the mean FPT, obtained using Eq. (2), also agree well with the stochastic simulations for multistep mRNA degradation.

The noise in the mRNA level is quantified by the squared CV, which is the ratio between the variance and the square of the mean. For the multistep-degradation model, we derive the noise dynamics in Appendix Sec B 2, given as follows:

CVx2(t)=1⟨x(t)⟩+⟨B2⟩−⟨B⟩μ⟨x(t)⟩2∑i=0R−1∑j=0R−1(i+j)!i!j!2i+j+1×[1−e−2μt∑ℓ=0i+j(2μt)ℓℓ!].(10)CV_x^2(t) = \frac{1}{\langle x(t)\rangle} + \frac{\langle B^2\rangle-\langle B\rangle}{\mu\langle x(t)\rangle^2} \sum_{i=0}^{R-1}\sum_{j=0}^{R-1}\frac{(i+j)!}{i!j!2^{i+j+1}} \times\left[1-e^{-2\mu t}\sum_{\ell=0}^{i+j}\frac{(2\mu t)^\ell}{\ell!}\right]. \tag*{(10)}

The first term represents the Poisson contribution to mRNA fluctuations, whereas the second captures the additional fluctuations associated with bursty production. The magnitude of the latter depends on both the burst-size statistics and the degradation dynamics, which determine how long transcripts produced within the same burst survive together. Taking the long-time limit (t→∞t \to\infty), the steady-state noise becomes

CVxs2=1⟨xs⟩(1+⟨B2⟩−⟨B⟩⟨B⟩qR),qR=1−(2RR)4R,(11)CV_{x_s}^2 = \frac{1}{\langle x_s\rangle}\left(1+\frac{\langle B^2\rangle-\langle B\rangle}{\langle B\rangle}q_R\right), \qquad q_R=1-\frac{\binom{2R}{R}}{4^R}, \tag*{(11)}

which agrees with the general queueing-theoretic expression [28] when the molecular lifetime distribution is specialized to the Erlang distribution arising from RR sequential degradation steps. The factor qRq_R captures the effect of multistep degradation, which increases monotonically with RR, hence increasing the steady-state mRNA noise. For unit-sized bursts (B=1B=1), this additional contribution vanishes, and the steady-state abundance remains Poissonian, independently of the number of degradation steps RR [24]. We validate the noise dynamics in mRNA levels for multi-step mRNA degradation with stochastic simulation in Figure 2 (d) and Figure 2 (e), respectively, for constant and geometric burst size distributions. For constant bursts (⟨B2⟩=⟨B⟩2\langle B^2\rangle=\langle B\rangle^2), with unit burst size, ⟨B⟩=1\langle B\rangle=1, Eq. (11) shows that the steady-state noise is Poissonian, where CVxs2=1/⟨xs⟩CV_{x_s}^2=1/\langle x_s\rangle, which is independent of RR. However, for ⟨B⟩>1\langle B\rangle>1, CVxs2CV_{x_s}^2 increases with RR at fixed mean mRNA lifetime. However, for geometrically distributed bursts, ⟨B2⟩=⟨B⟩(1+2⟨B⟩)\langle B^2\rangle=\langle B\rangle(1+2\langle B\rangle); consequently, Eq. (11) predicts that the steady-state noise increases monotonically with RR for any ⟨B⟩>0\langle B\rangle>0 at fixed ⟨xs⟩\langle x_s\rangle. For the same mean burst size ⟨B⟩\langle B\rangle, geometric bursts produce larger fluctuations than constant ones because of their additional burst-size variability, which is also evident in Figure 2 (d) and Figure 2 (e).

In contrast to the behavior of the steady-state abundance noise, the first-passage-time noise decreases as the number of degradation steps, RR, increases, as shown by the stochastic simulation results in Figure 2(g) and Figure 2(i), respectively, for constant and geometric burst distributions. This reduction arises through a mechanism distinct from multistep burst arrival: at fixed mean mRNA lifetime, increasing the number of degradation steps accelerates the mean mRNA response and increases the logarithmic sensitivity SxS_x in Eq. (3), thereby contributing to lower FPT noise. Thus, degradation memory can improve timing precision even when it does not reduce steady-state abundance noise.

The behavior of the optimal threshold further distinguishes multi-step degradation from the multi-step burst arrival process. For multi-step degradation, X0X_0 remains close to or above the classical value of 0.5 over most of the parameter range examined and shifts toward higher values as ⟨B⟩\langle B\rangle increases (see Appendix Fig. 5 (g, j)). At the smallest burst sizes and largest values of RR, the optimal threshold remains close to 0.5. Thus, unlike multistep burst arrival, multistep degradation does not produce a systematic shift of X0X_0 toward lower thresholds. Degradation memory therefore primarily improves temporal precision while preserving a relatively high optimal threshold, whereas burst-arrival memory can modify both FPT precision and the location of the optimal threshold.

Combined Effects of Multistep Transcription and Degradation

We next consider multistep burst arrival and mRNA degradation within a single model Figure 3(a). Burst production requires completion of NN sequential transcriptional stages, each with rate kk, whereas each transcript independently traverses RR degradation stages, each with rate μ\mu, before removal. Burst sizes are deterministic, B=⟨B⟩B = \langle B\rangle, and the mRNA abundance xx counts transcripts in all degradation stages. To isolate the effects of changing the number of stages, we hold the mean interburst time, N/kN/k, and the mean mRNA lifetime, R/μR/\mu, fixed.

Combined effects of multistep transcription and mRNA degradation on FPT noise for constant burst

Figure 3. Combined effects of multistep transcription and mRNA degradation on FPT noise for constant burst. Symbols denote SSA results; curves show analytical predictions or approximations as specified below. (a) Schematic of reaction scheme. Transcription proceeds through NN sequential stages at rate kk, producing a deterministic mRNA burst upon completion and resetting the cycle. Each transcript independently traverses RR degradation stages at rate μ\mu before removal. (b) Mean mRNA dynamics for (N,R)=(2,1),(2,10),(7,1)(N,R)=(2,1),(2,10),(7,1), and (7,10)(7,10). Black curves show Eq. (12). The mean dynamics is independent of NN and approaches the common steady state more rapidly as RR increases. (c) Transient mRNA noise, CV2(t)CV^2(t), which decreases with increasing NN. Solid curves show Eq. (13). Inset: steady-state noise in the N−RN-R plane for ⟨B⟩=5\langle B\rangle=5. Increasing NN reduces noise for all ranges of RR. However, increasing RR raises the steady-state noise for N<⟨B⟩N<\langle B\rangle and lowers it for N>⟨B⟩N>\langle B\rangle. (d) Mean FPT as a function of the normalized threshold X/⟨xss⟩X/\langle x_{\mathrm{ss}}\rangle. Solid curves show the threshold-crossing estimates obtained from the exact mean dynamics using Eq. (2). The predicted mean FPT is independent of NN at fixed RR, so the corresponding analytical curves overlap. Increasing RR reduces the mean FPT independent of threshold values. (e) FPT noise, CVT2CV_T^2, versus normalized threshold. Curves show the small-noise approximation, Eq. (3), evaluated using the exact transient mRNA mean and noise. FPT noise is nonmonotonic in the threshold, with a minimum whose value and location depend on NN and RR. Increasing both NN and RR reduces FPT noise. All results use ⟨B⟩=5\langle B\rangle=5, μ=R/10\mu=R/10, and k=Nk=N, fixing the mean interburst time at N/k=1N/k=1 and the mean mRNA lifetime at R/μ=10R/\mu=10 and mean steady-state mRNA number ⟨xss⟩=(⟨B⟩kR)/(μN)=50\langle x_{\mathrm{ss}}\rangle=(\langle B\rangle kR)/(\mu N)=50.

