of 26
*For correspondence:
collink@illinois.edu
These authors contributed
equally to this work
Present address:
Faculty of
Medicine, King Mongkut’s
Institute of Technology
Ladkrabang, Bangkok, Thailand;
§
Department of Biomedical
Sciences, Cedars Sinai Medical
Center, Los Angeles, United
States;
#
Johns Hopkins University
School of Nursing, Baltimore,
United states;
Department of
Microbiology, University of
Illinois at Urbana-Champaign,
Urbana, United States
Competing interest:
See
page 21
Funding:
See page 21
Received:
15 March 2019
Accepted:
27 October 2019
Published:
28 October 2019
Reviewing editor:
Julie
Overbaugh, Fred Hutchinson
Cancer Research Center, United
States
Copyright Ladinsky et al. This
article is distributed under the
terms of the
Creative Commons
Attribution License,
which
permits unrestricted use and
redistribution provided that the
original author and source are
credited.
Mechanisms of virus dissemination in
bone marrow of HIV-1–infected
humanized BLT mice
Mark S Ladinsky
1†
, Wannisa Khamaikawin
2†‡
, Yujin Jung
, Samantha Lin
2
,
Jennifer Lam
2#
, Dong Sung An
2
, Pamela J Bjorkman
1
, Collin Kieffer
*
1
Division of Biology and Biological Engineering, California Institute of Technology,
Pasadena, United States;
2
School of Nursing, UCLA AIDS Institute, University of
California, Los Angeles, Los Angeles, United States
Abstract
Immune progenitor cells differentiate in bone marrow (BM) and then migrate to
tissues. HIV-1 infects multiple BM cell types, but virus dissemination within BM has been poorly
understood. We used light microscopy and electron tomography to elucidate mechanisms of HIV-1
dissemination within BM of HIV-1–infected BM/liver/thymus (BLT) mice. Tissue clearing combined
with confocal and light sheet fluorescence microscopy revealed distinct populations of HIV-1 p24-
producing cells in BM early after infection, and quantification of these populations identified
macrophages as the principal subset of virus-producing cells in BM over time. Electron tomography
demonstrated three modes of HIV-1 dissemination in BM: (
i
) semi-synchronous budding from T-cell
and macrophage membranes, (
ii
) mature virus association with virus-producing T-cell uropods
contacting putative target cells, and (
iii
) macrophages engulfing HIV-1–producing T-cells and
producing virus within enclosed intracellular compartments that fused to invaginations with access
to the extracellular space. These results illustrate mechanisms by which the specialized environment
of the BM can promote virus spread locally and to distant lymphoid tissues.
DOI: https://doi.org/10.7554/eLife.46916.001
Introduction
The rapid systemic dissemination of HIV-1 from the site of initial infection to distant lymphoid tissues
is a critical component of the acute phase of HIV-1 infection. Bone marrow (BM) within the spongy
matrix of long bones functions as both a primary and secondary lymphoid organ and is the main site
for hematopoiesis of immune cells destined for tissues throughout the body. It contains cells com-
monly associated with HIV-1 pathogenesis (e.g., CD4+ T-cells, macrophages, and dendritic cells) and
BM-specific cells such as megakaryocytes and committed lymphoid and myeloid progenitor cells
that are also permissive to HIV-1 infection (
Alexaki and Wigdahl, 2008
;
Allen and Dexter, 1984
;
Folks et al., 1988
). HIV-1 infection leads to abnormalities in hematopoiesis (
Redd et al., 2007
;
Thiebot et al., 2001
;
Yamakami et al., 2004
), and virus replication can be detected in the BM of
SIV-infected non-human primates (NHPs) within days after infection (
Mandell et al., 1995
). Hemato-
poietic stem cells (HSCs) in BM can differentiate into multiple HIV-1 permissive cell types that retain
the capacity to enter the bloodstream and travel to distant tissue sites in a controlled fashion
(
Geissmann et al., 2010
;
Pabst, 2018
;
Pittet et al., 2014
), a process that could facilitate systemic
virus dissemination. The presence of varied BM cell types permissive to HIV-1 infection, the ability of
BM cells to migrate systemically, and the proposed implication of HSCs and progenitor cells (HSPCs)
as cell types that contribute to the virus reservoir (
Carter et al., 2011
;
Carter et al., 2010
;
McNamara et al., 2013
), make BM an important, but poorly understood, tissue for probing biologi-
cal mechanisms of HIV-1 dissemination and pathogenesis.
