Coronavirus rotational diffusivity

Shortly after the confirmation of first infection, one team had already discovered and published¹ in exquisite detail about the new coronavirus, along with how it differs from previous viruses. We call the virus particle causing the COVID-19 disease SARS-CoV-2, a spherical capsid covered with hollow spikes termed peplomers. Since the virus is not motile, it relies on its own random thermal motion, specifically, the rotational component of this thermal motion, to align its peplomers with targets. From Fig. 1(B) of Ref. 2, we learn that to perfuse capsid contents into a cell, precisely two adjacent peplomers must align with a dimeric target, nominally rectangular (110 × 160 Ų). Furthermore, this alignment must be long enough for fusion. Once fused, perfusion progresses to infection.

Whereas much of the prior work on flow and the virus focuses on infection of the organism,^3–21 this work targets the transport properties of the coronavirus particle and their implications in transport phenomena of cellular infection. Although our work is mainly driven by curiosity, it may deepen our understanding or even accelerate drug treatment or vaccine development, especially where these interfere with cellular infection.

The governing transport property for the virus to attack successfully is the rotational diffusivity of the SARS-CoV-2 particle (see Footnote 2 in p. 62 of Ref. 22). Too much rotational diffusivity and the alignment intervals will be too short to properly infect, and the peplomer is wasted. Too little rotational diffusivity, and too few alignments are produced to properly infect. The rotational diffusivity of a particle depends intimately on its shape.

Whereas in engineering, the complex viscosity function has a broad diversity of applications including polymer or suspension processing, for virus suspensions, its main use is for determining rotational diffusivity. In this paper, we calculate the complex viscosity and thus the rotational diffusivity of four classes of virus particles of ascending geometric complexity: tobacco mosaic, gemini, adeno, and corona. The gemini²³ and adeno²⁴ viruses share icosahedral bead arrangements, and for the corona virus, we use polyhedral solutions to the Thomson problem to arrange its peplomers.^25,26 We employ general rigid bead–rod theory to calculate complex viscosities and rotational diffusivities, from first principles, of the virus suspensions (Refs. 27–34; see EXAMPLE 16.7-1 of Ref. 22 or EXAMPLE 13.6-1 of Ref. 35).

We are attracted to general rigid bead–rod theory first for its flexibility. We design each macromolecular structure, here virus particles, by rigidly connecting nearest bead centers with massless dimensionless rods. We are attracted to general rigid bead–rod theory second for the accuracy of its simplest special cases, the rigid dumbbell suspensions, for which many transport properties are predicted (see this reviewed in Sec. I of Ref. 31).

General rigid bead–rod theory proceeds from the continuity equation for the macromolecular configuration, called the diffusion equation [Eq. (13.2-13) of Ref. 22]. By continuity, we mean that the diffusion equation conserves orientation, preserving one and only one orientation per macromolecule. Hassager solves the diffusion equation for the configuration distribution function in small-amplitude oscillatory flows, which, for rigid macromolecules, reduces to the orientation distribution function, ψ(θ, ϕ, t).

Consider, for instance, a coronavirus particle close enough to fuse with a dimeric receptor.² We refer the coronavirus particle to spherical coordinates and consider the receptor target orthogonal to its equator (θ = π/2) with the long axis of the dimeric receptor along the longitudinal direction. For this special infection opportunity, the probability of finding a peplomer aligned with said receptor is given by [see Eq. (9) of Ref. 36]

where for the nominally rectangular binding target,

For fusion, we, of course, require two peplomers with said alignment,² and thus, the probability of finding this falls well below p.

General rigid bead–rod theory connects ψ with macromolecular shapes, including those of viruses. In this way, the virus shape confers the transport properties to its suspension, including viscosity, elasticity, and diffusivities, be they rotational or translational. Little is known experimentally about the diffusivity of viruses, especially the rotational diffusivity. For instance, the translational diffusivity of the adenovirus has been measured by photon-correlation spectroscopy.³⁷ The rotational diffusivity of tobacco mosaic viruses has also been measured by light scattering,^38–41 transient electric birefringence,⁴² and flow birefringence.⁴³ The rotational diffusivity is deducible from the translational one by the identity given in Sec. II below.

One of the challenges of ab initio calculations from general rigid bead–rod theory on coronaviruses is that the peplomer arrangement is not known. However, we do know that the spikes are charge-rich,^44,45 and we can presume, charged identically. Furthermore, we know that the coronavirus spikes are anchored into its viral membrane and not into its capsid (Sec. 1. of Ref. 46), unlike the adenovirus spikes. Hence, the coronavirus spikes are free to be rearranged by their electrostatic repulsions. We thus expect the peplomers to arrange themselves by repelling one another into the polyhedral solutions to the Thomson problem.^25,26 By the Thomson problem, we mean how identically charged particles will organize themselves onto a sphere by minimizing system potential energy. In this work, we are thus using minimum potential energy peplomer arrangements for our coronavirus model particles.