Under these conditions, the exact mean dynamics are

⟨x(t)⟩=k⟨B⟩Nμ∑j=1R[1−e−μt∑ℓ=0j−1(μt)ℓℓ!].(12)\langle x(t)\rangle= \frac{k\langle B\rangle}{N\mu}\sum_{j=1}^{R}\left[1-e^{-\mu t}\sum_{\ell=0}^{j-1}\frac{(\mu t)^\ell}{\ell!}\right]. \tag*{(12)}

The derivation is given in Appendix C. At fixed mean interburst time, increasing NN changes the temporal regularity of burst production without changing its mean rate; hence, the mean mRNA trajectory is independent of NN. Increasing RR at fixed mean lifetime narrows the lifetime distribution and accelerates the approach to the common steady state. These properties are confirmed by the analytical curves and stochastic simulations in Figure 3(b).

Unlike the mean, the transient noise depends on both NN and RR. Writing the mean burst arrival rate λ=k/N\lambda= k/N, the exact squared coefficient of variation for constant burst can be expressed as

CVx2(t)=⟨x(t)⟩+λ⟨B⟩(⟨B⟩−1)IR(t)+2λ⟨B⟩2JN,R(t)⟨x(t)⟩2,t>0.(13)CV_x^2(t) = \frac{\langle x(t)\rangle+\lambda\langle B\rangle(\langle B\rangle-1)I_R(t)+2\lambda\langle B\rangle^2J_{N,R}(t)}{\langle x(t)\rangle^2},\qquad t>0. \tag*{(13)}

Here, IR(t)=∫0tSR(u)2 duI_R(t)=\int_0^t S_R(u)^2\,du, where SR(u)S_R(u) is the survival probability of an individual transcript, and JN,R(t)J_{N,R}(t) is the survival-weighted integral of correlations between distinct burst arrivals. Their explicit expressions and derivation are given in Appendix C. The three terms in the numerator represent the individual-molecule contribution, the contribution from molecules produced in the same burst, and the contribution from correlations between different bursts, respectively. For Poisson burst arrivals (N=1N=1), J1,R(t)=0J_{1,R}(t)=0. Multistep arrival modifies the last term and suppresses the transient fluctuations in the cases shown in Figure 3(c), while leaving the mean trajectory unchanged. The exact transient prediction agrees with the simulations.

The steady-state mean and the large-copy-number approximation to the steady-state noise are

⟨xs⟩=k⟨B⟩RNμ.(14)\langle x_s\rangle= \frac{k\langle B\rangle R}{N\mu}. \tag*{(14)}
CVxs2≃1⟨xs⟩[1+(⟨B⟩N−1)qR],qR=1−(2R)!4R(R!)2.(15)CV_{x_s}^2 \simeq\frac{1}{\langle x_s\rangle}\left[1+\left(\frac{\langle B\rangle}{N}-1\right)q_R\right],\qquad q_R=1-\frac{(2R)!}{4^R(R!)^2}. \tag*{(15)}

Since qRq_R increases monotonically with RR, the effect of multistep degradation depends on the ratio ⟨B⟩/N\langle B\rangle/N. At fixed ⟨xs⟩\langle x_s\rangle, increasing RR reduces the steady-state noise when ⟨B⟩<N\langle B\rangle<N, leaves it unchanged within this approximation when ⟨B⟩=N\langle B\rangle=N, and increases it when ⟨B⟩>N\langle B\rangle>N. The inset of

Fig. 3(c) illustrates the opposite trends. For N=1N = 1, Eq. (15) recovers Eq. (11) for constant burst, ⟨B2⟩=⟨B⟩2\langle B^2\rangle= \langle B\rangle^2.

We then examine the first-passage time to a threshold XX in total mRNA abundance. The mean-crossing estimate in Eq. (2) depends only on the mean trajectory and is therefore independent of NN at fixed RR. Increasing RR lowers this estimate, consistent with the faster accumulation of mRNA [Figure 3(d)]. The simulations follow the predicted trends, with larger deviations at high thresholds.

The small-noise approximation in Eq. (3) depends on both the transient mRNA noise and the slope of the mean trajectory at the mean-crossing time. Consequently, steady-state abundance noise alone does not determine FPT noise. As shown in Figure 3(e), CVT2CV_T^2 is nonmonotonic in the threshold, and both its minimum and the threshold at which it occurs depend on NN and RR. Increasing NN and RR reduces FPT noise. Among the combinations considered, (N,R)=(7,10)(N, R) = (7, 10) yields the lowest FPT noise. Deviations from the small-noise prediction become more pronounced at high copy-number noise. These results show that temporal precision reflects both production regularity and the transient response shaped by degradation.

Discussion and Conclusions

We have investigated how molecular memory generated by multistep burst arrival and multistep mRNA degradation affects the precision of threshold-crossing times in stochastic gene expression. Previous studies of Markovian models found that FPT noise is minimized at a threshold close to one-half of the steady-state expression level [7, 8] [4]. Our analytical calculations and stochastic simulations show that the optimal threshold is not universal for nonexponential waiting times. Both forms of memory reduce FPT noise, but through distinct mechanisms. At fixed mean burst arrival time, increasing the number of burst-arrival steps regularizes production and suppresses copy-number fluctuations without altering the mean accumulation trajectory. Within the small-noise approximation, this reduction in abundance noise translates into lower FPT noise. Multistep degradation instead alters the transient trajectory and its time sensitivity. When the mean mRNA lifetime is fixed, increasing the number of degradation steps accelerates the mean mRNA response and increases the logarithmic sensitivity, thereby contributing to lower FPT noise. For constant burst, production memory also shifts the optimal normalized threshold below one-half, whereas degradation memory generally leaves the optimum close to or above one-half. Thus, production and degradation memory can reinforce one another in reducing FPT noise while having distinct, and in some regimes opposing, effects on the location of its minimum.

The predicted dependence of timing precision on molecular memory is experimentally testable. Live-cell studies have revealed substantial variation in transcriptional bursting kinetics among mammalian genes, and recent work has emphasized the importance of resolving the relevant timescales underlying transcriptional bursting [29, 32]. For an inducible reporter whose expression activates a tunable downstream threshold, one could simultaneously estimate burst waiting-time distributions, burst-size distributions, and threshold-crossing-time statistics. The present theory predicts that promoters with more regular burst arrival should exhibit reduced FPT noise; moreover, a shift of the optimal threshold below one-half should be most apparent when burst-size fluctuations are small. Phage λ\lambda provides a complementary test bed in which the holin threshold and translational efficiency can be perturbed while lysis times are measured in single cells [8, 13]. Synthetic manipulation of RNA processing or degradation pathways could further test the prediction that degradation memory improves timing precision without reproducing the same threshold shift as production memory.

The analytical treatment relies on a small-noise approximation and on a regime in which burst arrival is fast compared with mRNA degradation. These assumptions allow the FPT noise to be related directly to transient abundance fluctuations and the slope of the mean trajectory. Deviations between the analytical approximation and stochastic simulations are expected when burst sizes become large, when burst-amplitude fluctuations are strong, or when threshold crossing is dominated by discrete overshoot events. Nevertheless, the simulations show that the qualitative conclusions persist beyond the strict approximation regime: multistep arrival regularizes timing, burst-size statistics determine whether the optimal threshold shifts, and multistep degradation provides an additional route to lower timing variability. These passive kinetic-memory effects are complementary to active timing control by feedforward or feedback circuits and to robustness against external disturbances [11, 14, 15, 17, 18]. Extensions incorporating explicit promoter switching, protein-level thresholds, extrinsic noise, cell division and dilution, or correlated production and degradation memory would further clarify how non-Markovian gene-expression dynamics regulate cellular decision times.

In summary, molecular memory is not merely a correction to Markovian models of gene expression; it provides a mechanism for controlling temporal precision. Production memory suppresses fluctuations in stochastic input and can shift the optimal threshold to earlier expression levels when bursts are sufficiently regular, whereas degradation memory improves timing through modified transcript persistence and transient accumulation. These findings establish that the precision and optimal design of intracellular timing are determined not only by mean expression levels, but also by the temporal organization and amplitude statistics of molecular events.