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
1 of 26
RESEARCH ARTICLE
Humanized mice (hu-mice) are a cost-effective model for imaging HIV-1 within tissues and the
only HIV-1 infection from which one can routinely obtain BM from long bone samples. Studies in
HIV-1–infected hu-mice have recapitulated and advanced important observations of HIV-1 disease,
including acute spread and systemic dissemination, latency, and target cell depletion (
Marsden and
Zack, 2015
;
Marsden and Zack, 2017
). Several hu-mouse models have been used to study HIV-1
pathogenesis, with BM, liver, thymus (BLT) hu-mice arguably containing the most complete reper-
toire of human immune cells of hu-mouse models (
Denton and Garcia, 2012
;
Marsden and Zack,
2015
). BLT mice are individually created by transplanting human liver and thymus tissues together
with autologous CD34+ HSCs into NOD/SCID/IL2R
g
null (NSG) mice. Importantly, BM from HIV-1–
infected BLT hu-mice includes human immune cells that produce virus transcripts during infection
(
Denton et al., 2008
;
Nixon et al., 2013
).
Immunofluorescence (IF) microscopy can be used to survey specific cell populations within tissues
of BLT hu-mice and other models of HIV-1 infection. Larger volume imaging (mm
3
-cm
3
) is possible
using tissue clearing methods, which render samples effectively transparent by removing lipids and
other biomolecules to enhance light penetration into the tissue (
Richardson and Lichtman, 2015
;
Tainaka et al., 2016
;
Treweek and Gradinaru, 2016
). Combining tissue clearing with light sheet
fluorescence microscopy (LSFM) allows in situ interrogation of large tissue volumes with single cell
resolution and rapid generation of quantifiable spatial information that would be difficult to achieve
with traditional immunohistochemistry techniques (
Richardson and Lichtman, 2015
;
Tainaka et al.,
2014
;
Yang et al., 2014
). Tissue clearing has allowed imaging of brain connectivity in mice
(
Chung et al., 2013
;
Susaki et al., 2014
), whole-body metastasis in mouse models of cancer
(
Guldner et al., 2016
;
Kubota et al., 2017
), and the distribution of HIV-1–infected cells within gut-
associated lymphoid tissue (GALT) and spleen (
Kieffer et al., 2017a
;
Kieffer et al., 2017b
).
Although the inherent density of bone prevents adequate clarification using many tissue clearing
techniques, BM imaging is possible using Bone CLARITY, which was developed to clear mouse
femurs in order to address the spatial distribution of specific cell populations (
Greenbaum et al.,
2017
).
Lymphoid tissues can be imaged at higher resolution using electron microscopy (EM), with elec-
tron tomography (ET) allowing 3D visualization of tissue volumes at ultrastructural resolution
(
McIntosh et al., 2005
). We previously reported ET studies of HIV-1 dissemination in hu-mouse
GALT (
Ladinsky et al., 2014
) and a longitudinal study of HIV-1 in hu-mouse GALT and spleen during
acute infection (
Kieffer et al., 2017b
). After using immuno-EM to verify the identity of virions in tis-
sue as HIV-1, we used ET of optimally-preserved tissues to distinguish, characterize and quantify free
mature virions, free immature virions, and budding virions within 3D tissue volumes (
Kieffer et al.,
2017b
;
Ladinsky et al., 2014
). These studies revealed 3D profiles of free and budding HIV-1 virions
and documented large pools of extracellular cell-free virions localized between cells and occasional
instances of virological synapses, supporting roles for both cell-free and cell-to-cell virus dissemina-
tion in these hu-mouse tissues (
Kieffer et al., 2017b
;
Ladinsky et al., 2014
). In contrast, our ET
studies of murine leukemia virus (MLV) in secondary lymphoid mouse tissues showed extensive zones
of contact between infected macrophages and B-cells, virions in the cell-cell interface (virological
synapses), and transfer of virions from infected donor cells to uninfected target cells via uropod
extensions (
Sewald et al., 2015
). The potential roles of cell-free versus cell-associated virus transmis-
sion via cellular structures for HIV-1 in BM remain unexplored.
