Intracellular Bacterial Replication and Host-Cell Lysis

Why Separate Intracellular and Extracellular Bacteria?

A falling extracellular bacterial count does not necessarily mean that infection is resolving. Bacteria may enter host cells, replicate there, and later return to the extracellular space when infected cells lyse. A useful model must therefore track where viable bacteria are, not only their total number. This distinction is important when interpreting measurements made from different sample fractions or when evaluating an intervention that changes uptake, intracellular survival, or host-cell lysis.

The model below separates uninfected host cells, infected host cells, viable intracellular bacteria, and viable extracellular bacteria. It is deliberately a homogeneous, deterministic model: each infected cell is represented by the same average rates, and bacterial numbers are treated as continuous quantities. The model is intended to describe uptake, replication, and lysis, rather than every immune or tissue process that might affect them.

Definition: Let \(H\) be the density of uninfected host cells, \(I\) the density of infected host cells, \(Q\) the density of viable intracellular bacteria, and \(B\) the density of viable extracellular bacteria. Densities of host cells are measured in cells per volume and bacterial densities in bacteria per volume. Let \(\beta\) be the uptake coefficient, \(r_i\) and \(r_e\) the intracellular and extracellular bacterial replication rates, \(\delta_i\) the intracellular bacterial loss rate, \(\mu_i\) the nonlytic loss rate of infected cells, \(k_l\) the host-cell lysis rate, \(c_e\) the extracellular bacterial clearance rate, and \(\rho\) the viable fraction of intracellular bacteria released at lysis.

A Compartment Model and Its Assumptions

Assume that host-cell density \(H\) is approximately constant over the time window of interest. Each successful uptake event infects one previously uninfected host cell and introduces one viable bacterial founder. Uptake is proportional to \(B\) and \(H\), so its rate per volume is \(\beta BH\); \(\beta\) has units of volume per cell per time when the density units are used consistently. Infected cells are lost either nonlytically, at rate \(\mu_i\), or by lysis, at rate \(k_l\). Intracellular bacteria replicate at rate \(r_i\), are lost at rate \(\delta_i\), and are removed from the intracellular compartment when their host cell is lost.

Extracellular bacteria replicate at rate \(r_e\), are cleared at rate \(c_e\), and are removed from that compartment by uptake. Lysis releases the fraction \(\rho\) of the intracellular bacteria present in the lysed cells as viable extracellular bacteria; \(0\leq\rho\leq1\). With these assumptions, the balance equations are:

\begin{aligned} \frac{dI}{dt} &= \beta BH-(\mu_i+k_l)I,\\ \frac{dQ}{dt} &= \beta BH+(r_i-\delta_i-\mu_i-k_l)Q,\\ \frac{dB}{dt} &= (r_e-c_e)B-\beta BH+\rho k_l Q. \end{aligned}

Each term represents an entry into or exit from a compartment. For example, \(\beta BH\) adds infected cells and intracellular bacteria, while removing bacteria from \(B\). The term \(-k_lQ\) removes intracellular bacteria when their host cells lyse, and \(+\rho k_lQ\) returns the viable released fraction to \(B\). These terms do not say that lysis creates bacteria: replication changes bacterial number, while lysis transfers viable bacteria between compartments, with the unreleased or nonviable fraction lost from the measured viable burden.

The equations assume that uptake events are single-founder infections, that rates are constant over the modeled interval, and that infected cells share a common lysis hazard. Holding \(H\) fixed is an approximation, appropriate only when depletion or replenishment of uninfected cells has little effect on uptake. If host-cell depletion matters, \(H\) needs its own balance equation. Likewise, if uptake is not proportional to \(BH\), the uptake term must be replaced by a mechanism supported by the setting.

Intracellular Burden Within an Infected-Cell Cohort

A useful way to interpret intracellular measurements is to consider a cohort already infected at time zero, with no further uptake into that cohort. Its infected-cell density declines as cells are lost, while its aggregate intracellular burden changes through replication, bacterial loss, and removal of bacteria when infected cells are lost. The cohort equations are \(dI/dt=-(\mu_i+k_l)I\) and \(dQ/dt=(r_i-\delta_i-\mu_i-k_l)Q\).

Dividing aggregate burden by infected-cell density gives the mean viable burden per infected cell, \(m=Q/I\). Differentiating this ratio shows why cell loss and bacterial growth should not be conflated:

\begin{aligned} \frac{dm}{dt} &=\frac{I(dQ/dt)-Q(dI/dt)}{I^2}\\ &=\frac{I(r_i-\delta_i-\mu_i-k_l)Q+Q(\mu_i+k_l)I}{I^2}\\ &=(r_i-\delta_i)m. \end{aligned}

Thus, under the model’s nonselective loss assumption, lysis and nonlytic infected-cell loss reduce the number of cells and the aggregate intracellular burden but do not directly change the mean burden among surviving infected cells. In contrast, the aggregate burden \(Q\) has net rate \(r_i-\delta_i-\mu_i-k_l\). These are different quantities: a mean per infected cell can rise while total intracellular burden falls if infected cells are lost sufficiently quickly.