The rotational alignment of the virus particle studied herein is prefusion and not to be confused with the postfusion diffusive rotational search of the spike-protein unfolding that accompanies binding.⁴⁷ 

Using general rigid bead–rod theory, we propose the construction of virus particles from sets of beads whose positions are fixed relative to one another. For example, the SARS-CoV-2 particle geometry is a spherical capsid surrounded by a constellation of protruding peplomers. We understand that the number of peplomers per virus particle differs from particle to particle and seems to decrease with time after inoculation. We suspend our bead–rod models of virus particles into a Newtonian solvent. We begin by neglecting interactions of the solvent velocity fields, be they (i) between nearest beads within the virus particle^48,49 or (ii) between nearest virus particles. To any such collection of bead masses, we can associate a moment of inertia ellipsoid (MIE) whose center is the center of mass and whose principal moments of inertia match those of the virus particle. The MIE thus determines the orientability of the virus particle and thus the virus rotational diffusivity. Our use of moment of inertia ellipsoids is not to be confused with replacing the virus particle with an ellipsoid of revolution, with its own hydrodynamic environment.⁵⁰ We know of no previous calculation of the moments of inertia ellipsoid of virus particles, and we think that this missing physics can deepen our understanding of SARS-CoV-2.

To model the virus particle, we locate each bead of mass m_i with the position vector of the ith bead r_i, where the virus particle center of mass R satisfies

so that

where N is the total number of beads and $M≡∑i=1Nmi$ is the virus particle mass. Since we construct our virus particles with identical beads of diameter d and mass m, then M…mN, and thus, the center of mass is

which we will use below.

We next install viral coordinates at the center of mass of the virus, and we orient these Cartesian coordinates such that $δ^3$ is along the polar axis of the moment of inertia ellipsoid. For our virus particles, $δ^3$ is through the particle. In this study, to allow us to explore the surface density of peplomers, the peplomer arrangement, and even the triadic details of the three-glycoprotein spikes, we will use a finely beaded sphere for the capsid. By necessity of general rigid bead–rod theory, our capsid and peplomer beading must be equally fine.

The position vector of the ith bead with respect to the virus center of mass is given by

We define the principal moments of inertia I₁, I₂, and I₃ by [Eqs. (16.7-17) and (16.7-18) of Ref. 22 or (13.6-17) and (13.6-18) of Ref. 35]

where the subscript i is the bead number. We design each virus particle structure by first rigidly connecting nearest bead centers with massless widthless rods. Throughout our work, L is the distance between the nearest bead centers. We then complete the general rigid bead–rod construction by rigidly connecting the remaining bead centers to their nearest neighbors. For the SARS-CoV-2 particle, L is the center to center distance between osculating beads forming the capsid. Although the peplomer is a spike with a bulbous triadic head, in this work, we will model it as a single bead not touching the capsid.

Since the virus particle structure is axisymmetric, so will be its moment of inertia ellipsoid. By axisymmetric, we mean that both the virus particle and its moment of inertia ellipsoid have at least one axis of symmetry.²⁷ Furthermore, if the virus particle structure is axisymmetric, at least two of its principal moments of inertia equate, at any angle from the molecular axis, so that I₁ = I₂. Our usage of axisymmetric is not to be confused with the common geometric meaning of continuous rotational symmetry about an axis.

Hassager derives the expression for the dimensionless shear relaxation function for general rigid bead–rod theory,

in which [Eq. (16.7-38) of Ref. 22 or Eqs. (13.6-44), (13.6-45), and (13.6-46) of Ref. 35]

and the particle rotation constant is^29,31

where 0 ≤ b ≤ 3/5 and 0 ≤ aν ≤ 7/2. The three quantities a, b, and ν thus define completely the differences in linear viscoelastic behaviors arising between different axisymmetric macromolecular structures. Whereas we associate a with the Dirac delta function contribution to the relaxation function, we associate b with the dying exponential.

The relaxation time of the corresponding virus particle suspension can be expressed as

in which the bead friction coefficient is given by

We define a characteristic time for all virus particle suspensions as

which nondimensionalizes as

where φ is the bead volume fraction, and for osculating beads, where L = d,

and

Dividing Eq. (14) by Eq. (16) normalizes the relaxation time,

We can then use Eq. (10) to calculate the polymer contribution to the stress tensor in any linear viscoelastic flow, including oscillatory shear flow, from [Eq. (1) of Ref. 30]

where all symbols are defined in Tables I and II.