To address mechanisms of HIV-1 dissemination in BM, we employed a multiscale imaging
approach: Bone CLARITY (
Greenbaum et al., 2017
) combined with IF, confocal, and LSFM to quan-
tify the distributions of specific cell types associated with HIV-1 within intact volumes of BM; and ET
to detect individual virions and virion-producing cells to characterize aspects of HIV-1 infection at
the subcellular level. These imaging studies allowed visualization of HIV-1 distribution within larger
volumes of tissue from a hu-mouse model and revealed distinct mechanisms of virus dissemination
within BM.
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
2 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
Results
Generation of HIV-1–infected BLT mice
To investigate mechanisms of HIV-1 dissemination in BM, BLT hu-mice were generated by transplan-
tation of human fetal liver-derived CD34+ HSPCs via retro-orbital vein plexus injection and trans-
plantation of pieces of human liver and thymus under the mouse kidney capsule in 8-week-old
irradiated NSG mice. Human multi-lineage hematopoietic cell populations (hCD45+ hematopoietic,
hCD3+ T-cell, hCD19+ B-cell, hCD4+ T-cell and hCD8+ T-cell) were reconstituted in peripheral
blood at 13 weeks post transplantation (
Figure 1—figure supplement 1A,B
). BLT hu-mice were
infected with CCR5-tropic HIV-1
NFNSX
at a dose of 200 ng of p24 gag via retro-orbital vein plexus at
14 weeks. Animals were euthanized and sternums and femurs were harvested to investigate HIV-1
dissemination in BM at 5 days (n = 1), 10 days (n = 2), 63 days (n = 1) and 92 days (n = 1) post-infec-
tion (
Figure 1—figure supplement 1C
). The hCD4/CD8 ratios in peripheral blood were unaltered at
5 and 10 days post-infection and subsequently declined to low levels (
Figure 1—figure supplement
1D
). Accordingly, the HIV-1 viral load in peripheral blood was low at 0 (m410) and 5 (m419) days
post-infection, but rapidly reached greater than 1.25

10
5
copies of HIV-1 RNA/mL for all later
timepoints (
Figure 1—figure supplement 1E
).
IF surveys of BM cell distributions from HIV-1-infected hu-mice
Sternums and femurs from HIV-1–infected BLT hu-mice were fixed and cleared by Bone CLARITY
(
Greenbaum et al., 2017
) (
Figure 1A
).
~
1–2 mm regions of cleared sternum containing a central
channel of BM were imaged by confocal fluorescence microscopy after immunostaining for cell
nuclei, HIV-1 p24, human CD4 (hCD4)-expressing cells, and human CD68 (hCD68)-expressing macro-
phages (
Figure 1B,C
). Compared with other lymphoid tissues, BM does not contain a dense distribu-
tion of cells (
Figure 1B
). This approach allowed us to localize hCD4+ cells, hCD68+ macrophages,
and cells that were negative for both markers in addition to highlighting the stark differences in size
and morphology exhibited by these cell types. (
Figure 1C
). HIV-1 p24+ co-localized with both hCD4
+ and hCD68+ cells. These results showed that hCD4+ and hCD68+ cells are present within regions
of intact, cleared BM from HIV-infected BLT hu-mice, and confirmed HIV-1 p24 protein expression
associated with both cell types as evidenced by co-staining with HIV-1 p24.
To obtain information from larger BM volumes, we analyzed femurs from HIV-1–infected hu-mice.
