Efficient transmission and persistence of low-frequency SIVmac251 variants in CD8-depleted rhesus macaques with different neuropathology

Abstract

Infection of CD8-depleted rhesus macaques with the genetically heterogeneous simian immunodeficiency virus (SIV)mac251 viral swarm provides a rapid-disease model for simian acquired immune deficiency syndrome and SIV-encephalitis (SIVE). The objective was to evaluate how the diversity of the swarm influences the initial seeding of the infection that may potentially affect disease progression. Plasma, lymphoid and non-lymphoid (brain and lung) tissues were collected from two infected macaques euthanized at 21 days post-infection (p.i.), as well as longitudinal specimens and post-mortem tissues from four macaques followed throughout the infection. About 1300 gp120 viral sequences were obtained from the infecting SIVmac251 swarm and the macaques longitudinal and post-mortem samples. Phylogenetic and amino acid signature pattern analyses were carried out to assess frequency, transmission dynamics and persistence of specific viral clusters. Although no significant reduction in viral heterogeneity was found early in infection (21 days p.i.), transmission and replication of SIV variants was not entirely random. In particular, two distinct motifs under-represented (<4 %) in the infecting swarm were found at high frequencies (up to 14 %) in all six macaques as early as 21 days p.i. Moreover, a macrophage tropic variant not detected in the viral swarm (<0.3 %) was present at high frequency (29–100 %) in sequences derived from the brain of two macaques with meningitis or severe SIVE. This study demonstrates the highly efficient transmission and persistence in vivo of multiple low frequency SIVmac251 founder variants, characterized by specific gp120 motifs that may be linked to pathogenesis in the rapid-disease model of neuroAIDS.

Introduction

Development of AIDS-related neurological disorders is increasing in human immunodeficiency virus type 1 (HIV-1)-infected patients in the era of highly active anti-retroviral therapy (HAART) (Anthony et al., 2005; Bell, 2004). It has been suggested that these disorders may be caused by ‘macrophage dysregulation’, where the accumulation, activation and infection of macrophages infiltrating brain tissues leads to neuropathology (Salemi et al., 2005; Williams & Hickey, 2002). The macaque model provides a unique platform for studying brain-related disorders as the natural course of simian immunodeficiency virus (SIV) in experimentally infected rhesus macaques (Macaca mulatta) follows a similar course to that of HIV-1 in non-ART-treated patients (Desrosiers, 1990). In particular, infection with the genetically heterogeneous SIVmac251 viral swarm constitutes a well-established model of neuroAIDS (Burudi & Fox, 2001). The histopathology and neuropathology of macaques closely resembles the development of HIV-associated dementia (HAD) in humans (Burudi & Fox, 2001; Lackner, 1994; Mankowski et al., 2002; Zink et al., 1998, 2006). Both diseases are characterized by the accumulation of perivascular macrophages, microglia and macrophage nodules, multi-nucleated giant cells and gliosis (Burudi & Fox, 2001; Lackner, 1994; Mankowski et al., 2002; Zink et al., 1998, 2006).

Identifying particulars of the initial seeding and subsequent evolution of the virus in lymphoid and non-lymphoid tissues is critical for understanding the development of AIDS-related neurological disorders (Burudi & Fox, 2001; Lackner, 1994; Mankowski et al., 2002; Zink et al., 1998, 2006). Previous studies reported the presence of specific neurotropic strains in the brain or cerebrospinal fluid (Pillai et al., 2006; Power et al., 1995) and suggested that macrophage-tropic variants present at late-stage brain infection may originate from lymphoid tissues (Liu et al., 2000); however, the effect of genetic heterogeneity of the initial transmitting/founder viral population on the evolutionary dynamics of brain infection and disease progression in the new host remains unclear.

The CD8-depleted macaque model infected with the SIVmac251 swarm provides two advantages for investigating the association of genetic variation and disease. First, the CD8-depleted model is a rapid-disease model of neuroAIDS, usually leading to meningitis and SIV-encephalitis (SIVE) in most macaques within 65–120 days (Mankowski et al., 2002). Alterations observed this accelerated model of neuroAIDS result from unrestricted viral expansion in the setting of immunodeficiency rather than from CD8⁺-lymphocyte depletion alone (Ratai et al., 2011; Williams & Burdo, 2012). Second, transmission dynamics are comparable to HIV-1 (Shankarappa et al., 1999). For example, intra-vaginal infection with SIVmac251 results in a much less diverse population of transmitting variants than intravenous infection (Greenier et al., 2001). A recent study demonstrated that multiple transmitted variants in both intravenous and intra-rectal SIVmac251 infections appeared to be randomly selected from the initial swarm and that the viruses acquired by either route represented the full genetic spectra of the inoculates (Keele et al., 2009). The present study investigated the effect of the genetic diversity of the SIVmac251 swarm on the initial seeding of the infection and disease progression in six intravenously inoculated CD8⁺-depleted macaques. Longitudinal blood and lymphoid tissue samples were collected from the macaques, as well as post-mortem samples from lymphoid (bone marrow and lymph node) and non-lymphoid (brain and lung) tissues. These samples provide a strong overview of the changes in the initial intravenous inoculum, which could not be ethically performed in humans.