Key result: For a closed cohort with constant rates, \(Q(t)=Q(0)\exp[(r_i-\delta_i-\mu_i-k_l)t]\), whereas \(m(t)=m(0)\exp[(r_i-\delta_i)t]\). Positive net intracellular burden growth requires \(r_i>\delta_i+\mu_i+k_l\); positive mean burden growth among surviving infected cells requires only \(r_i>\delta_i\).

Worked Example: Separating Mean Burden from Total Burden

Suppose a hypothetical cohort begins with \(I(0)=1{,}000\) infected cells per mL and \(Q(0)=10{,}000\) viable intracellular bacteria per mL. Let \(r_i=0.80\) per day, \(\delta_i=0.20\) per day, \(\mu_i=0.05\) per day, and \(k_l=0.15\) per day. Assume no new uptake. Calculate the cohort burdens after 3 days.

The aggregate intracellular net rate is \(0.80-0.20-0.05-0.15=0.40\) per day. Therefore,

Q(3)=10{,}000\exp(0.40\times3) =10{,}000\exp(1.20) \approx 33{,}200\ \text{bacteria/mL}.

The infected-cell loss rate is \(0.05+0.15=0.20\) per day, so

I(3)=1{,}000\exp(-0.20\times3) =1{,}000\exp(-0.60) \approx 549\ \text{cells/mL}.

The mean burden is initially \(10{,}000/1{,}000=10\) bacteria per infected cell. After 3 days it is \(33{,}200/549\approx60.5\) bacteria per infected cell, also obtained from \(10\exp[(0.80-0.20)3]\approx60.5\). Here both aggregate burden and mean burden rise, but for different net rates. The rise in aggregate burden includes the consequence of cell losses; it does not mean every infected cell survived.

Lysis as a Source of Extracellular Bacteria

Lysis couples the intracellular and extracellular compartments. Its instantaneous viable release flux is \(\rho k_lQ\), measured in bacteria per volume per time. Even when extracellular bacteria are being cleared faster than they replicate, this input can temporarily make their measured density rise. Conversely, a declining extracellular count can coexist with a growing intracellular burden if release is small, delayed relative to the observation window, or outweighed by extracellular clearance.

For a simple closed infected-cell cohort, let \(g_i=r_i-\delta_i-\mu_i-k_l\), and suppose there is no further uptake after time zero. Then \(Q(t)=Q(0)e^{g_it}\). If the extracellular net rate \(a=r_e-c_e\) is constant, the extracellular equation is \(dB/dt=aB+\rho k_lQ(0)e^{g_it}\). For \(a\ne g_i\), integrating this linear equation gives:

B(t)=B(0)e^{at} +\rho k_lQ(0)\frac{e^{g_it}-e^{at}}{g_i-a}.

The first term is the trajectory that would occur from the initial extracellular bacteria alone. The second term is the contribution from lysis-driven release. It is nonnegative for \(t\geq0\), because the numerator and denominator have the same sign. If \(a=g_i\), the corresponding expression is \(B(t)=e^{at}[B(0)+\rho k_lQ(0)t]\). These formulas rely on constant rates and the stated no-further-uptake assumption; with ongoing uptake, \(B\) and \(Q\) must be solved as a coupled system.

Worked Example: Intracellular Persistence Despite Extracellular Decline

Consider a hypothetical infected-cell cohort with \(Q(0)=10{,}000\) bacteria/mL and \(g_i=0.40\) per day. Let the extracellular net rate be \(a=-0.70\) per day, the lysis rate \(k_l=0.15\) per day, the viable release fraction \(\rho=0.80\), and \(B(0)=100\) bacteria/mL. Assume no further uptake. Calculate \(B(3)\).

Without lysis-driven input, the initial extracellular burden would decline to \(100e^{-0.70(3)}\approx12.2\) bacteria/mL. Including release:

\begin{aligned} B(3) &=100e^{-0.70(3)} +(0.80)(0.15)(10{,}000) \frac{e^{0.40(3)}-e^{-0.70(3)}}{0.40-(-0.70)}\\ &\approx12.2+1{,}090.91(3.3201-0.1225)\\ &\approx3{,}501\ \text{bacteria/mL}. \end{aligned}

The intracellular cohort grows because \(g_i>0\), although extracellular bacteria in the absence of release would decline because \(a<0\). The larger extracellular count at day 3 is attributable to the model’s release input, not to positive extracellular net growth. In a real interpretation, measurement of extracellular bacteria alone would not identify which source generated the increase.