Low-level autofluorescence from cleared and decolorized femurs was recorded using LSFM to gener-
ate a reference volume for imaging after immunostaining (
Figure 1D
). Imaging of a
~
5 mm long
and
~
2 mm diameter region of immunostained femur from a 10 day post-infection (PI) hu-mouse
showed numerous hCD4+ and hCD68+ cells dispersed throughout the BM but relatively few p24+
cells (
Figure 1D
). Segmentation and quantification to detect the density of the individual cell popu-
lations showed that 626 per mm
3
hCD4+ cells, 766 per mm
3
hCD68+ T-cells, and five per mm
3
p24+ cells were present within the sample volume. Additional timepoints PI were imaged and quan-
tified, demonstrating a constant density of hCD68+ cells, a decreased density of hCD4+ cells, and a
low density of p24+ cells that increased slightly over time (
Figure 1E
). The reduction of hCD4+
T-cells is consistent with NHP studies showing SIV-induced T-cell depletion in BM (
Hoang et al.,
2019
). Co-localization analysis indicated that the majority of p24+ cells were hCD68+, with this level
remaining consistent over time, whereas the percentage of p24+/hCD4+ cells decreased with time
(
Figure 1E
).
ET surveys of cell and virus distributions in infected BM
In direct comparisons of ET imaging of BM obtained by needle aspiration versus after dissection and
removal from long bones, we found that BM extracted by needle aspiration was contaminated with
blood from the vasculature, and we could only obtain physiologically-relevant regions of BM that
were distinct from contaminating mature red blood cells at the EM level by extracting BM samples
from long bones (unpublished results). We therefore used intact BM tissue taken from HIV-1–
infected BLT mice after dissection and removal from long bones for ET. Specifically, BM tissues were
obtained by removing
~
1 cm of the sternum bone immediately after the mouse was sacrificed and
then placing it in a fixative after which the sternum was opened and intact BM was removed without
disruption of the underlying tissue structure.
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
3 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
Figure 1.
Clearing and imaging of HIV-1–infected BLT BM. (
A
) Examples of fixed (left) and cleared/decolorized (right) sternums and femurs from HIV-1-
infected BLT hu-mice. Squares are 5 mm by 5 mm. Images of tissues before and after clearing were captured on an iPhone 5 (Apple). (
B
) Representative
confocal Z-slice of cleared sternum from an HIV-1–infected hu-mouse 63 days post-infection immunostained for hCD4 (magenta), hCD68 (red), HIV-1
p24 (green), and nuclei (blue). (
C
) Gallery of zoomed confocal Z-slices showing characteristic sizes and morphologies of hCD4+ cells (magenta), hCD68
+ cells (red), and p24+ cells (green). Arrows indicate p24+ cells. (
D
) Volume of autofluorescence from cleared femur of an HIV-1 infected hu-mouse 10
days post-infection captured with light sheet microscope (left). Segmented model of bone from the same volume (center left). Dashed red box shows
region of interest for subsequent panels. Light sheet volume of a region of BM immunostained for hCD4 (magenta), hCD68 (red), and HIV-1 p24 (green)
(center right). Segmented model from the same dataset showing individual cell distributions within BM (right). (
E
) Quantification of individual cell
population densities over time. (
F
) Percent of total p24+ cells co-localizing with hCD68+ cells (red), hCD4+ cells (magenta), or not co-localized with
hCD68+ or hCD4+ cells (green) at specific times post-infection. Error bars represent standard deviations from the mean of 3 measurements from
separate volumes of tissue greater than 0.5 mm
3
each. Scale bars: (B = 20
m
m; C = 5
m
m; D = 4 mm, left panels; 500
m
m, right panels).