Results

Macaque pathology

All six macaques were successfully infected with the SIVmac251 swarm and quickly displayed an elevated viral copy number in the plasma. Viral load in plasma was already detected in all animals at 3 days post-infection (p.i.), peaked between 7 and 14 days p.i., and remained elevated throughout the disease (Fig. 1). D01 and D02 were euthanized at 21 days p.i. to examine viral sequences from brain tissues during early infection. The remaining macaques, D03, D04, D05 and D06 were monitored until death due to severe weight loss (D03, euthanized at 75 days p.i.), colitis (D04, 91 days p.i.), CMV infection (D05, 95 days p.i.) and brain haemorrhage (D06, 118 days p.i.). All four macaques had a significant drop in CD4 counts between 8 and 26 days p.i. They all recovered in the following weeks and consistently showed CD4 >300 at the time of death (data not shown). Since these macaques are CD8 depleted, the CD4 absolute counts are not relevant for disease progression as the CD4 population increases to compensate for the CD8 loss. Three macaques (D03, D04 and D05) developed SIV encephalitis (SIVE) and one macaque (D06) had meningitis, which were diagnosed at necropsy by histopathology of brain tissues (Table 1).

Fig. 1.

SIV copy number and the CD8 absolute count over time in infected macaques. (a) SIV copy number (y-axis) is given in eq ml⁻¹; time (x-axis) is in days p.i. Inoculated macaques were treated with an anti-CD8 antibody at 6 days p.i. Macaques D01 and D02 were euthanized at 21 days p.i. Macaques D03, D04, D05 and D06 followed throughout the duration of the infection until death. (b) CD8 absolute counts (y-axis); time (x-axis) is given in days p.i. for macaques D03, D04, D05 and D06.

Table 1. Macaque pathology report and summary of gp120 sequences collected from infected macaques.

Macaque*	Pathology report†	Symptoms	Tissue	Time (days p.i.)	No. sequences
D01	Euthanized	Lymphoid hyperplasia, minimal chronic inflammation in multiple sites.	Plasma	21	46
		Bone marrow	21	9
		BAL	21	8
		Meninges	21	27
		Frontal lobe	21	26
		Parietal lobe	21	29
		Temporal lobe	21	25
D02	Euthanized	Inflammatory cells present in the meninges of frontal, parietal and temporal lobes and along the optic nerve.	Plasma	21	49
		Bone marrow	21	36
		Frontal lobe	21	29
		Parietal lobe	21	19
		Temporal lobe	21	22
D03	SIVE	Euthanized due to weight loss (probably due to the suppurative tonsillitis, chronic sialadenitis and eosinophilic gastritis). Eosinophilic perivasculitis in the brain, mild interstitial pneumonia and giant cell ganglionitis.	Plasma	21	17
		Plasma	61	21
		BAL	61	8
		Plasma	75	20
		Bone marrow	75	14
		BAL	75	21
		Lymph node	75	12
		Meninges	75	23
		Frontal lobe	75	20
		Parietal lobe	75	36
		Temporal lobe	75	17
D04	SIVE	Colitis: C. coli recovered from the gut and consistent with the lesions and clinical signs.	Plasma	21	20
		Plasma	61	21
		BAL	61	25
		Plasma	91	25
		Bone marrow	91	16
		Lymph node	91	16
		Frontal lobe	91	8
		Parietal lobe	91	15
		Temporal lobe	91	24
D05	SIVE	Typical CMV opportunistic infection that develops late in the disease process.	Plasma	21	27
		Plasma	61	12
		BAL	61	23
		Plasma	89	28
		Bone marrow	95	47
		Lymph node	95	10
		Frontal lobe	95	34
		Parietal lobe	95	21
		Temporal lobe	95	28
D06	Meningitis	Meninges of the brain, spinal cord and epineurium of ganglions are expanded by oedema, haemorrhage, macrophage and neutrophil infiltrates and admixed with numerous intracellular granular materials. Lymphoid hyperplasia of multiple lymph nodes and spleen, giant cell pneumonia and lymphoplasmacytic infiltrates multiple organs.	Plasma	21	23
		Plasma	61	23
		BAL	61	5
		Lymph node	61	12
		Plasma	118	22
		Bone marrow	118	24
		BAL	118	27
		Lymph node	118	24
		Frontal lobe	118	9
		Parietal lobe	118	21
		Temporal lobe	118	11

^*Macaques were infected with the PBMCs passaged (11 days) SIVmac251 viral swarm and treated with an anti-CD8 antibody 6 days after the inoculum.

^†Primates D01 and D02 were euthanized early at 21 days p.i. Primates D03 through D06 were euthanized at end-stage events.

SIV diversity in rhesus macaques during early infection (21 days p.i.)

A total of 1135 longitudinal SIV sequences were obtained from plasma- and macrophage-infiltrated tissues collected from the infected macaques (Table 1): 354 from plasma (median 62, range 46–68 per macaque), 220 from lymphoid tissues (median 34, range 9–60), 117 from bronchial lavage (BAL) (median 21, range 0–32) and 444 from brain tissues (median 77, range 41–107). The range for each tissue at a specific time point was 5–49 viral sequences, which provides a reasonable overview of SIV genetic variability during disease progression in the macaque model.