References

Transient Mean and Variance for Multistep Burst Arrival

We derive the transient mean and variance of the mRNA abundance for the multistep burst-arrival model considered in Sec. III. The transcriptional cycle consists of NN sequential states,

D0→kD1→k⋯→kDN−1→kD0+BM,(16)D_0 \xrightarrow{k} D_1 \xrightarrow{k} \cdots\xrightarrow{k} D_{N-1} \xrightarrow{k} D_0 + BM, \tag*{(16)}

where completion of the final transition produces a burst of mRNAs. Individual mRNAs undergo first-order degradation,

M→μ∅.(17)M \xrightarrow{\mu} \varnothing. \tag*{(17)}

We denote the burst-size distribution by

ρb=Pr⁡(B=b),b=0,1,2,…,(18)\rho_b = \Pr(B=b), \qquad b=0,1,2,\ldots, \tag*{(18)}

with moments

⟨B⟩=∑b=0∞bρb,⟨B2⟩=∑b=0∞b2ρb.(19)\langle B\rangle= \sum_{b=0}^{\infty} b\rho_b, \qquad\langle B^2\rangle= \sum_{b=0}^{\infty} b^2\rho_b. \tag*{(19)}

The chemical master equation for state D0D_0 is

∂P0(x,t)∂t=k∑b=0∞ρbPN−1(x−b,t)−kP0(x,t)+μ(x+1)P0(x+1,t)−μxP0(x,t),(20)\frac{\partial P_0(x,t)}{\partial t} = k\sum_{b=0}^{\infty}\rho_b P_{N-1}(x-b,t) - kP_0(x,t) + \mu(x+1)P_0(x+1,t) - \mu xP_0(x,t), \tag*{(20)}

and, for the remaining transcriptional states (i=1,2,…,N−1i=1,2,\ldots,N-1),

∂Pi(x,t)∂t=kPi−1(x,t)−kPi(x,t)+μ(x+1)Pi(x+1,t)−μxPi(x,t),(21)\frac{\partial P_i(x,t)}{\partial t} = kP_{i-1}(x,t) - kP_i(x,t) + \mu(x+1)P_i(x+1,t) - \mu xP_i(x,t), \tag*{(21)}

where

Pi(x,t)=Pr⁡[x(t)=x, Di(t)=1],i=0,…,N−1,(22)P_i(x,t)=\Pr[x(t)=x,\ D_i(t)=1], \qquad i=0,\ldots,N-1, \tag*{(22)}

denote the joint probability that there are xx mRNAs and that the transcriptional cycle is in state DiD_i at time tt. The probability of occupying transcriptional state DiD_i is

pi(t)=⟨Di(t)⟩=∑x=0∞Pi(x,t),(23)p_i(t)=\langle D_i(t)\rangle=\sum_{x=0}^{\infty}P_i(x,t), \tag*{(23)}
FPT noise with mean burst size and transcription steps

Figure 4. FPT noise with ⟨B⟩\langle B\rangle and NN for constant burst and multi-step transcription. FPT noise with normalized threshold (X/⟨xs⟩X/\langle x_s\rangle) for different mean burst size, ⟨B⟩\langle B\rangle when burst is constant and number of transcription steps, (a) N=1N = 1, (b) N=5N = 5, and (c) N=10N = 10. In the large noise limit, our approximated SNA calculation (solid lines Eq. (3) does not hold with SSA simulation (symbols) [39]. (d) Steady-state Fano factor of mRNA level, (e) minima value of FPT noise, and (f) minima value of normalized threshold in the ⟨B⟩−N\langle B\rangle- N plane, obtained from SSA simulation [39] over 10410^4 realizations. (g)-(i) Geometric burst distribution: the same analysis is presented in the last row for the geometric burst distribution, where X0≳0.5X_0 \gtrsim0.5, unlike the constant burst case, where X0X_0 can fall below 0.5. Parameter values: ⟨xs⟩=50\langle x_s\rangle= 50, ⟨τ⟩=2\langle\tau\rangle= 2, k=N/⟨τ⟩k = N/\langle\tau\rangle, and μ=⟨B⟩/(⟨xs⟩⟨τ⟩)\mu= \langle B\rangle/(\langle x_s\rangle\langle\tau\rangle) are tuned to keep ⟨xs⟩\langle x_s\rangle and ⟨τ⟩\langle\tau\rangle fixed for different NN and ⟨B⟩\langle B\rangle.

and the moments of the total mRNA abundance are

⟨xn(t)⟩=∑i=0N−1∑x=0∞xnPi(x,t).(24)\langle x^n(t)\rangle= \sum_{i=0}^{N-1}\sum_{x=0}^{\infty} x^n P_i(x,t). \tag*{(24)}

The state probabilities satisfy

dp0dt=kpN−1−kp0,(25)\frac{d p_0}{d t}=k p_{N-1}-k p_0, \tag*{(25)}
dpidt=kpi−1−kpi,i=1,…,N−1.(26)\frac{d p_i}{d t}=k p_{i-1}-k p_i,\qquad i=1,\ldots,N-1. \tag*{(26)}

Let Di(t)∈{0,1}D_i(t)\in\{0,1\} denote the occupancy of transcriptional state ii, such that

∑i=0N−1Di(t)=1,(27)\sum_{i=0}^{N-1}D_i(t)=1, \tag*{(27)}

and we obtain

pi(t)=1N,(28)p_i(t)=\frac{1}{N}, \tag*{(28)}

by assuming burst arrivals are faster than the mRNA lifetime.

Exact mean dynamics of mRNA

Let m(t)=⟨x(t)⟩m(t)=\langle x(t)\rangle denote the mean total mRNA abundance. A burst-production event changes xx by BB, whereas a degradation event changes xx by −1-1. The exact mean equation is therefore

dmdt=k⟨B⟩pN−1(t)−μm(t).(29)\frac{d m}{d t}=k\langle B\rangle p_{N-1}(t)-\mu m(t). \tag*{(29)}

For an arbitrary initial transcriptional phase and m(0)=0m(0)=0, Eq. (29) gives

m(t)=k⟨B⟩∫0tpN−1(t′)e−μ(t−t′) dt′.(30)m(t)=k\langle B\rangle\int_0^t p_{N-1}(t')e^{-\mu(t-t')}\,dt'. \tag*{(30)}

Thus, the transient mean generally depends on the initialization of the transcriptional cycle. For the stationary phase initialization in Eq. (28), the mean equation reduces to

dmdt=k⟨B⟩N−μm(t).(31)\frac{d m}{d t}=\frac{k\langle B\rangle}{N}-\mu m(t). \tag*{(31)}

Starting from an empty mRNA population, m(0)=0m(0)=0, its exact solution is

⟨x(t)⟩=k⟨B⟩μN(1−e−μt)(32)\langle x(t)\rangle=\frac{k\langle B\rangle}{\mu N}\left(1-e^{-\mu t}\right) \tag*{(32)}

where the steady-state mean mRNA number in the limit t→∞t\to\infty is

⟨xs⟩=k⟨B⟩μN.(33)\langle x_s\rangle=\frac{k\langle B\rangle}{\mu N}. \tag*{(33)}

Exact second-moment and variance equations

The exact differential equation for the second moment is

d⟨x2⟩dt=2k⟨B⟩⟨xDN−1⟩+k⟨B2⟩⟨DN−1⟩−2μ⟨x2⟩+μ⟨x⟩.(34)\frac{d\langle x^2\rangle}{dt}=2k\langle B\rangle\langle xD_{N-1}\rangle+k\langle B^2\rangle\langle D_{N-1}\rangle-2\mu\langle x^2\rangle+\mu\langle x\rangle. \tag*{(34)}