DOI: https://doi.org/10.7554/eLife.46916.002
The following figure supplement is available for figure 1:
Figure 1 continued on next page
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
4 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
BM from sternums of HIV-1–infected BLT hu-mice was prepared for ET by prefixation with glutar-
aldehyde and paraformaldehyde followed by high-pressure freezing and freeze-substitution (HPF/
FS) and resin embedding (
Kieffer et al., 2017b
;
Ladinsky et al., 2014
). Although we cannot use the
same samples for high-resolution ET and immuno-EM because antibody epitopes on resin-embed-
ded samples are typically destroyed (
Ladinsky and Howell, 2007
), many of the cells in BM can be
identified by morphology as we describe below. With the exception of not including budding or free
virions, uninfected BLT mouse BM was not distinguishable from HIV-1–infected BLT mouse BM in
terms of cell density, cell types, or cell morphologies (data not shown).
Large areas of HIV-1–infected BM were surveyed as EM overviews comprising 50–500 montaged
frames at
~
6500 x magnification (
Figure 2A
). These surveys revealed a diverse array of cell morphol-
ogies consistent with cells found in BM, including T-cells, B-cells, macrophages, megakaryocytes,
and HSCs/HSPCs (
Figure 2A
;
Figure 2—figure supplement 1
). Compared with tissues such as
Figure 1 continued
Figure supplement 1.
Human cell reconstitution and HIV-1 infection in BLT hu-mice.
DOI: https://doi.org/10.7554/eLife.46916.003
Figure 2.
EM and LM imaging of cell types found in BM. (
A
) EM overview of a vascularized region, showing a variety of resident cell types within BM
(labeled). (
B
) EM tomographic slice showing typical morphology of a BM T-cell (spherical shape and large nucleus-to-cytoplasm ratio). (
C
) Confocal IF
image showing an hCD4+ cell with a spherical shape and large nucleus-to-cytoplasm ratio. (
D
) Montaged projection EM overviews of a
polymorphonuclear BM macrophage in three serial 400 nm sections, demonstrating connectivity between the various lobes of the complex nucleus
(arrowheads). (
E
) Representative confocal slice of an hCD68+ cell (red) showing a multilobed nucleus (blue). Distinctly multinuclear macrophages were
not observed in BM. (
F
) Segmented 3-D volume of the cell in (
E
) showing the complex surface morphology (red) and a single nucleus (blue) with
multiple interconnected lobes.
DOI: https://doi.org/10.7554/eLife.46916.004
The following figure supplements are available for figure 2:
Figure supplement 1.
T-cells and macrophages display different modes of cell-to-cell contact for potential virus transfer.
DOI: https://doi.org/10.7554/eLife.46916.005
Figure supplement 2.
HIV-1–bearing uropods are associated with macrophages.
DOI: https://doi.org/10.7554/eLife.46916.006
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
5 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
GALT and spleen, the density of cells in hu-mouse BM was relatively low, as also found for human
BM (
Brass, 2005
). IF of an equivalent region of hu-mouse BM showed a similar distribution of cells,
including hCD4+ cells and hCD68+ cells, dispersed amongst other cell types (
Figure 1B
). By EM,
T-cells can be distinguished by their round compact shape, large nucleus, and minimal cytoplasm
(
Zucker-Franklin, 1975
) (
Figure 2B
). Cells identified by EM as T-cells were morphologically similar
to hCD4+ cells detected by IF within cleared BM (
Figure 1C
;
Figure 2C
). Cells identified as BM
T-cells regularly displayed uropods (cytoplasmic protrusions containing organelles and structures
promoting cell adhesion and signaling;
Sa
́
nchez-Madrid and Serrador, 2009
;
Sewald et al., 2015
)
that formed synaptic contacts with neighboring cells (
Figure 2—figure supplement 1A,G
), sugges-
tive of the migration characteristics exhibited by some innate immune cells (
Hind et al., 2016
;
La
̈
mmermann and Sixt, 2009
;
Sewald et al., 2015
).
Macrophages are larger than T-cells and are characterized by a complex morphology that
includes deep surface invaginations, multi-lobed nuclei, and populations of vesicles and granules of
varying size and electron density (
Orenstein, 2007
;
Orenstein and Wahl, 1999
) (
Figure 1C
; 2D-F).