The nucleotide diversity in the 21 days p.i. plasma samples from the infected macaques was compared with the diversity of the inoculum to determine whether a transmission bottleneck had occurred. Diversity within plasma samples was between 0.7±0.1 and 0.9±0.1 % (Table 2) and did not significantly differ from infecting swarm (Student’s t-test, P>0.01), suggesting that multiple variants were transmitted and persisted until at least 21 days p.i. The lack of significant difference between the viral swarm and the plasma samples seemed to suggest that no transmission bottleneck occurred. In the four macaques that were not euthanized at 21 days p.i., diversity at the second time point (61 days p.i.) and at necropsy (75–118 days p.i.) ranged from 0.8±0.2 to 1.2±0.2 %. Moreover, within time-point diversity and mean divergence from the infecting viral swarm consistently increased over time in all four macaques. Increasing divergence and diversity have also been observed within HIV-infected patients (Shankarappa et al., 1999), indicating that the CD8-depleted SIV/macaque rapid-disease model recapitulates, although on a much shorter timescale, the HIV intra-host evolutionary patterns. In all macaques, the V1 domain was the most variable in terms of both nucleic acid substitutions and insertion/deletions, with lengths ranging from 36 to 47 aa. The number of glycosylation sites (1–3) was similar among macaques at all time points and there was no evidence of preferential transmission of any particular variant.

Table 2. Genetic characterization of SIV envelope sequences from longitudinal plasma samples.

na, Not applicable.

Macaque	Time point (days p.i.)	No. sequences	Mean diversity*	Mean divergence†	V1 length	V1 N-linked glycosylation sites
SIVmac251_2006‡	na	193	0.6 (0.1)	NA	36–47	2–3
D01	21	46	0.8 (0.1)	0.8 (0.1)	36–42	2–3
D02	21	49	0.7 (0.1)	0.7 (0.1)	36–47	2–3
D03	21	17	0.8 (0.2)	0.7 (0.1)	39–41	2–3
D03	61	21	1.0 (0.2)	1.0 (0.2)	38–41	2–3
D03	75	20	1.0 (0.2)	0.9 (0.2)	38–42	2
D04	21	20	0.7 (0.1)	0.6 (0.1)	38–47	2
D04	61	21	0.8 (0.2)	0.8 (0.1)	37–41	2
D04	91	25	0.9 (0.2)	0.9 (0.2)	38–42	2
D05	21	27	0.9 (0.1)	0.7 (0.1)	38–42	2–3
D05	61	12	1.0 (0.2)	0.9 (0.1)	38–42	2
D05	89	28	1.2 (0.2)	1.0 (0.2)	38–42	2–3
D06	21	23	0.8 (0.2)	0.8 (0.1)	38–42	2
D06	61	23	1.0 (0.2)	1.0 (0.2)	39–42	2
D06	118	22	1.1 (0.2)	1.2 (0.2)	38–42	2–3

^*Average nucleotide diversity (P-distance) of sequences within a specific SIV viral swarm. Values are in per cent; se are given in parentheses.

^†Mean nucleotide divergence (P-distance) of viral swarm sequences compared to the 1991 PBMCs passaged SIVmac251 viral swarm. Values are in per cent; se are given in parentheses.

^‡PBMCs passaged 1991 (11 days) SIVmac251 viral swarm from the New England Primate Center. GenBank accession numbers: JF741274–JF741921.

Cluster analysis

The transmission of specific SIVmac251 gp120 genotypes was further investigated by using a cluster analysis. To take into account the genetic diversity of the viral swarm, sequences in the inoculum were grouped on the basis of their nucleotide identity. A sequence cluster within the infecting viral swarm was defined as a group of sequences sharing ≥99.5 % nucleotide identity. The algorithm classified 193 SIVmac251 gp120 sequences within 17 discrete clusters, labelled from 1 (most frequent) to 17 (less frequent). Cluster 1 was the most represented, accounting for 41.5 % of the SIVmac251 gp120 sequences. Clusters 2–5 each included 4.2–12.4 % of the sequences, clusters from 6–17 each included <4 % of the sequences, while 2.1 % unique and highly divergent sequences were unassigned to a specific cluster (Table 3).

Table 3. Frequency of SIVmac251 gp120 sequence clusters in macaques at 21 days p.i.

Cluster*	Cluster (%)†	Percentage of gp120 sequences from plasma at 21 days p.i.‡
		D01	D02	D03	D04	D05	D06
1	41.5	33.0 (28.3)	33.5 (26.5)	26.5 (17.6)	28.9 (20)	52.5 (48.1)	33 (30.4)
2	12.4	3.0 (2.2)	0.6 (0.0)	0.0 (0.0)	11.0 (5.0)	5.6 (0.0)	7.6 (4.3)
3	10.4	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	17.5 (15)	0.0 (0.0)	0.0 (0.0)
4	5.2	2.2 (2.2)	1.0 (0.0)	5.9 (5.9)	0.0 (0.0)	7.4 (7.4)	0.0 (0.0)
5	4.2	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
6	3.1	4.3 (4.3)	2.0 (2.0)	0.0 (0.0)	0.0 (0.0)	7.4 (7.4)	0.0 (0.0)
7	2.6	2.2 (0.0)	1.1 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
8	2.6	1.8 (0.0)	8.9 (6.1)	2.9 (0.0)	3.9 (0.0)	1.2 (0.0)	12.3 (4.3)
9	2.6	6.1 (4.3)	6.1 (6.1)	5.9 (5.9)	10.0 (10.0)	7.4 (7.4)	8.7 (8.7)
10	2.1	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
11	2.1	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	10.0 (10.0)	0.0 (0.0)	0.0 (0.0)
12	2.1	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	3.7 (3.7)	0.0 (0.0)
13	2.1	3.6 (2.2)	4.4 (2.0)	0.0 (0.0)	0.0 (0.0)	3.7 (3.7)	0.0 (0.0)
14	1.6	11.2 (8.7)	19.4 (16.3)	11.8 (5.9)	16.2 (15)	3.7 (3.7)	8.0 (0.0)
15	1.6	3.3 (2.2)	2.0 (2.0)	11.8 (5.9)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
16	1.0	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
17	1.0	4.3 (4.3)	0.5 (0.0)	5.9 (5.9)	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)
UC	2.1	1.1 (0.0)	0.0 (0.0)	11.8 (5.9)	2.5 (0.0)	0.0 (0.0)	0.0 (0.0)
Total sequences	193	46	49	17	20	27	23
Classified (%)	98.2	76.1 (58.7)	79.6 (61)	82.5 (53)	100 (75)	92.6 (81.4)	69.6 (47.7)