In this work, we apply these derivations to virus particles, specifically to the calculation of the rotational diffusivity, given by the identity (see Footnote 2 of p. 62 of Ref. 22)

about which, for virus particles, remarkably little is known. Substituting Eq. (20) into Eq. (22) and nondimensionalizing,

Substituting Eq. (17) into this and rearranging gives the dimensionless rotational diffusivity

from which we uncover a characteristic time for each virus particle suspension λ_s. The quantity ν thus defines completely the rotational diffusivity of a virus particle.

In the tradition of the transport sciences, we define the rotatory diffusivity as (see Footnote 2 of p. 62 of Ref. 22)

which, for any axisymmetric macromolecule, from general rigid bead–rod theory, gives

which has the dimensions of diffusivity and which is four times the translational diffusivity,

or

In this paper, we depart from said transport tradition of using the rotatory diffusivity, D_rot, and frame our results in terms of the rotational diffusivity, D_r.

The challenge in determining the rotational diffusivity of a virus particle, from first principles, begins with modeling its intricate geometry with beads, locating the position of each bead. Once overcome, the next challenge is to use this geometry to arrive at the transport properties for the SARS-CoV-2 particle. From these, we will deepen our understanding of how these remarkable particles can align their peplomers both for long enough and often enough to infect.

For this work, we chose general rigid bead–rod theory for its flexibility and accuracy (Sec. I of Ref. 31). However, for bead–rod structures as complex as coronaviruses, drawing the bead–rod models presented a challenge, which we met using solid modeling computer-aided design.⁵¹ This challenge arises when progressing from the R values in Eq. (5) to bead–rod imagery, for instance, when going from Table III for the R values of our tobacco mosaic and gemini viruses to our images in Figs. 1–5, respectively.

In Secs. IV–VII, we calculate the rotational diffusivity along with the complex viscosity of four classes of virus particles of ascending geometric complexity: tobacco mosaic, gemini, adeno, and corona. Section IV affords a comparison of our general bead–rod theory with measured behavior of the complex viscosity. Section V is purposed to explore how fine structural detail affects virus rotational diffusivity. Section VI affords a comparison with the measured value of the translational diffusivity, and Sec. VII, affords an exploration of how the detailed structure of SARS-CoV-2 affects its rotational diffusivity.

One measures the complex viscosity in oscillatory shear flow generated by confining the fluid to a simple shear apparatus and then by subjecting one solid–liquid boundary to a coplanar sinusoidal displacement, generating the corresponding cosinusoidal shear rate

such that the rate of deformation tensor is given by

Using the characteristic relaxation time of the virus suspension, λ, we can nondimensionalize Eq. (29) as

where λω and $λγ̇0$ are the Deborah and Weissenberg numbers. In this paper, we focus on small-amplitude oscillatory shear flow (SAOS). For this flow field, for the molecular definition of small amplitude, general rigid bead–rod theory yields

whose left side is the macromolecular Weissenberg number. From Eq. (32), we learn that structures with higher ν will have lower limits for linear viscoelasticity.

Substituting Eqs. (10) and (29) into Eq. (21) yields the polymer contribution to the shear stress

in which [Eqs. (40) and (41) of Ref. 31]

where

is the complex viscosity.^52,53 In this paper, we plot the real and imaginary parts of the responses as functions of frequency, following the work of Ferry (Secs. 2.A.4–2.A.6 of Ref. 54) or Bird et al. (Sec. 4.4 of Ref. 55).

As ω → 0, for the polymer contribution to the zero-shear viscosity, we get

which we use for Tables VI–IX.

Following EXAMPLE 5.2-6 of Ref. 55 and specifically by setting n′ = n = 1 in Eq. (5.2-4) of Ref. 55, we can define a structure-dependent characteristic time,

which is the ratio of the first normal stress coefficient to the viscosity at zero shear rate and thus reflects fluid elasticity. We insert Eq. (44) of Ref. 31 to get the structure-dependent characteristic time

into which we insert Eq. (20) to get

which we will use below.

In this section, we test the use of general rigid bead–rod theory for predicting the complex viscosity of viruses by comparing with the measured values for tobacco mosaic virus suspensions. Although this particular virus has the form of a nanotube (see Fig. 1 of Ref. 24), since its bore is narrow, we shall approximate this rigid and rod-like virus with an osculated shish-kebab (see Table VI). From general rigid bead–rod theory we know that, for the osculated shish-kebab (TABLE XV of Ref. 31),

in which λ₀ is given by Eq. (18) and d is the diameter of the tobacco mosaic virus (d ≃ 18 nm from Ref. 56).