Although some IF and projection EM images appeared to show multiple nuclei in presumptive mac-
rophages, serial-section ET of large 3D volumes demonstrated that individual lobes of nuclei were
connected in the
~
25 cells that we examined (
Figure 2D
), suggesting that BM macrophages are not
polynuclear; that is they do not contain multiple separate nuclei. Analogous IF data from cleared BM
confirmed these observations by showing continuous, multi-lobed nuclei (
Figure 2E,F
). BM macro-
phages also displayed numerous pseudopods emanating from their surfaces. Macrophage pseudo-
pods (
Figure 2—figure supplement 1B
) could be distinguished from uropods emanating from
T-cells (
Figure 2—figure supplement 1A
) in that they were smaller and contained organized cyto-
skeletal filaments, but lacked membranous organelles (
Figure 2—figure supplement 1A–F
). Uro-
pods were also found on macrophages (
Figure 2—figure supplement 2
), but more rarely than
found on BM T-cells. Taken together, EM and IF surveys of BM from HIV-1–infected hu-mice
revealed the presence, distribution, and morphological details of a subset of cell populations that
could be involved in HIV-1 dissemination.
Using ET to gain ultrastructural insight into mechanisms of virus dissemination within HIV-1–
infected BM, we identified three potential dissemination mechanisms.
Dissemination mechanism #1: Synchronous virion release from the cell
surface
Virions budding from the plasma membrane were detected by ET as early as 5 days PI on cells mor-
phologically identified as T-cells (
Figure 3A,B
) and macrophages (
Figure 3C,D
), consistent with
larger-volume IF imaging of cleared tissue (
Figure 1
). Interestingly, virions were often seen budding
from the surface of T-cells at points corresponding to narrow regions of underlying cytoplasm
(
Figure 3B
), suggesting either that virus components and associated host machinery can assemble in
a space-limited environment and/or that the morphology of HIV-1–infected cells is dynamic. The
finding of budding virions along the plasma membrane of virus-producing cells that were not in
direct contact with other cells (common in BM due to its low cell density) suggested that the release
of free virions from the plasma membrane is an in vivo HIV-1 dissemination mechanism.
HIV-1 virions budding from the surface of virus-producing cells in the BM often showed a series
of 2–4 thin, but electron-dense, bands that circumscribed the constricting bud neck (
Figure 3E
;
Fig-
ure 3—figure supplement 1
). We previously used immuno-EM to demonstrate that the CHMP1B
and CHMP2A ESCRT-III proteins are present at locations in budding virions where these bands of
electron density form (
Ladinsky et al., 2014
). In addition, the location of the bands within the bud
neck is consistent with their identification as components of ESCRT-III, which transiently polymerize
into spirals at the necks of budding virions to facilitate plasma membrane abscission and virus
release (
Sundquist and Krausslich, 2012
) (
Figure 3E
;
Figure 3—figure supplement 1
). Finally, their
resemblance to EM structures of polymerized ESCRT-III proteins and similarity to a ‘dome’ model of
ESCRT-mediated membrane scission (
Scho
̈
neberg et al., 2017
;
Sundquist and Krausslich, 2012
)
(
Figure 3—figure supplement 1A
) further suggests that the bands represent polymerized ESCRT-III
spirals (
Johnson et al., 2018
;
McCullough et al., 2015
;
McCullough et al., 2018
;
Sundquist and
Krausslich, 2012
).
In surveys of HIV-1 budding profiles emanating from virus-producing BM cells we confirmed that
putative ESCRT-III densities were present only during a stage in the budding process when the bud
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
6 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
Figure 3.
Detection of viruses budding from cells in BM. (
A
) EM overview of a BM region containing a virus-producing T-cell. (
B
) Tomogram of an HIV-1
budding profile emanating from the surface of a T-cell showing little cytoplasm between the nuclear envelope (NE) and the budding plasma membrane
(PM). N = nucleus; NP = nuclear pore. Inset: 3-D model of the HIV-1 bud (green; plasma membrane; orange, immature HIV-1 core). (
C
) EM overview of
a BM region containing a virus-producing macrophage (M
F
). (
D
) Tomogram showing two HIV-1 budding profiles emanating from a single stalk.