^*A cluster is defined by a group of viral swarm sequences sharing ≥99.5 % nucleotide identity. The clusters are numbered and ordered according to the number of viral swarm sequences present in each cluster. Unclustered sequences, (UC) represent those sequences that did not share ≥99.5 % with any other sequence observed in the viral swarm.

^†Percentage of sequences in the SIVmac251 viral swarm used to infect the macaques represented in a given cluster.

^‡Plasma sequences genetically related to a given cluster differ 0–6 nt from one or more sequences of that cluster. Frequencies outside parentheses are weighted to include sequences ambiguously and unambiguously assigned to a cluster. Frequencies in parentheses are unweighted to include sequences unambiguously assigned to a cluster.

Among the six infected macaques, 47.8–81.5 % (unweighted, see Methods) and 69.6–100 % (weighted, see Methods) of the sequences sampled at the earliest time point (21 days p.i.) were successfully classified within specific clusters (Table 3). When the weighted cluster assignment was analysed, clusters 2, 3, 4 and 5 were found in five, one, four and zero macaques, respectively (Table 3). Moreover, two clusters (9 and 14), which were present at low frequency (<5 %) in the inoculum, were present in all six macaques. Since the six primates were infected intravenously using the same experimental conditions, the sequences obtained for all the primates were combined to evaluate whether the transmission of specific clusters occurred randomly. Approximately 1/3 of the sequences (35 %) present in all the macaques combined were related to cluster 1. As indicated by the sequences observed in the individual primates (Table 3 and Fig. 2), clusters 2, 3 and 5 were significantly less represented in the macaques than in the viral swarm (Fisher’s exact test, P<0.05 with a power of detection ≥80 %), indicating that less sequences from these clusters than expected by chance alone were transmitted to the primates. On the other hand, clusters 9 and 14, two under-represented (<4 %) clusters in the swarm, were significantly more represented in the infected macaques (P≤0.05), with a frequency greater than expected by chance alone (Fig. 2). Based on Fisher’s exact test, the power for detecting cluster 9 was about 58 % and greater than 99 % for cluster 14. Similar results were observed in the unweighted analysis (data not shown). Similar results were observed though, when a more stringent cut-off of 3 nt differences was used for comparing the viral swarm clusters to the plasma sequences (data not shown).

Fig. 2.

Early transmission of SIVmac251 gp120 clusters in intravenous CD8-depleted macaques. The black bar denotes the frequency of the cluster within the viral swarm. The white bar represents the frequency of the weighted cluster in all the primates combined. The asterisk shows P-values ≤0.05 based on the Fisher’s exact test.

Signature pattern analysis

Signature pattern analysis detected specific V2/V3 amino acid motifs characteristic of the sequences belonging to clusters 9 (R_AQN_S_V) and 14 (L_V), not previously reported as sites to confer macrophage-tropism (Fig. 3a). Moreover, in macaque D05, one sequence from plasma at 21 days p.i. appeared divergent when compared with all other plasma sequences. Such sequence was not close to any of the sequences in the infecting swarm (~3 % diversity), but a blast search showed close similarity (0.5 % diversity) to other SIVmac239 isolates described previously (Kestler et al., 1990; Strickland et al., 2011). Upon further investigation, this sequence displayed a unique motif in the V2 region; characterized by 3 aa substitution relative to the reference sequence SIVmac239 (GenBank accession no. M33262) downloaded from the HIV databases (http://www.hiv.lanl.gov/content/index): M165I; K176E, N199D, hereafter referred to as the ‘I_E_D’ motif (Fig. 3b). To date these variants are unique to our macaques. However, the 176E residue has previously been described in two SIVmac239-infected rhesus macaques, in which the K176E mutation conferring macrophage tropism in vitro in alveolar macrophages arose independently (Kodama et al., 1993; Mori et al., 1992, 2000).

Fig. 3.

Representative sequences with distinctive gp120 amino acid signature from infected macaques. Sequences were aligned to the 1A11 (AC: M76764) and the SIVmac239 (AC: M33262) clone. Identical amino acids are indicated by a dot. Motifs detected by signature pattern analysis are highlighted with a box. Sequences labelled as SIVmac251 are from viral swarm used to infect the macaques. Sequences labelled as D01–D06 are from different infected macaques; sampling time is given in days p.i. (a) Representative SIVmac251 and plasma sequences belonging to clusters 9 and 14. UC, Unclassified sequences. (b) Representative sequences of the macrophage-tropic viral variant I_E_D detected in plasma and different post-mortem tissues from infected macaques.