Defining

V(t)=Var⁡[x(t)]=⟨x2(t)⟩−⟨x(t)⟩2,(35)V(t)=\operatorname{Var}[x(t)]=\langle x^2(t)\rangle-\langle x(t)\rangle^2, \tag*{(35)}

and using Eq. (29), we obtain

dVdt=−2μV+μ⟨x⟩+k⟨B2⟩⟨DN−1⟩+2k⟨B⟩Cov⁡(x,DN−1).(36)\frac{dV}{dt}=-2\mu V+\mu\langle x\rangle+k\langle B^2\rangle\langle D_{N-1}\rangle+2k\langle B\rangle\operatorname{Cov}(x,D_{N-1}). \tag*{(36)}

Steady-state covariance and abundance noise

To obtain variance dynamics from Eq. (36) we need to know the mixed moment between the mRNA abundance and the burst-producing transcriptional state. We define

ui(t)=⟨x(t)Di(t)⟩,i=0,…,N−1.(37)u_i(t)=\langle x(t)D_i(t)\rangle,\qquad i=0,\ldots,N-1. \tag*{(37)}

These mixed moments obey the closed linear system

du0dt=−(k+μ)u0+kuN−1+k⟨B⟩⟨DN−1⟩,(38)\frac{du_0}{dt}=-(k+\mu)u_0+ku_{N-1}+k\langle B\rangle\langle D_{N-1}\rangle, \tag*{(38)}
duidt=kui−1−(k+μ)ui,i=1,…,N−1.(39)\frac{du_i}{dt}=ku_{i-1}-(k+\mu)u_i,\qquad i=1,\ldots,N-1. \tag*{(39)}

For an initially empty mRNA population,

ui(0)=0,i=0,…,N−1.(40)u_i(0)=0,\qquad i=0,\ldots,N-1. \tag*{(40)}

For the stationary phase distribution, ⟨Di⟩=1/N\langle D_i\rangle=1/N. Setting the time derivatives in Eqs. (38) and (39) to zero gives the steady-state solution

ui,s=(kk+μ)iu0,s.(41)u_{i,s}=\left(\frac{k}{k+\mu}\right)^i u_{0,s}. \tag*{(41)}

The mixed moment involving the burst-producing state is

uN−1,s=⟨B⟩N1(1+μ/k)N−1.(42)u_{N-1,s}=\frac{\langle B\rangle}{N}\frac{1}{(1+\mu/k)^N-1}. \tag*{(42)}

Since

⟨xs⟩⟨DN−1⟩=k⟨B⟩μN2,(43)\langle x_s\rangle\langle D_{N-1}\rangle=\frac{k\langle B\rangle}{\mu N^2}, \tag*{(43)}

the steady-state covariance is

Cov⁡(x,DN−1)s=⟨B⟩N[1(1+μ/k)N−1−kμN].(44)\operatorname{Cov}(x,D_{N-1})_s=\frac{\langle B\rangle}{N}\left[\frac{1}{(1+\mu/k)^N-1}-\frac{k}{\mu N}\right]. \tag*{(44)}

For N=1N=1, this covariance vanishes. For N>1N>1, it is negative because

(1+μ/k)N−1>Nμ/k.(45)(1+\mu/k)^N-1>N\mu/k. \tag*{(45)}

This negative correlation between the mRNA abundance and the burst-producing state contributes to the suppression of mRNA fluctuations by multistep burst arrival. Using Eq. (44) in the steady-state limit of Eq. (36) gives

CVxs2=12⟨xs⟩[⟨B2⟩⟨B⟩+2⟨B⟩(1(1+μ/k)N−1−kμN)].(46)CV_{x_s}^{2}=\frac{1}{2\langle x_s\rangle}\left[\frac{\langle B^2\rangle}{\langle B\rangle}+2\langle B\rangle\left(\frac{1}{(1+\mu/k)^N-1}-\frac{k}{\mu N}\right)\right]. \tag*{(46)}

Fast-arrival approximation for the transient variance

A closed approximation for the variance dynamics can be obtained when all nonzero relaxation modes of the burst-arrival process are fast compared with mRNA degradation. After the short arrival-process transient, we approximate the combined production and covariance contribution in Eq. (36) by its steady-state value: The approximate transient variance equation is therefore

dVdt≈−2μV+μ⟨x(t)⟩+2μVs−μ⟨xs⟩,(47)\frac{dV}{dt}\approx-2\mu V+\mu\langle x(t)\rangle+2\mu V_s-\mu\langle x_s\rangle, \tag*{(47)}

where

Vs=Var⁡[xs]=CVxs2⟨xs⟩2.(48)V_s=\operatorname{Var}[x_s]=CV_{x_s}^{2}\langle x_s\rangle^2. \tag*{(48)}

Using the mean dynamics in Eq. (32) and the initial condition V(0)=0V(0)=0, the integrating-factor solution of Eq. (47) is

V(t)=Vs(1−e−2μt)−⟨xs⟩e−μt(1−e−μt).(49)V(t)=V_s\left(1-e^{-2\mu t}\right)-\langle x_s\rangle e^{-\mu t}\left(1-e^{-\mu t}\right). \tag*{(49)}

Finally, dividing Eq. (49) by

⟨x(t)⟩2=⟨xs⟩2(1−e−μt)2(50)\langle x(t)\rangle^2=\langle x_s\rangle^2\left(1-e^{-\mu t}\right)^2 \tag*{(50)}

gives

CVx2(t)≈CVxs2(1+e−μt1−e−μt)−e−μt⟨x(t)⟩.(51)CV_x^2(t)\approx CV_{x_s}^2\left(\frac{1+e^{-\mu t}}{1-e^{-\mu t}}\right)-\frac{e^{-\mu t}}{\langle x(t)\rangle}. \tag*{(51)}

Multistep mRNA Degradation

We derive the steady-state mean mRNA abundance for the multistep-degradation model considered in Sec. IV. Bursts arrive according to a Poisson process with rate kk. Each burst produces a random number BB of mRNAs, with mean ⟨B⟩\langle B\rangle, and all newly produced mRNAs enter the first degradation state, M0M_0. Each transcript then passes through RR sequential degradation states,

M0→μM1→μ⋯→μMR−1→μ∅.(52)M_0 \xrightarrow{\mu} M_1 \xrightarrow{\mu} \cdots\xrightarrow{\mu} M_{R-1} \xrightarrow{\mu} \varnothing. \tag*{(52)}

The total mRNA abundance is defined as

x(t)=∑r=0R−1Mr(t).(53)x(t) = \sum_{r=0}^{R-1} M_r(t). \tag*{(53)}

Exact Mean Dynamics

Let us denote the mean mRNA as

mr(t)=⟨Mr(t)⟩.(54)m_r(t) = \langle M_r(t)\rangle. \tag*{(54)}

Because the reaction network is linear, the equations governing the mean copy numbers close exactly:

dm0dt=k⟨B⟩−μm0,(55)\frac{d m_0}{d t} = k\langle B\rangle- \mu m_0, \tag*{(55)}
dmrdt=μmr−1−μmr,r=1,…,R−1.(56)\frac{d m_r}{d t} = \mu m_{r-1} - \mu m_r,\qquad r=1,\ldots,R-1. \tag*{(56)}

We consider the following initial condition

mr(0)=0,r=0,…,R−1.(57)m_r(0) = 0,\qquad r=0,\ldots,R-1. \tag*{(57)}

Denoting the Laplace transform of mr(t)m_r(t) by m~r(s)\widetilde{m}_r(s), Eq. (55) gives

m~0(s)=k⟨B⟩s(s+μ).(58)\widetilde{m}_0(s) = \frac{k\langle B\rangle}{s(s+\mu)}. \tag*{(58)}

For the subsequent degradation states, Eq. (56) gives the recursion

m~r(s)=μs+μm~r−1(s)=k⟨B⟩μrs(s+μ)r+1.(59)\widetilde{m}_r(s) = \frac{\mu}{s+\mu}\widetilde{m}_{r-1}(s) = \frac{k\langle B\rangle\mu^r}{s(s+\mu)^{r+1}}. \tag*{(59)}