Using Measurements to Distinguish the Compartments

Compartment-specific measurements are needed to distinguish replication from redistribution. For example, extracellular samples quantify \(B\), while a validated procedure that removes or neutralizes extracellular bacteria before cell disruption can estimate viable intracellular bacteria \(Q\). Measurements should be aligned in time and interpreted with the sampling process in mind: incomplete removal of extracellular bacteria can make an intracellular assay appear higher, and cell lysis during handling can transfer bacteria between measured fractions.

A log-slope of aggregate intracellular burden in a closed cohort estimates the net rate \(g_i=r_i-\delta_i-\mu_i-k_l\), not the replication rate \(r_i\) by itself. Separately measuring infected-cell counts can help distinguish changes in aggregate burden from changes in burden per infected cell, but it does not by itself identify every rate. For example, the mean per-cell slope estimates \(r_i-\delta_i\) under the cohort assumptions; replication and intracellular killing remain combined.

Worked Example: Interpreting Compartment-Specific Slopes

In a hypothetical experiment, extracellular bacteria are neutralized before host cells are disrupted. Viable intracellular burden in a tracked infected-cell cohort changes from \(200\) to \(800\) bacteria over 4 hours. In a separate extracellular-only control, viable burden changes from \(1{,}000\) to \(250\) bacteria over the same interval. Assume each burden follows an exponential trajectory over these measurements.

The intracellular aggregate net slope is

\frac{\ln(800/200)}{4} =\frac{\ln(4)}{4} \approx0.347\ \text{per hour}.

This supports positive net growth of the measured intracellular cohort during the interval. It does not establish that intracellular replication alone is \(0.347\) per hour: in the model, the slope combines replication, bacterial loss, nonlytic infected-cell loss, and lysis.

The extracellular-only slope is

\frac{\ln(250/1{,}000)}{4} =\frac{\ln(0.25)}{4} \approx-0.347\ \text{per hour}.

Thus, the hypothetical measurements show increasing intracellular burden alongside declining extracellular burden. They are consistent with compartment-specific persistence, but the slopes alone cannot determine whether later extracellular counts will rise: that also depends on lysis, viable release, and extracellular clearance.

Diagnostics, Alternatives, and Limitations

The model is useful when the main question concerns transfers between extracellular and intracellular compartments. It is not a substitute for checking whether the assumed compartments and rates match the experiment. Useful diagnostics include plotting intracellular and extracellular measurements separately, checking that fitted trajectories reproduce both rather than only their sum, and examining whether inferred rates are stable across the time window. Systematic curvature on a log-burden plot may indicate time-varying rates, changing host-cell availability, or a delay not represented by the model. An extracellular increase that precedes the measured lysis signal would also challenge a model in which lysis is the only intracellular source.

A simpler extracellular-only model may be adequate if intracellular burden is negligible or the study measures only a short interval before meaningful uptake. It cannot explain a release-driven rebound. The compartment model here is more mechanistic, but it may still be inadequate if uptake is saturated, host-cell depletion is substantial, lysis has a distributed delay, or infected cells differ systematically in bacterial burden or lysis timing. A more detailed model can represent those processes explicitly, but additional states and parameters require data that can support their estimation.

A common mistake is to call any increase in extracellular bacteria “extracellular growth.” That conclusion does not follow when infected cells can release bacteria. The balance equation is \(dB/dt=(r_e-c_e)B-\beta BH+\rho k_lQ\); therefore, \(dB/dt>0\) can occur even when \(r_e-c_e<0\), provided release exceeds the combined extracellular losses. Another mistake is to interpret a positive slope in total intracellular burden as a direct estimate of replication. The correction is to identify the measured compartment and use the net rate that corresponds to it, while acknowledging which mechanisms remain combined.

Key takeaway: Intracellular bacterial replication, loss of infected cells, and lysis jointly determine intracellular burden, while extracellular burden also reflects growth, clearance, uptake, and release. Interpret each observed slope in the compartment and measurement context that produced it.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

In a closed infected-cell cohort, what is the net rate governing aggregate intracellular burden \(Q\)?
If extracellular net growth is negative, can extracellular burden still rise? State the condition in terms of the model’s fluxes.
For the cohort in the first worked example, what is the mean intracellular burden per infected cell at time zero?
Why does an intracellular log-slope not generally identify the intracellular replication rate alone?

Solutions. (1) The aggregate net rate is \(r_i-\delta_i-\mu_i-k_l\), because replication adds bacteria while intracellular loss and infected-cell loss remove them from \(Q\). (2) Yes. From \(dB/dt=(r_e-c_e)B-\beta BH+\rho k_lQ\), burden rises when \(\rho k_lQ>(c_e-r_e)B+\beta BH\), for \(r_e-c_e<0\). (3) \(Q(0)/I(0)=10{,}000/1{,}000=10\) bacteria per infected cell. (4) The observed aggregate slope combines replication with bacterial loss, nonlytic infected-cell loss, and lysis; without additional information those processes cannot be separated by that slope alone.