All sequences from each of the six monkeys were then analysed to assess the presence of these three motifs. The percentages of the sequences from each time point and tissue that harboured the three motifs are given in Table 4. The R_AQN_S_V motif was found in brain macrophage tissue in four of six monkeys and the L_V motif was found in six of six monkeys. Both the R_AQN_S_V and the L_V motifs were observed primarily in the monocytes/macrophages-infiltrated tissues such as BAL and brain (Table 4). Normal BAL is ~50–75 % macrophages, 10–30 % lymphocytes and 5–15 % neutrophils. Moreover, in macaque D04 and D05, the motifs were also detected in sequences amplified from the lymph node (Table 4). Therefore, we cannot exclude that some of the sequences amplified from BAL tissues were actually infecting T-cells. On the other hand, there are no CD4⁺ T-lymphocytes present in the brain and, to avoid potential contamination from blood microvessels, the primates were exsanguinated prior to the dissections of brain tissues. Sequences were obtained from RNA and the only HIV productive infection in the brain that has conclusively been described occurs in macrophage subpopulations (Williams & Hickey, 2002). The I_E_D motif was found in the brain macrophages of three of six monkeys, where it was present at high frequency (29–100 %) in sequences derived from the brain at post-mortem of two macaques with meningitis or SIVE, and observed in the temporal lobe of one monkeys euthanized at 21 days p.i. (Table 4). The motif was never found in sequences from plasma, apart from the initial observation in monkey D05. It was also present in BAL macrophages at 61 days p.i. and post-mortem in macaque D06, and post-mortem in macaque D03 (Table 4).

Table 4. Frequency of viral sequences with distinctive gp120 motifs in plasma, lymphoid and non-lymphoid tissues from SIVmac251-infected macaques.

Macaque	Time (days p.i.)	Tissue*	R_AQN_S_V (%)	Motif†
				L_V (%)	I_E_D (%)
D01 Euthanized	21	Plasma	4	9
	21	Bone marrow
	21	BAL
	21	Meninges	11
	21	Frontal lobe	15
	21	Parietal lobe	7	7
	21	Temporal lobe	28		24
D02 Euthanized	21	Plasma	6	6
	21	Bone marrow	14	3
	21	Frontal lobe	52
	21	Parietal lobe		16
	21	Temporal lobe		14
D03 SIVE	21	Plasma	6
	61	Plasma	19	5
	61	BAL
	75	Plasma	10	5
	75	Bone marrow		7
	75	BAL			100
	75	Lymph node
	75	Meninges			100
	75	Frontal lobe	40
	75	Parietal lobe		53
	75	Temporal lobe		12
D04 SIVE	21	Plasma	10	5
	61	Plasma	5	5
	61	BAL	8	8
	91	Plasma		12
	91	Bone marrow		25
	91	Lymph node		6
	91	Frontal lobe		38
	91	Parietal lobe
	91	Temporal lobe
D05 SIVE	21	Plasma	7	4	4
	61	Plasma	17
	61	BAL		4
	89	Plasma	7	18
	95	Bone marrow	4	13
	95	Lymph node	10	10
	95	Frontal lobe			82
	95	Parietal lobe
	95	Temporal lobe		4	29
D06 Meningitis	21	Plasma	4
	61	Plasma	4
	61	BAL			20
	61	Lymph node
	118	Plasma
	118	Bone marrow
	118	BAL			85
	118	Lymph node
	118	Frontal lobe		11
	118	Parietal lobe	24	19
	118	Temporal lobe	9
Macaques			7	7	10
SIVmac251			3	2

^*Tissues in bold were derived from post-mortem brain samples.

^†The gp120 amino acid motifs detected by signature pattern analysis are the ones described in Fig. 2 (R_AQN_S_V and L_V are the motifs characteristic of clusters 9 and 14, respectively; I_E_D is the V2 motif characteristic of macrophage tropic strains).

SIV phylogenetic patterns within infected macaques

To investigate further the distribution of the three motifs, maximum-likelihood (ML) genealogies were inferred including all gp120 sequences generated for each macaque (Fig. 4). For the macaques euthanized at 21 days p.i. (D01 and D02), sequences derived from plasma and non-lymphoid tissues were interspersed in the tree. For the macaques with longitudinal sequences (D03–D06), a similar trend was observed, indicating multiple introductions of the virus into different tissues. On the other hand, sequences with a specific motif usually clustered within a monophyletic clade from each monkey. This pattern was most striking for sequences harbouring the I_E_D motif, which typically clustered together on a long branch mostly including tissues sampled later in infection (60–120 days p.i.). The sequences with the I_E_D motif displayed about 3 % divergence when compared to the other sequences from the same macaque. Such a divergence is expected to accumulate during 3–4 months evolution under a Poisson distributed mutational process with a rate of around 0.001 nt substitution per year. Additionally, the sequences with the I_E_D motif from the different macaques were not identical but displayed 0.1–0.9 % heterogeneity within each macaque and 0.2–6 % between the macaques. These sequences were detected in brain and/or BAL, suggesting a potential compartmentalization in macrophage-infiltrated tissues.