We will next test Eqs. (34) and (35) against the well-known behaviors of the complex viscosities of the tobacco mosaic suspensions (see Ref. 57 and Fig. 9-3 of Ref. 54). Proceeding specifically from the data in Fig. 14.5-1 of Ref. 22 and mindful of [Eq. (14.4-23) of Ref. 22],

so that

and mindful of Eqs. (4.4-16) and (4.4-17) of Ref. 55,

so that for dilute virus suspensions, where

we get

so that

We construct Fig. 6 using the best fit values of [η]₀ = 23.35 and, at T = 310.0 K, λ = 5.20 × 10⁻⁴ s and, at T = 298.2 K, λ = 8.10 × 10⁻⁴ s. By Eq. (22), these correspond to rotational diffusivities at T = 310.0 K and D_r = 321 s⁻¹ and at T = 298.2 K and D_r = 206 s⁻¹.

In passing, mindful of the caption of Fig. 14.5-1 of Ref. 22, we calculate the dimensional relaxation times for the tobacco mosaic virus suspensions in Fig. 6 at T = 310.0 K,

and at T = 298.2 K,

The fitted value [η]₀ = 23.35 falls below the reported theoretical value of [η]₀ = 27 (see the caption of Fig. 14.5-1 of Ref. 22).

Solving Eq. (52) for the integer number of beads in the osculated shish-kebab,

where

and where the right side of Eq. (55) is rounded to the nearest integer. Using Eq. (18) and for d ≃ 18 nm, at T = 310.0 K, we get

and at T = 298.2 K,

Thus, using Eq. (57) with the fitted value of λ = 5.20 × 10⁻⁴ s for T = 310.0 K gives Λ = 283, and so, Eq. (55) gives N = 12. Similarly, using that with λ = 8.10 × 10⁻⁴ s for T = 298.2 K gives Λ = 282, and so, N = 12. From this, we learn that the tobacco mosaic virus can be modeled with an osculated shish-kebab of 12 beads, for which Table III lists the position vectors (see Fig. 1).

Recall that, below Eq. (52), we found that at T = 310.0 K, D_r = 321 s⁻¹ (or λ₀D_r = 1.17 × 10⁻³), and at T = 298.2 K, D_r = 206 s⁻¹ (or λ₀D_r = 1.18 × 10⁻³). These values compare closely with the value predicted by Eq. (23) for an osculated shish-kebab of N = 12. Furthermore, from the available measurements (all non-rheological), the rotational diffusivity range for the tobacco mosaic virus at room temperature (20°C–25°C) is^38–43 

Our value of D_r = 206 s⁻¹, fitted to complex viscosity measurements (Fig. 6), falls just below this range.

In this section, we have approximated this rigid and rod-like virus with an osculated shish-kebab. However, its detailed structure of a narrow-bore nanotube consisting of the osculated helix of beads shown in Fig. 1. of Ref. 24 can be captured using Eqs. (5) and (6) of Ref. 28, where L = d. We leave this for another day.

In this section, we use general rigid bead–rod theory to model the gemini virus as twin truncated icosahedra. To illustrate these twin truncated icosahedra, we construct Fig. 2 from the position vectors in Table III. We then compare this twin icosahedral structure to its coarser simpler cousin, two osculating beads (L = 2R). From Table VII, we learn that the finer twin icosahedral structure of the gemini virus increases both λ/λ₀ and (η₀ − η_s)/nkTλ over two osculating beads. From this, we deepen our understanding of the role played by the finer twin truncated icosahedral structure.

By comparing the values of λ₀D_r in Table VII, we discover that a twin icosahedral gemini model gives a lower rotational diffusivity than for twin osculating beads. Using the values of 2b/aν for the gemini virus in Table VII, with Eqs. (34) and (35), we construct Fig. 7 from which we first learn that for twin truncated icosahedra, (η′ − η_s)/(η₀ − η_s) descends less sharply than for twin osculating beads. We also learn that for the twin truncated icosahedral structure, $η′′$/(η₀ − η_s) falls below that of the twin osculating beads. We thus find that the fine structure of the gemini virus matters, raising viscosity and lowering elasticity.

To arrive at the rotational diffusivity of the adenovirus, we follow the method of Sec. II. The adenovirus capsid (not including peplomers) is made up of 252 capsomers in an icosahedral arrangement (plate II of Ref. 58). Twelve of these are “pentons” with five nearest neighbors, and the other 240 are “hexons” with six nearest neighbors.