Figure 3 continued on next page
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
7 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
neck was
~
50–80% of the diameter of the budding virion itself (
Figure 3E
;
Figure 3—figure supple-
ment 1B
). The bands were not present when the bud neck was larger or smaller than that ratio, cor-
responding to earlier or later budding stages, respectively (
Figure 3—figure supplement 1B
).
Although ET of tissue samples produces a series of static images, we can use surveys of budding
profiles to deduce aspects of a dynamic process such as budding. For example, the observation that
the putative ESCRT-III densities are only present on buds that have relatively thick necks suggests
that the presence of ESCRT-III spirals connotes a transient, and specific, temporal stage in the HIV-1
budding process. This suggestion is consistent with recent TIRF microscopy studies of individual
budding viruses showing ESCRT-III accumulation at a specific stage of HIV-1 Gag accumulation and
membrane closure prior to membrane constriction (
Johnson et al., 2018
).
Virus-producing T-cells in BM often showed numerous nascent virions budding from multiple sites
along the plasma membrane (
Figure 3E
). Because our ET images represent snapshots of specific
time points in the budding process captured by rapid fixation, we infer that the nascent virions rep-
resent simultaneous budding events. In such cases, ET showed that all buds on a given cell had simi-
lar neck diameters, and that all either showed or did not show presumptive ESCRT-III bands. For
example, a virus-producing BM T-cell was found to have six budding profiles on its surface within a
400 nm thick section (
Figure 3E
). ET of each bud showed a similar neck diameter and presumptive
ESCRT-III bands circumscribing each bud neck. Nascent virions of different neck diameters or that
did not include presumptive ESCRT-III bands (
Figure 3E
, outer panels) were not observed on this
cell. These observations suggest that all virions budding from this cell were at a similar stage of
egress at the time the BM sample was fixed, supporting the hypothesis that release of virus from
virus-producing T-cells in BM occurs in a coordinated or synchronous fashion (
Ladinsky et al., 2014
);
that is we infer that if all of the buds present on a region of the cell are showing ESCRT bands, they
are all at roughly the same point in the budding process. Furthermore, the absence of buds with
ESCRT bands within that same region of the cell implies that the budding events were initiated and
proceeded at multiple points within roughly the same definable time window.
Dissemination mechanism #2: Cell-to-cell transmission of virions
attached to T-cell uropods
HIV-1 virions were found associated with, and budding from, the plasma membranes of large uro-
pods that projected from T-cells and contacted nearby cells (
Figure 4A–C
). One or two uropods
were typically seen per cell. Uropod plasma membranes usually displayed numerous caveolae-like
invaginations (
Figure 4D
), implying that uropods are an active zone for endocytosis and signaling
(
Cheng and Nichols, 2016
;
Sa
́
nchez-Madrid and Serrador, 2009
). The uropod mode of cell-to-cell
contact contrasted with the common contact mode of macrophages, in which numerous pseudo-
pods emanating from a macrophage surface engaged with multiple neighboring cells, sometimes
enveloping entire cells (
Figure 2—figure supplement 1B
). ET and 3D modeling of a zone-of-contact
between a T-cell uropod and a potential target cell showed a region between two uropod processes
that included electron-dense material into which a virion was budding (
Figure 4F,G
). Mature virions
were also found within plasma membrane invaginations of virus-producing cells that were contacting
potential target cells (
Figure 2—figure supplement 1G
). Cumulatively, these observations describe
a potential mechanism for virus dissemination in the BM by which uropods actively facilitate the
direct transfer of virions between virus-producing cells and target cells across a virological synapse,
which we define as a region of close proximity between a donor cell and a potential target cell, often
Figure 3 continued
Macrophages often exhibited multiple (up to 5) viruses emanating from a single stalk. (
E
) Projection EM image showing HIV-1 budding profiles
emanating from a virus-producing T-cell in BM (center). Six budding profiles were present on the cell within the volume of the reconstructed tomogram.