Fig. 4.

ML genealogies of gp120 sequences from infected macaques. Each tree includes all plasma and tissue sequences sampled from macaque D01 (a), D02 (b), D03 (c), D04 (d), D05 (e) or D06 (f), respectively. Branches are scaled in substitutions per site according to the scale bar (0.005) at the bottom left corner and coloured to represent the tissue of origin as follows: plasma and lymphoid 21 days p.i. (blue), plasma and lymphoid 61 days p.i. (green), plasma and lymphoid post-mortem (red), brain (cyan), BAL 61 days p.i. (dark brown), BAL post-mortem (light brown). Sequences containing the R_ AQN_S_V, L_V and I_E_D motifs are indicated by an oval, triangle and square, respectively. Branches with bootstrap support >75 % are denoted with an asterisk. Bootstrap values >50 % for the divergent branch in D05 and D06 are provided.

Discussion

Despite the advent of HAART, it is still unclear why a significant number of patients continue to develop HIV-associated neurocognitive disorders (HAND), with HAD being the most severe (Anthony et al., 2005; Bell, 2004). It is possible that genetic determinants for HAD exist, as suggested by studies that found different rates of HAD pathogenesis in similar populations infected with different HIV subtypes (Sacktor et al., 2009). The present work sought to determine how the diversity of the infecting viral swarm influences the initial seeding and subsequent progression of the infection.

A first-pass analysis of overall genetic similarity indicated similar nucleotide diversity in the inoculum as compared to the 21 days p.i. plasma sequences in each monkey. This initial result, suggested that no bottleneck occurred during the transmission and that sequences were transmitted randomly. However, a more detailed clustering analysis revealed a preferential transmission (or lack of transmission) of specific genetic variants. Three clusters representing cumulatively about one third of the sequences in the inoculum were found at significantly lower frequency than expected in the macaques, while two low frequency clusters were present at significantly higher frequency. No physical mucosal or immune barrier for infection could account for this observation at 21 days p.i., as all the CD8-depleted primates were intravenously infected with a heterogeneous SIVmac251 viral swarm. Although there is a rebound of the CD8-lymphocytes repertoire overtime, the adaptive immune pressure is absent in early transmission. Therefore, the result may suggest a differential fitness among the sequences in the SIVmac251 viral swarm in terms of transmission and/or replication dynamic in early infection. Furthermore, the motifs defining the clusters found in the macaques at higher frequency than expected stochastically were also detected in multiple tissues at increasing frequency over the course of infection, suggesting efficient replication and potentially higher fitness than other transmitted variants. Since all monkeys developed SIVE or meningitis, these analyses could reveal the potential establishment of neurovirulent strains. It is possible that the emergence of such motifs may be resulting from the absence of T-cell control in the CD8-depleted model and/or innate restriction factors (i.e. TRIM5α, Tetherin, APOBEC3g, etc.) may influence the outgrowth of particular cluster; however, their presence mostly in macrophage-infiltrated tissues (BAL and brain) shows a potential tropism for macrophages, which may substantiate the hypothesis of their link to neurovirulence. Moreover, the monophyletic clades observed in the phylogenetic trees showed the early establishment of compartmentalized viral strains in the brain, a well-known phenomenon in patients that develop HAD (Lane et al., 1995; Nickle et al., 2003; Salemi et al., 2005; Smit et al., 2001). Unfortunately, viral strains were not obtained immediately following CD8-depletion. However, in the CD8-depleted model the animals remain in an acute phase until they exhibit end-stage disease pathologies. The viraemia level remains high throughout the course of the disease. Although there is a drop in the viraemia at 21 days p.i., viraemia is still several logs higher than that observed in a non-depleted model and there are no statistical differences in the levels at the initial peak and 21 days p.i. As such, although it cannot be excluded that fitter variants may have been detected earlier, the 21 days p.i. time point provides a good overview of the early events, as proven by the fact that the sequences in the infected animals are still very similar to those observed in the viral swarm. Additional studies in a non-CD8-depleted model are currently under way to validate these observations in macaques with an unperturbed immune system at the time of infection.

Five of the macaques harboured strains containing a macrophage tropic V2 motif (Kodama et al., 1993; Mori et al., 1992, 2000) that was not detected in the infecting swarm. Two scenarios could explain the observed high frequency of this viral variant in the macaques and its absence in the inoculum: (i) the mutations leading to the amino acid motif emerged de novo in different macaques due to convergent evolution; (ii) they originated from a transmitted/founder variant present at extremely low frequency (<0.3 %) in the viral swarm, which then out-competed the other transmitted variants due to a higher fitness. Although the first possibility cannot be excluded, several observations are more consistent with the second explanation. First, except for the single sequence in the 21 days p.i. plasma from macaque D05, the I_E_D variant was found later in infection (>60 days p.i.), and at a very high frequency (20–100 %), in BAL macrophages from macaques with longitudinal sampling (i.e. macaques that were not euthanized at 21 days p.i.), suggesting that some time was needed for this variant to out-compete the first viral population that colonized the tissue. Unfortunately, BAL macrophages from D04 and D05 at the last time point were not available, so the observation could not be confirmed in these macaques. Second, the I_E_D variant appears as a divergent monophyletic clade in the genealogies, suggesting that it evolved only once in each macaque. If this motif conferred such an advantage that it evolved de novo in five of six macaques, we might expect that it also evolved along several independent trajectories within a single macaque as well. To test the functional significance of these potential neurovirulent motifs, it would be interesting to test in the future whether SIV clones harbouring one or a combination of the three motifs have indeed enhanced macrophage tropism or would cause faster development of neuroAIDS in the macaque model.