Inverting Eq. (59), we obtain

mr(t)=k⟨B⟩μ[1−e−μt∑j=0r(μt)jj!],r=0,…,R−1.(60)m_r(t) = \frac{k\langle B\rangle}{\mu}\left[1-e^{-\mu t}\sum_{j=0}^{r}\frac{(\mu t)^j}{j!}\right],\qquad r=0,\ldots,R-1. \tag*{(60)}

Summing Eq. (60) over all degradation states gives the exact mean dynamics of the total mRNA abundance:

⟨x(t)⟩=k⟨B⟩μ[R−e−μt∑j=0R−1(R−j)(μt)jj!].(61)\langle x(t)\rangle= \frac{k\langle B\rangle}{\mu}\left[R-e^{-\mu t}\sum_{j=0}^{R-1}(R-j)\frac{(\mu t)^j}{j!}\right]. \tag*{(61)}

Taking the long-time limit yields

⟨xs⟩=lim⁡t→∞⟨x(t)⟩=k⟨B⟩Rμ.(62)\langle x_s\rangle= \lim_{t\to\infty}\langle x(t)\rangle= \frac{k\langle B\rangle R}{\mu}. \tag*{(62)}

The normalized transient mean can therefore be written as

⟨x(t)⟩⟨xs⟩=1−e−μtR∑j=0R−1(R−j)(μt)jj!.(63)\frac{\langle x(t)\rangle}{\langle x_s\rangle}=1-\frac{e^{-\mu t}}{R}\sum_{j=0}^{R-1}(R-j)\frac{(\mu t)^j}{j!}. \tag*{(63)}

In our simulations, the steady-state mean is held fixed by choosing

k=μ⟨xs⟩R⟨B⟩.(64)k=\frac{\mu\langle x_s\rangle}{R\langle B\rangle}. \tag*{(64)}

Equations (B10) and (B12) are exact and do not require a small-noise or timescale-separation approximation. Moreover, the burst-size distribution affects the mean dynamics only through ⟨B⟩\langle B\rangle.

Exact Variance Dynamics

We next derive the exact variance dynamics of the total mRNA abundance. The lifetime of an individual mRNA is Erlang distributed with shape parameter RR and rate μ\mu. Its survival probability at age uu is

SR(u)=e−μu∑j=0R−1(μu)jj!.(65)S_R(u)=e^{-\mu u}\sum_{j=0}^{R-1}\frac{(\mu u)^j}{j!}. \tag*{(65)}

Consider a burst that occurred at time uu before observation. Conditional on the burst size BB, the number of mRNAs from this burst that survive until the observation time, denoted by Y(u)Y(u), follows a binomial distribution:

Y(u)∣B∼Binomial⁡(B,SR(u)).(66)Y(u)\mid B\sim\operatorname{Binomial}(B,S_R(u)). \tag*{(66)}

Its first two conditional moments are

⟨Y(u)∣B⟩=BSR(u),(67)\langle Y(u)\mid B\rangle=BS_R(u), \tag*{(67)}
⟨Y2(u)∣B⟩=BSR(u)[1−SR(u)]+B2SR2(u).(68)\langle Y^2(u)\mid B\rangle=BS_R(u)\left[1-S_R(u)\right]+B^2S_R^2(u). \tag*{(68)}

Averaging over the burst-size distribution gives

⟨Y(u)⟩=⟨B⟩SR(u),(69)\langle Y(u)\rangle=\langle B\rangle S_R(u), \tag*{(69)}
FPT noise for multi-step mRNA degradation and bursty single-step production

Figure 5. FPT noise for multi-step mRNA degradation and bursty single-step production. (a) The schematic of the gene network, where transcription is a single-step process with bursty production of size BB and frequency kk, and degradation is an RR-step process with equal rate μ\mu. FPT noise with normalized threshold (X/⟨xs⟩)(X/\langle x_s\rangle) for different mean burst size, ⟨B⟩\langle B\rangle when burst is constant and number of degradation steps, (a) R=1R=1, (b) R=5R=5, and (c) R=10R=10. (e) Steady-state Fano factor of mRNA level, (f) minima value of FPT noise, and (g) minima value of normalized threshold in the ⟨B⟩−R\langle B\rangle-R plane, obtained from SSA simulation [39] over 10310^3 realizations. (e-g) We consider constant burst, i.e. σB2=0\sigma_B^2=0, (h-j) and geometric burst, i.e, σB2=⟨B⟩(1+⟨B⟩)\sigma_B^2=\langle B\rangle(1+\langle B\rangle). Parameter values: μ=R/10\mu=R/10 to keep the mRNA lifetime fixed at 10 and k=⟨xs⟩μ/(R⟨B⟩)k=\langle x_s\rangle\mu/(R\langle B\rangle) to keep mean steady-state mRNA number fixed at 50 for varying RR and ⟨B⟩\langle B\rangle.

⟨Y2(u)⟩=⟨B⟩SR(u)+(⟨B2⟩−⟨B⟩)SR2(u).(70)\left\langle Y^2(u)\right\rangle=\langle B\rangle S_R(u)+\left(\left\langle B^2\right\rangle-\langle B\rangle\right)S_R^2(u). \tag*{(70)}

Since the bursts arrive according to a Poisson process with rate kk, the variance of the total mRNA abundance is

Var⁡[x(t)]=k∫0t⟨Y2(u)⟩ du.(71)\operatorname{Var}[x(t)]=k\int_0^t\left\langle Y^2(u)\right\rangle\,du. \tag*{(71)}

Using Eq. (70), this becomes

Var⁡[x(t)]=⟨B⟩k∫0tSR(u) du+k(⟨B2⟩−⟨B⟩)∫0tSR2(u) du.(72)\operatorname{Var}[x(t)]=\langle B\rangle k\int_0^t S_R(u)\,du+k\left(\left\langle B^2\right\rangle-\langle B\rangle\right)\int_0^t S_R^2(u)\,du. \tag*{(72)}

We define

I2(t)=∫0tSR2(u) du.(73)I_2(t) = \int_0^t S_R^2(u)\,\mathrm{d}u. \tag*{(73)}

Substituting Eq. (65) into Eq. (73) gives

I2(t)=1μ∑i=0R−1∑j=0R−1(i+j)!i!j!2i+j+1×[1−e−2μt∑ℓ=0i+j(2μt)ℓℓ!].(74)I_2(t) = \frac{1}{\mu}\sum_{i=0}^{R-1}\sum_{j=0}^{R-1}\frac{(i+j)!}{i!j!2^{i+j+1}}\times\left[1-e^{-2\mu t}\sum_{\ell=0}^{i+j}\frac{(2\mu t)^\ell}{\ell!}\right]. \tag*{(74)}

Equations (B21) and (B23) provide the exact variance dynamics for an arbitrary burst-size distribution with finite first and second moments.

For a constant burst of size B=bB=b,

⟨B⟩=b,⟨B2⟩=b2,(75)\langle B\rangle=b,\qquad\langle B^2\rangle=b^2, \tag*{(75)}

and hence

Var⁡[x(t)]=⟨x(t)⟩+kb(b−1)I2(t).(76)\operatorname{Var}[x(t)]=\langle x(t)\rangle+kb(b-1)I_2(t). \tag*{(76)}

In particular, for unit bursts, b=1b=1,

Var⁡[x(t)]=⟨x(t)⟩,FF(t)=1,(77)\operatorname{Var}[x(t)]=\langle x(t)\rangle,\qquad\mathrm{FF}(t)=1, \tag*{(77)}

at all times, independently of the number of degradation steps.