An icosahedron has 20 triangular faces, 30 edges, and 12 vertices. Each vertex holds a penton, shared by five edges and five faces. The faces are made of two nested triangular arrangements, the inner triangle being made of six hexons. Each of 120 edge hexons is shared by two faces. Each of the 120 face hexons belongs to a single face, with all six of its neighbors lying in that face.

The edges are taken to have length 5L, with the interparticle distance L equal to the diameter d of the capsomer. The 12 vertices sit on the edges of three rectangles with Cartesian coordinates given by cyclic permutations of (±β, ±1, 0) scaled by 5L/2, where $β=1+5/2$ is the golden ratio. This exploits the fact that $1+β2+(β−1)2=2$⁠. We designate the position vectors V_i of the 12 vertices in Table IV. The edge vectors are thus E_ij = V_j − V_i, where, for example, E₀₁₁₁ = V₁₁–V₀₁.

To describe face beads, we need twelve sets of five triplets ijk arising from a counterclockwise enumeration of the five neighboring vertices associated with one vertex. We encode this as follows: The notation 01: 02, 11, 06, 05, 09 means faces ijk = 010 211, 011 106, 010 605, 010 509, 010 902. Table V lists our twelve sets.

We associate two edge beads on five edges and two face beads (this is a chiral choice) on five faces with each vertex, so that each of the 12 vertices V_i with i = 1, 2, …, 12 is associated with five triplets ijk and the following 21 points: the vertex itself, V_i, to which we add vectors for the ten edge beads, $15Eij$ and $25Eij$⁠, and ten face beads, $15Eij+Eik$ and $152Eij+Eik$⁠. Following this method, we arrive at the position vectors R_i of the adenovirus capsid beads so that i = 1, 2, …, 252.

Oliver, who measured the translational diffusivity of the adenovirus,³⁷ overlooked the identity [Eq. (28)] when he wrote “the rotational diffusivity of the adenovirus appears to be zero.” Perhaps this explains why the transport property, rotational diffusivity, has been largely overlooked in virology.

From general rigid bead–rod theory [Eq. (23)], for the characteristic time of the adenovirus, we get

and then, inserting Eq. (26) gives

into which we insert the identity Eq. (27) to get

into which we next insert Eq. (16),

Now, from Ref. 37, we have the measured value of the translational diffusivity for the adenovirus at body temperature,

Substituting this into Eq. (63),

which establishes the correspondence between our general rigid bead–rod model of the adenovirus (see in Table VIII) and the adenovirus particle itself.

From the available microscopy (see Fig. 11 of Ref. 59), r_c ≃ 116 nm (between opposing vertices), and the range for the virus radius, made dimensionless with the capsid radius, is given by

We thus position one bead for each adenovirus spike along each of the 12 vertices using r_$v$/r_c = 5/4 (see Fig. 3), which satisfies Eq. (66). Using r_c ≃ 116 nm and mindful of the adenovirus geometry (see Fig. 3) and its dimensions (see Fig. 4), we get L = 12.4 nm. Inserting this into Eq. (65) gives

namely, the characteristic time, from general rigid bead–rod theory, for the adenovirus.

The coronavirus particle is a biological material whose dimensions are thus known to within biological experimental error. From the available microscopy, r_c ≃ 100 nm–133 nm,⁶⁰ and thus, the range for the virus radius, made dimensionless with the capsid radius, is given by (see Table X and Fig. 8)

Each trimeric peplomer head, consisting of three glycoproteins, is equilateral triangular when viewed along the spike axis (see Fig. 14). For the purposes of the general rigid bead–rod theory, we must replace this trimer with a sphere of radius r_b. For this sphere, we choose a diameter, 2r_b, matching the length of the equilateral triangle (compare Fig. 8 with Fig. 14). From the available published SARS-CoV spike structure, r_b ≃ 6.5 nm and the peplomer height r_$v$ − r_c ≃ 13.0 nm.⁶¹ From this, we learn that the SARS-CoV spike is equidimensional, that is, 2r_b/(r_$v$ − r_c) ≃ 1. In general rigid bead–rod theory, we approximate the bulbous SARS-CoV-2 triglycoprotein head with a single bead so that 2r_b = d. From the available microscopy, we can see that the range for the triglycoprotein head diameter, made dimensionless with the capsid diameter, is given by (see Table X and Fig. 8)

When viewed through the lens of general rigid bead–rod theory, we learn that the rotational diffusivity of the coronavirus and its associated rheological properties are conferred by the particle shape and not by the ratio r_c/r_b.