Keele et al. (2009) recently found that low-dose rectal inoculation of SIVmac251 of non-CD8⁻ depleted macaques led to productive infection by a minimum of 1–9 viruses randomly acquired from the swarm. Although the difference in transmission route (rectal versus intravenous) may account for the disagreement between the present study and Keele et al. (2009), the most likely explanation may be related to the CD8-depletion of in our rapid-disease model. The depletion of CD8 in macaques within the first 12 days after the infection has been shown to result in highly reproducible and rapid development of simian acquired immune deficiency syndrome (SAIDS) and elevated incidence of SIVE (Lifson et al., 2001; Schmitz et al., 1999), as well as a sustained higher plasma viral load during the asymptomatic period (Burdo et al., 2010). These factors are likely to increase the spectrum of viral variants that may potentially replicate in the new host. The lack of cytotoxic T-cell control could have fostered the replication/expansion of the macrophage-tropic and neurotropic variants (found at high frequency in the macaques’ BAL and brain tissues) that may otherwise be cleared during early infection by the intact immune system of non-depleted macaques. Further studies are currently under way using a non-CD8-depleted model to validate the importance of the replication/expansion of these viral variants. However, it is important to note that the data reported here are consistent with the ‘macrophage dysregulation’ model (Williams & Hickey, 2002), which proposes that the accumulation, activation and infection of macrophages infiltrating brain tissues can eventually lead to neuropathology initiated by the production of new viral variants that the immune system is unable to control (Lamers et al., 2011; Salemi et al., 2005; Williams & Hickey, 2002).

The present study demonstrated, for the first time, the efficient transmission and persistence in vivo of multiple low frequency SIVmac251 founder variants with specific gp120 motifs that productively replicated in both lymphoid and non-lymphoid tissues. Moreover, our study demonstrates that the infection of CD8-depleted macaques with a highly heterogeneous inoculum results in the outgrowth of low frequency variants and neuropathology. Thus, our model represents an ideal system in which to evaluate the presence of neurotropic viruses and characterize macrophage tropic signatures that may be linked to neuroAIDS.

Methods

Macaques, viral infection and CD8⁺ T-lymphocyte depletion.

Six rhesus macaques were intravenously inoculated with the SIVmac251 (1 ng SIV p27) viral swarm, 2006 stock (Strickland et al., 2011). The genetic characterization of this SIVmac251 viral swarm showed the presence of distinct phylogenetic lineages, with a mean nucleotide diversity of 0.6 % in the gp120 region. To achieve rapid-disease progression with high incidence of SIVE, macaques were CD8⁺-depleted by treatment with a human anti-CD8 antibody, cM-T807, administered subcutaneously (10 mg kg⁻¹) at 6 days p.i. and intravenously (5 mg kg⁻¹) at 8 and 12 days p.i. CD8-depletion was monitored by flow cytometry prior to antibody treatment and weekly thereafter. Two macaques (D01 and D02) were sacrificed at 21 days p.i. to investigate early genotypes in brain tissues, while the other four were monitored until terminal events (Table 1). Macaques were anaesthetized with ketamine-HCl, euthanized by an intravenous pentobarbital overdose and exsanguinated. All macaques were housed at Tulane National Primate Research Center in accordance with standards of the American Association for Accreditation of Laboratory Animal Care.

Sample collection and RNA extraction.

Blood, BAL, peripheral lymph nodes and bone marrow, were collected at 21 (macaques D01–D06) and 61 days p.i., and at the time of death (D03–D06). Plasma was obtained from EDTA-treated whole blood. Approximately 10 ml blood was collected in EDTA tubes. Peripheral blood mononuclear cells (PBMCs) were separated from EDTA anti-coagulated whole blood by density-gradient centrifugation with Ficoll-Paque Premium (GE Healthcare). Plasma was obtained from EDTA-treated whole blood by two rounds of centrifugation at high speed for 15 min. A peripheral lymph node was excised from the axillary or inguinal area of the macaque. A partial biopsy was performed to leave functional lymph node tissue behind. One millilitre of bone marrow was aspirated by using a Jamshidi needle and cells were frozen in aliquots of approximately 1×10⁷ until processing. Brain tissues were harvested post-mortem from each macaque following a previously described protocol (Chakrabarti et al., 1991; Davis et al., 1992; Johnson et al., 1988). Total RNA from monkey tissues (10⁵–10⁶ cells per tissue sample) was extracted using RNase-free glycogen as an RNA carrier. Snap frozen biopsies and necropsy specimens (~20 mg frozen tissue) were homogenized using a Power Gen 125 hand-held homogenizer with disposable tips. Optimal cutting temperature (OCT)-embedded frontal, temporal and parietal cortices were sectioned at 10 µM and stained with anti-SIVp28 (Microbix) to confirm productive infection in the area being sectioned. If positive, ten 10 µM sections were cut from the adjacent tissue. The samples were washed twice in ice-cold PBS to remove residual OCT medium, and the tissue was processed as above.

Viral sequencing.