Taking the long-time limit of Eq. (74) gives

I2(∞)=1μ∑i=0R−1∑j=0R−1(i+j)!i!j!2i+j+1=Rμ[1−(2RR)4R].(78)\begin{aligned} I_2(\infty) &= \frac{1}{\mu}\sum_{i=0}^{R-1}\sum_{j=0}^{R-1}\frac{(i+j)!}{i!j!2^{i+j+1}}\\ &= \frac{R}{\mu}\left[1-\frac{\binom{2R}{R}}{4^R}\right]. \tag*{(78)} \end{aligned}

Using

⟨xs⟩=k⟨B⟩Rμ,(79)\langle x_s\rangle=\frac{k\langle B\rangle R}{\mu}, \tag*{(79)}

the steady-state variance becomes

Var⁡[xs]=⟨xs⟩{1+⟨B2⟩−⟨B⟩⟨B⟩[1−(2RR)4R]}.(80)\operatorname{Var}[x_s]=\langle x_s\rangle\left\{1+\frac{\langle B^2\rangle-\langle B\rangle}{\langle B\rangle}\left[1-\frac{\binom{2R}{R}}{4^R}\right]\right\}. \tag*{(80)}

Therefore, the steady-state Fano factor is

FFs=1+⟨B2⟩−⟨B⟩⟨B⟩[1−(2RR)4R].(81)\mathrm{FF}_s=1+\frac{\langle B^2\rangle-\langle B\rangle}{\langle B\rangle}\left[1-\frac{\binom{2R}{R}}{4^R}\right]. \tag*{(81)}

For constant burst of size bb, this reduces to

FFs=1+(b−1)[1−(2RR)4R].(82)\mathrm{FF}_{s}=1+(b-1)\left[1-\frac{\binom{2R}{R}}{4^{R}}\right]. \tag*{(82)}

Finally, the exact transient Fano factor is

FF(t)=Var⁡[x(t)]⟨x(t)⟩=1+k(⟨B2⟩−⟨B⟩)I2(t)⟨x(t)⟩,t>0.(83)\mathrm{FF}(t)=\frac{\operatorname{Var}[x(t)]}{\langle x(t)\rangle}=1+\frac{k\left(\langle B^{2}\rangle-\langle B\rangle\right)I_{2}(t)}{\langle x(t)\rangle},\qquad t>0. \tag*{(83)}

These exact abundance moments are used to validate the stochastic simulations.

mRNA Mean and Noise Dynamics with Multistep Burst Arrival and Degradation

We derive the transient and steady-state moments of the total mRNA abundance when both burst arrival and degradation proceed through multiple stages. Throughout, bursts have a fixed positive integer size ⟨B⟩\langle B\rangle, so that ⟨B2⟩=⟨B⟩2\langle B^{2}\rangle=\langle B\rangle^{2}. Individual transcripts degrade independently, and x(t)x(t) denotes the total number of transcripts across all degradation stages.

Model and mean dynamics

Each burst is generated upon completion of NN sequential transcriptional stages, each with rate kk. The interburst interval τ\tau therefore has the Erlang density

fN(τ)=kNτN−1e−kτ(N−1)!,τ≥0,⟨τ⟩=Nk.(84)f_{N}(\tau)=\frac{k^{N}\tau^{N-1}e^{-k\tau}}{(N-1)!},\qquad\tau\geq0,\qquad\langle\tau\rangle=\frac{N}{k}. \tag*{(84)}

Thus, the stationary mean burst-arrival rate is λ=1/⟨τ⟩=k/N\lambda=1/\langle\tau\rangle=k/N. Here, τ\tau denotes an interburst interval, whereas ⟨τ⟩\langle\tau\rangle denotes its mean.

Each transcript passes through RR degradation stages, each with rate μ\mu, before removal. Its survival probability at age tt is

SR(t)=e−μt∑j=0R−1(μt)jj!,∫0∞SR(t) dt=Rμ.(85)S_{R}(t)=e^{-\mu t}\sum_{j=0}^{R-1}\frac{(\mu t)^{j}}{j!},\qquad\int_{0}^{\infty}S_{R}(t)\,\mathrm{d}t=\frac{R}{\mu}. \tag*{(85)}

For the transient calculation, we take x(0)=0x(0)=0 and sample the initial transcriptional phase uniformly over the NN stages. The arrival process is therefore stationary from t=0t=0, although the mRNA population is initially empty. In particular, the first waiting time is the stationary residual waiting time, rather than a complete Erlang interburst interval when N>1N>1.

A burst of age uu contributes ⟨B⟩SR(u)\langle B\rangle S_{R}(u) surviving transcripts on average. Integrating over burst ages gives mean mRNA number

⟨x(t)⟩=λ⟨B⟩∫0tSR(u) du=k⟨B⟩Nμ∑j=1R[1−e−μt∑ℓ=0j−1(μt)ℓℓ!].(86)\begin{aligned} \langle x(t)\rangle=\lambda\langle B\rangle\int_{0}^{t}S_{R}(u)\,\mathrm{d}u \\ &= \frac{k\langle B\rangle}{N\mu}\sum_{j=1}^{R}\left[1-e^{-\mu t}\sum_{\ell=0}^{j-1}\frac{(\mu t)^{\ell}}{\ell!}\right]. \tag*{(86)} \end{aligned}

Taking t→∞t \to\infty yields the steady-state mean

⟨xs⟩=λ⟨B⟩Rμ=k⟨B⟩RNμ.(87)\langle x_{\mathrm{s}}\rangle=\lambda\langle B\rangle\frac{R}{\mu}=\frac{k\langle B\rangle R}{N\mu}. \tag*{(87)}

Exact transient noise

Correlations between distinct burst arrivals are described by the renewal density

hN(u)=∑n=1∞fN∗n(u)=∑n=1∞knNunN−1e−ku(nN−1)!,u>0,(88)h_N(u)=\sum_{n=1}^{\infty}f_N^{*n}(u)=\sum_{n=1}^{\infty}\frac{k^{nN}u^{nN-1}e^{-ku}}{(nN-1)!},\qquad u>0, \tag*{(88)}

where fN∗nf_N^{*n} denotes the nn-fold convolution of fNf_N. Conditional on a burst at time zero, hN(u) duh_N(u)\,du is the expected number of bursts in an interval of width dudu at lag uu. Its Laplace transform is

h~N(s)≡∫0∞e−suhN(u) du=(k/(k+s))N1−(k/(k+s))N,s>0.(89)\widetilde{h}_N(s)\equiv\int_0^{\infty}e^{-su}h_N(u)\,du=\frac{(k/(k+s))^N}{1-(k/(k+s))^N},\qquad s>0. \tag*{(89)}

For a stationary renewal process, the density of ordered pairs of distinct bursts separated by a positive lag uu is λhN(u)\lambda h_N(u). The excess over the uncorrelated value is therefore λ[hN(u)−λ]\lambda[h_N(u)-\lambda].

The number of surviving transcripts from a burst of age uu is binomial with parameters ⟨B⟩\langle B\rangle and SR(u)S_R(u). Its second factorial moment is ⟨B⟩(⟨B⟩−1)SR(u)2\langle B\rangle(\langle B\rangle-1)S_R(u)^2. Separating pairs of transcripts produced in the same burst from those produced in distinct bursts gives

⟨x(t)[x(t)−1]⟩=λ⟨B⟩(⟨B⟩−1)∫0tSR(u)2 du+2λ⟨B⟩2∫0tSR(u)∫0uSR(v)hN(u−v) dv du.(90)\begin{aligned} \langle x(t)[x(t)-1]\rangle &=\lambda\langle B\rangle(\langle B\rangle-1)\int_0^t S_R(u)^2\,du\\ &\quad+2\lambda\langle B\rangle^2\int_0^t S_R(u)\int_0^u S_R(v)h_N(u-v)\,dv\,du. \tag*{(90)} \end{aligned}

Using Var⁡[x(t)]=⟨x(t)2⟩−⟨x(t)⟩2\operatorname{Var}[x(t)]=\langle x(t)^2\rangle-\langle x(t)\rangle^2, and defining