Each bud had a similar neck diameter and discernable bands of electron density (arrowheads) potentially representing polymerized ESCRT-III fission
machinery.
DOI: https://doi.org/10.7554/eLife.46916.007
The following figure supplement is available for figure 3:
Figure supplement 1.
HIV-1 budding profiles displaying ESCRT-III spirals.
DOI: https://doi.org/10.7554/eLife.46916.008
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
8 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
including zones of direct contact, that facilitates transfer of material from the donor cell to the target
cell.
Dissemination mechanism #3: Macrophage phagocytosis of virus-
producing T-cells and release of intracellular virions
BM macrophages (identified by hCD68 labeling) associated with HIV-1 p24 were detected by IF
microscopy (
Figure 1
), prompting higher resolution analysis by ET to characterize and ascertain their
role in virus dissemination. BM macrophages are characterized by having large populations of endo-
cytic compartments, phagosomes and numerous morphologically-complex invaginations of the cell
surface (
Sewald et al., 2015
). Although both BM macrophages and neutrophils in BM from BLT hu-
mice were indeed polymorphonuclear, neutrophils can be easily distinguished from macrophages by
their smaller size and populations of large, evenly dense granules. Furthermore, neutrophils lack
extensive, pleomorphic surface invaginations that are common to macrophages and are often
enhanced after infection by HIV-1 (
Deneka et al., 2007
;
Graziano et al., 2016
). Such invaginations
appeared electron-lucent in BM macrophages (
Figure 2
;
Figure 7—figure supplement 1
).
Cumulatively, ET showed several modes of BM macrophages associated with HIV-1 virions,
including phagosomes containing ingested virus-producing cells (
Figure 5
), virions budding from the
Figure 4.
HIV-1 transfer via uropod. (
A–B
) Projection EM image of cell-cell contact via uropod in BM (panel A) with
modeled donor cell uropod (magenta) and target cell (green) (panel B). (
C
) Tomographic slice of the zone of
contact between the donor uropod and target cell. Several free HIV-1 virions are visible along the length of the
uropod. (
D
) Caveolae-like invaginations (CLI) along the uropod. (
E
) A virus budding from the plasma membrane of
the uropod. (
F
) A virus budding from a uropod in contact with a potential target cell and possibly forming a
virological synapse.
DOI: https://doi.org/10.7554/eLife.46916.009
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
9 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease
cell surface (
Figure 3C,D
), virions contained within small intracellular compartments in the cytoplasm
(
Figure 6
), and virions contained within surface invaginations (
Figure 7
;
Figure 7—figure supple-
ment 1
). We estimate BM macrophages (as identified morphologically) that did not contain HIV-1
virions outnumbered macrophages associated with virions by
~
20:1.
Figure 5.
Macrophages phagocytose virus-producing cells within BM, and mature HIV-1 survives phagocytic degradation. (
A
) A typical BM macrophage
with a partially degraded cell within a phagosome. (
B
) Tomogram detail of the region indicated by the rectangle in A, showing mature HIV-1 virions
(red arrowheads) within the phagosome, adjacent to the degrading cells. Inset: Mature HIV-1 virion within an endocytic compartment adjacent to the
phagosome. (
C
) An enlarged polymorphonuclear macrophage with two phagocytosed cells at different stages of degradation. The cell in the upper
phagosome is nearly completely degraded. (
D
) Tomogram detail of the region indicated by the square in C, showing a portion of the upper
phagosome region. Mature HIV-1 virions (red arrowheads) are present within the phagosome and within an adjacent endocytic compartment that is
continuous with the phagosome. Inset (upper right): higher magnification detail of an HIV-1 virion, confirming its identity by the presence of a cone-
shaped core. Inset (lower left): Detail of a free, immature HIV-1 virion demonstrating an incomplete ‘C-shaped’ core.
DOI: https://doi.org/10.7554/eLife.46916.010
Ladinsky
et al
. eLife 2019;8:e46916.
DOI: https://doi.org/10.7554/eLife.46916
10 of 26
Research article
Human Biology and Medicine
Microbiology and Infectious Disease