Viral RNA was isolated from plasma (300 µl) using the QiaAMP Viral RNA mini kit (Qiagen) and from monkey tissues using TRIzol reagent (Invitrogen). The eluted RNA was used for at least two separate reverse transcription reactions using two different env gene-specific primers or random hexamers using SuperScript III RT Enzyme (Invitrogen), as described previously (Keele et al., 2009; Strickland et al., 2011). Multiple dilutions of RNA were used to correct for sampling bias. Multiple nested PCRs were performed on the cDNA to amplify the full-length gp120 sequences as described previously (Keele et al., 2009; Strickland et al., 2011). Amplified samples were cloned using the TOPO TA cloning kit (Invitrogen) with care taken to avoid sample contamination during the colony picking stage. Sequencing was performed at the University of Florida Interdisciplinary Center for Biotechnology Research genomics core facility. Sequences from the infecting SIVmac251 inoculum were obtained as described previously (Keele et al., 2009; Strickland et al., 2011). All sequences were assembled with the CodonCode software (Code Corporation) and manually adjusted to take into account the overall PCR and sequencing error rate. Sequences were aligned with the clustal (Thompson et al., 1997) algorithm implemented in BioEdit (Hall, 1999), followed by a manual optimization protocol taking into account conserved glycosylation motifs (Lamers et al., 1996). All alignments were gap-stripped for further analyses. The new sequences were deposited in GenBank with accession numbers JF764947–JF766081. The alignments are available upon request.

Assessing PCR and sequencing errors.

PCR and cloning were performed under the same conditions described in the previous section using the 1A11 clone (AIDS Reagent program #2736 derived from the SIVmac251 swarm) as a template. To calculate PCR error rate, 96 clones were sequenced, in both forward and reverse direction. Sequences were quality trimmed to 700 bp and aligned using BioEdit. Using the 1A11 sequence as a reference, pair-wise analysis was performed to compute the number of nucleotide differences using mega 4.0, resulting in an overall rate of 1.2 errors per kb (1 %). To calculate the sequencing error rate, a single clone was sequenced directly 96 times and only one error was observed. All errors were due to point mutations at singleton positions and no recombinant sequences, which could result by Taq polymerase-induced template switching, were detected. Therefore, to correct for the overall PCR error rate, singletons (which, as expected, represented <1 % of the observed point mutations and appeared to be randomly distributed) were removed from the final alignments.

Cluster analysis.

The gap-stripped viral swarm sequences from the inoculum (SIVmac251, N = 193) were classified into discrete clusters using the complete linkage algorithm implemented in the program Clusterer (Klepac-Ceraj et al., 2006), where a cluster is defined as a group of sequences with ≥99.5 % nucleotide identity. The method is an unsupervised clustering algorithm where distances between clusters are determined by the greatest Hamming distance between any two sequences in different clusters.

Specific amino acid motifs defining the 17 clusters detected in the SIVmac251 inoculum were determined by signature pattern analysis using a 90 % threshold in the VESPA program (http://www.hiv.lanl.gov/content/sequence/VESPA/vespa.html). The remaining SIVmac251 strains not present in the cluster of interest were used as background.

Each of the sequences in the 21 days p.i. plasma samples from the infected macaques was then compared with the clusters in the SIVmac251 swarm. A plasma sequence was considered genetically related to a given cluster if it showed ≤6 nt differences when compared to each of one of the sequences of that specific cluster. The cut-off was chosen to take into account the potential accumulation of new mutations during the first 3 weeks of intra-host viral replication. Previous studies in experimentally infected macaques estimated an upper limit of 0.05 aa substitutions per site per year for the evolutionary rate of SIV gp120 (Burns & Desrosiers, 1991; Johnson et al., 1991). Therefore, under a Poisson process, the expected number of amino acid substitutions between the ancestral sequence in the swarm and the one sampled from the infected macaque t ( = 21) days after the infection is μt*L (μ = evolutionary rate; L = length of gene region = 433 aa), with variance μt*L, which results in 3±3 expected changes with 95 % confidence interval. Some sequences from plasma were equally distant to two clusters. To take into account the uncertainty of the founder sequence, such sequences were fractionally assigned to both clusters, weighted by the number of sequences found in the inoculum for that cluster. The Fisher’s exact test was used to evaluate the differences in the proportions between specific sequence clusters in the primates and in the viral swarm, by analysing either the unambiguously assigned sequences (‘unweighted’ analysis), or all sequences including the ones fractionally assigned to multiple clusters (‘weighted’ analysis).

Sequence diversity and phylogenetic analysis.

Genetic diversity in each dataset was calculated from pair-wise distance measures using uncorrected P-distances and 500 bootstrap replicates to estimate standard errors. Neighbour-joining phylogenies were inferred using the ML composite correction model (equivalent to the GTR model) and gamma-distributed rate variation (+G) across sites. Statistical support for branches was assessed by 1000 bootstrap replicates. Calculations were carried out with mega 4.0 (Tamura et al., 2007). ML calculations were performed with the online version of PHYML v3.0 (http://www.atgc-montpellier.fr/phyml/) (Guindon et al., 2009). The ML trees were inferred using the GTR+G model. Tree searching was performed using SPR and NNI with five random starting trees. Support for internal branches was assessed using the approximate likelihood-ratio test with 200 replicates (Guindon et al., 2009).