IR(t)=∫0tSR(u)2 du,(91)I_R(t)=\int_0^t S_R(u)^2\,du, \tag*{(91)}
JN,R(t)=∫0tSR(u)∫0uSR(v)[hN(u−v)−λ] dv du,(92)J_{N,R}(t)=\int_0^t S_R(u)\int_0^u S_R(v)[h_N(u-v)-\lambda]\,dv\,du, \tag*{(92)}

we obtain the exact transient variance

Var⁡[x(t)]=⟨x(t)⟩+λ⟨B⟩(⟨B⟩−1)IR(t)+2λ⟨B⟩2JN,R(t).(93)\operatorname{Var}[x(t)]=\langle x(t)\rangle+\lambda\langle B\rangle(\langle B\rangle-1)I_R(t)+2\lambda\langle B\rangle^2J_{N,R}(t). \tag*{(93)}

The corresponding squared coefficient of variation is

CVx2(t)=⟨x(t)⟩+λ⟨B⟩(⟨B⟩−1)IR(t)+2λ⟨B⟩2JN,R(t)⟨x(t)⟩2,t>0.(94)CV_x^2(t)=\frac{\langle x(t)\rangle+\lambda\langle B\rangle(\langle B\rangle-1)I_R(t)+2\lambda\langle B\rangle^2J_{N,R}(t)}{\langle x(t)\rangle^2},\qquad t>0. \tag*{(94)}

The restriction t>0t > 0 is required because ⟨x(0)⟩=0\langle x(0)\rangle= 0. The term involving IR(t)I_R(t) accounts for transcripts produced in the same burst, whereas the term involving JN,R(t)J_{N,R}(t) accounts for correlations between distinct bursts.

For Poisson burst arrivals (N=1N = 1), h1(u)=k=λh_1(u) = k = \lambda and hence J1,R(t)=0J_{1,R}(t) = 0. Equation (C11) then reduces to

CVx2(t)=1⟨x(t)⟩+λ⟨B⟩(⟨B⟩−1)⟨x(t)⟩2∫0tSR(u)2 du.(95)CV_x^2(t) = \frac{1}{\langle x(t)\rangle} + \frac{\lambda\langle B\rangle(\langle B\rangle- 1)}{\langle x(t)\rangle^2}\int_0^{t} S_R(u)^2\,du. \tag*{(95)}

Steady-state limit

To evaluate the long-time limit, define the survival overlap

CR(u)=∫0∞SR(v)SR(v+u) dv.(96)C_R(u) = \int_0^\infty S_R(v)S_R(v+u)\,dv. \tag*{(96)}

Changing variables in Eq. (92) gives

lim⁡t→∞JN,R(t)=∫0∞[hN(u)−λ]CR(u) du.(97)\lim_{t\to\infty} J_{N,R}(t) = \int_0^\infty[h_N(u)-\lambda]C_R(u)\,du. \tag*{(97)}

Thus, the exact steady-state variance is

Var⁡[xs]=⟨xs⟩+λ⟨B⟩(⟨B⟩−1)∫0∞SR(u)2 du+2λ⟨B⟩2∫0∞[hN(u)−λ]CR(u) du.(98)\operatorname{Var}[x_s] = \langle x_s\rangle+ \lambda\langle B\rangle(\langle B\rangle- 1)\int_0^\infty S_R(u)^2\,du + 2\lambda\langle B\rangle^2\int_0^\infty[h_N(u)-\lambda]C_R(u)\,du. \tag*{(98)}

For Erlang lifetimes,

qR≡∫0∞SR(u)2 du∫0∞SR(u) du=1−(2RR)4R.(99)q_R \equiv\frac{\int_0^\infty S_R(u)^2\,du}{\int_0^\infty S_R(u)\,du} = 1 - \frac{\binom{2R}{R}}{4^R}. \tag*{(99)}

Dividing Eq. (98) by Eq. (87) yields the exact Fano factor

FFN,R≡Var⁡[xs]⟨xs⟩=1+(⟨B⟩−1)qR+2⟨B⟩μR∫0∞[hN(u)−kN]CR(u) du.(100)\mathrm{FF}_{N,R} \equiv\frac{\operatorname{Var}[x_s]}{\langle x_s\rangle} = 1 + (\langle B\rangle- 1)q_R + \frac{2\langle B\rangle\mu}{R}\int_0^\infty\left[h_N(u)-\frac{k}{N}\right]C_R(u)\,du. \tag*{(100)}

The steady-state squared coefficient of variation is therefore

CVxs2=FFN,R⟨xs⟩.(101)CV_{x_s}^2 = \frac{\mathrm{FF}_{N,R}}{\langle x_s\rangle}. \tag*{(101)}

Finite-sum representation and limiting cases

The correlation integral in Eq. (100) admits a finite-sum representation. Define

HN(s)≡h~N(s)−λs=1(1+s/k)N−1−kNs,s>0.(102)H_N(s) \equiv\widetilde{h}_N(s) - \frac{\lambda}{s} = \frac{1}{(1+s/k)^N-1} - \frac{k}{Ns}, \qquad s > 0. \tag*{(102)}

Expanding the survival functions in Eq. (96) gives

CR(u)=e−μuμ∑p=0R−1cR,p(μu)p(103)C_R(u) = \frac{e^{-\mu u}}{\mu} \sum_{p=0}^{R-1} c_{R,p}(\mu u)^p \tag*{(103)}

with coefficients

cR,p=1p!∑i=0R−1∑ℓ=0R−1−p(i+ℓ)!i!ℓ!2i+ℓ+1.(104)c_{R,p} = \frac{1}{p!} \sum_{i=0}^{R-1} \sum_{\ell=0}^{R-1-p} \frac{(i+\ell)!}{i!\ell!2^{i+\ell+1}}. \tag*{(104)}

Since

∫0∞[hN(u)−λ]e−μuup du=(−1)pdpHN(s)dsp∣s=μ,(105)\int_0^\infty[h_N(u)-\lambda]e^{-\mu u}u^p\,du = (-1)^p \left.\frac{d^p H_N(s)}{ds^p}\right|_{s=\mu}, \tag*{(105)}

the exact Fano factor becomes

FFN,R=1+(⟨B⟩−1)qR+2⟨B⟩R∑p=0R−1cR,p(−μ)pdpHN(s)dsp∣s=μ.(106)\mathrm{FF}_{N,R} = 1 + (\langle B\rangle-1)q_R + \frac{2\langle B\rangle}{R}\sum_{p=0}^{R-1}c_{R,p}(-\mu)^p\left.\frac{d^p H_N(s)}{ds^p}\right|_{s=\mu}. \tag*{(106)}

For single-step burst arrival (N=1N=1), H1(s)=0H_1(s)=0, and Eq. (106) reduces to

FF1,R=1+(⟨B⟩−1)qR.(107)\mathrm{FF}_{1,R}=1+(\langle B\rangle-1)q_R. \tag*{(107)}

Thus, FF1,R=1\mathrm{FF}_{1,R}=1 for ⟨B⟩=1\langle B\rangle=1, whereas it increases with RR for ⟨B⟩>1\langle B\rangle>1. For single-step degradation (R=1R=1), q1=c1,0=1/2q_1=c_{1,0}=1/2, giving

FFN,1=1+⟨B⟩2+⟨B⟩[(k/(k+μ))N1−(k/(k+μ))N−kNμ].(108)\mathrm{FF}_{N,1}=\frac{1+\langle B\rangle}{2}+\langle B\rangle\left[\frac{(k/(k+\mu))^N}{1-(k/(k+\mu))^N}-\frac{k}{N\mu}\right]. \tag*{(108)}

Large-copy-number approximation

At fixed NN, RR, and ⟨B⟩\langle B\rangle, the large-copy-number limit corresponds to μ/k→0\mu/k\to0, because

μk=⟨B⟩RN⟨xs⟩.(109)\frac{\mu}{k}=\frac{\langle B\rangle R}{N\langle x_s\rangle}. \tag*{(109)}

The approximation below concerns this limit; increasing the burst size alone does not establish μ/k≪1\mu/k\ll1. Expanding Eq. (102) about s/k=0s/k=0 gives

HN(s)=−N−12N+N2−112Nsk+O(s2k2).(110)H_N(s)=-\frac{N-1}{2N}+\frac{N^2-1}{12N}\frac{s}{k}+O\left(\frac{s^2}{k^2}\right). \tag*{(110)}

At leading order, only the p=0p=0 term in Eq. (106) contributes. Moreover, Eqs. (103) and (99) imply cR,0=μCR(0)=RqRc_{R,0}=\mu C_R(0)=Rq_R. Substitution therefore yields

FFN,R=1+(⟨B⟩N−1)qR+O(μk),(111)\mathrm{FF}_{N,R}=1+\left(\frac{\langle B\rangle}{N}-1\right)q_R+O\left(\frac{\mu}{k}\right), \tag*{(111)}

or, equivalently,

CVxs2≃1⟨xs⟩[1+(⟨B⟩N−1)qR].(112)CV_{x_s}^2 \simeq\frac{1}{\langle x_s\rangle}\left[1+\left(\frac{\langle B\rangle}{N}-1\right)q_R\right]. \tag*{(112)}

The remainder in Eq. (111) is stated for fixed NN, RR, and ⟨B⟩\langle B\rangle; it is not a uniform error bound when these parameters also increase.

Since qR+1−qR=(2RR)4R(2R+2)>0q_{R+1}-q_R=\frac{\binom{2R}{R}}{4^R(2R+2)}>0, the leading-order change in the Fano factor at fixed NN and ⟨B⟩\langle B\rangle is

FFN,R+1−FFN,R≃(⟨B⟩N−1)(qR+1−qR),(113)\mathrm{FF}_{N,R+1}-\mathrm{FF}_{N,R}\simeq\left(\frac{\langle B\rangle}{N}-1\right)(q_{R+1}-q_R), \tag*{(113)}

provided both parameter sets lie within the approximation regime. Accordingly, the Fano factor decreases with RR when ⟨B⟩<N\langle B\rangle<N, is independent of RR to leading order when ⟨B⟩=N\langle B\rangle=N, and increases with RR when ⟨B⟩>N\langle B\rangle>N. At fixed ⟨xs⟩\langle x_s\rangle, the same trends apply to CVxs2CV_{x_s}^2.

For ⟨B⟩=1\langle B\rangle=1 and N>1N>1, the approximation predicts that multistep degradation enhances the noise suppression associated with regular burst arrival. In contrast, for N=1N=1 and ⟨B⟩>1\langle B\rangle>1, the interburst correlation term vanishes, and the contribution from transcripts produced in the same burst causes the Fano factor to increase with RR, as follows exactly from Eq. (107). These opposite trends reflect the competition between arrival regularity and the shared production times of transcripts within each burst.

References

  1. [5]L. Ham, M. A. Coomer, K. Öcal, R. Grima, and M. P. H. Stumpf, Nature Communications 15, 5286 (2024).
  2. [6]J. Paulsson, Physics of Life Reviews 2, 157 (2005).
  3. [7]A. Dal Co, M. Cosentino Lagomarsino, M. Caselle, and M. Osella, Nucleic Acids Research 45, 1069 (2017).
  4. [8]S. Kannoly, T. Gao, S. Dey, I.-N. Wang, A. Singh, and J. J. Dennehy, iScience 23, 101186 (2020).
  5. [9]K. R. Ghusinga, C. A. Vargas-Garcia, and A. Singh, Scientific Reports 6, 30229 (2016).
  6. [10]K. R. Ghusinga and A. Singh, in 2019 Sixth Indian Control Conference (ICC) (IEEE, 2019) pp. 280–285.
  7. [11]K. R. Ghusinga and A. Singh, in 2020 59th IEEE Conference on Decision and Control (CDC) (IEEE, 2020) pp. 5867–5872.
  8. [12]K. Rijal, A. Prasad, A. Singh, and D. Das, Physical Review Letters 128, 048101 (2022).
  9. [13]S. Kannoly, K. Singh, N. Juman, Z. Marufee Islam, I. Caballero Quiroga, A. Singh, and J. J. Dennehy, bioRxiv , 2025.05.14.654169 (2025), preprint, version 1, posted May 17, 2025.
  10. [14]S. Dey, S. Kannoly, P. Bokes, J. J. Dennehy, and A. Singh, in 2021 European Control Conference (ECC) (IEEE, 2021) pp. 2127–2132.
  11. [15]Z. Vahdat, K. R. Ghusinga, and A. Singh, in 2021 29th Mediterranean Conference on Control and Automation (MED) (IEEE, 2021) pp. 450–455.
  12. [16]K. Biswas and A. Ghosh, Physical Review E 101, 022418 (2020).
  13. [17]S. Rezaee, C. Nieto, and A. Singh, in 2023 27th International Conference on System Theory, Control and Computing (ICSTCC) (IEEE, 2023) pp. 505–510.
  14. [18]S. Rezaee, C. Nieto, and A. Singh, in 2024 IEEE 63rd Conference on Decision and Control (CDC) (IEEE, 2024) pp. 7772–7777.
  15. [19]K. Biswas, S. Dey, and A. Singh, Scientific Reports 14, 27199 (2024).
  16. [20]J. M. Pedraza and J. Paulsson, Science 319, 339 (2008).
  17. [21]C. Deneke, R. Lipowsky, and A. Valleriani, PLOS ONE 8, e55442 (2013).
  18. [22]D. E. Weidemann, J. Holehouse, A. Singh, R. Grima, and S. Hauf, Science Advances 9, eadh5138 (2023).
  19. [23]J. Zhang and T. Zhou, Proceedings of the National Academy of Sciences 116, 23542 (2019).
  20. [24]J. Szavits-Nossan and R. Grima, Biophysical Journal 123, 1034 (2024).
  21. [25]D. L. Bentley, Nature Reviews Genetics 15, 163 (2014).
  22. [26]N. L. Garneau, J. Wilusz, and C. J. Wilusz, Nature Reviews Molecular Cell Biology 8, 113 (2007).
  23. [27]M. Ma, J. Szavits-Nossan, A. Singh, and R. Grima, Mathematical Biosciences 373, 109204 (2024).
  24. [28]T. Jia and R. V. Kulkarni, Physical review letters 106, 058102 (2011).
  25. [29]D. M. Suter, N. Molina, D. Gaffield, K. Schneider, U. Schibler, and F. Naef, Science 332, 472 (2011).
  26. [30]K. B. Halpern, S. Tanami, S. Landen, M. Chapal, L. Szlak, A. Hutzler, A. Nizhberg, and S. Itzkovitz, Molecular Cell 58, 147 (2015).
  27. [31]V. Agarwal and D. R. Kelley, Genome Biology 23, 245 (2022).
  28. [32]J. V. W. Meeussen and T. L. Lenstra, Trends in Genetics 40, 160 (2024).
  29. [33]A. Dal Co, M. Cosentino Lagomarsino, M. Caselle, and M. Osella, Nucleic Acids Research 45, 1069 (2017).
  30. [34]H. Chen, L. Zhou, J. Teng, A. Chen, and T. Zhou, Physical Review E 114, 034122 (2026).
  31. [35]L. Liu, B. Kashyap, and J. Templeton, Journal of Applied Probability 27, 671 (1990).
  32. [36]N. Kumar, A. Singh, and R. V. Kulkarni, PLoS computational biology 11, e1004292 (2015).
  33. [37]Z. Vahdat and A. Singh, in 2024 IEEE 63rd Conference on Decision and Control (CDC) (IEEE, 2024) pp. 7792–7797.
  34. [38]K. R. Ghusinga, J. J. Dennehy, and A. Singh, Proceedings of the National Academy of Sciences 114, 693 (2017).
  35. [39]D. T. Gillespie, The Journal of Physical Chemistry 81, 2340 (1977).
  36. [40]S. J. Park, S. Song, G.-S. Yang, P. M. Kim, S. Yoon, J.-H. Kim, and J. Sung, Nature Communications 9, 297 (2018).

Paper details

vlnc:W_prj9rsde
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Agricultural and Biological Sciences

Not recorded for this paper

View statement
Looking for related papers…
Contents