Collateral Damage: What Effect Does Anti-CD4 and Anti-CD8α Antibody–Mediated Depletion Have on Leukocyte Populations?

Since 1986, injection of anti-CD4 or anti-CD8 Abs into mice has been a widely used technique to deplete CD4⁺ or CD8⁺ T cell subsets, respectively, in vivo (1–3). However, as the process of depletion relies on Ag–Ab interactions, concerns have been raised about collateral effects on non–T cells with surface expression of CD4 or CD8. The principal cell populations of concern are dendritic cells (DCs), with specific DC subsets expressing varying levels of CD4 or CD8α. DCs are key APCs (reviewed in Ref. 4), and collateral depletion of these cells could have far-reaching consequences for a host’s immune response, especially during viral infection. CD11b⁺ DCs express CD4, whereas CD8α⁺ DCs are identified by the presence of CD8α (5, 6). In addition, plasmacytoid DCs (pDCs) express a high level of CD4 and a varying level of CD8α (5, 6). Accordingly, significant reductions in the number of CD4⁺ DCs or CD8α⁺ DCs are reported after anti-CD4 or anti-CD8α Ab injections, respectively (7–9). To reduce the impact on DCs, Abs against CD8β, instead of the commonly used target CD8α, have been used (10, 11), as the CD8 surface complex differs between DCs (CD8α/CD8α homodimer) and T cells (CD8α/CD8β heterodimer) (12). However, certain subsets of CD8⁺ T cells, such as intraepithelial lymphocytes, express only CD8α and not CD8β on their surface and are thus not depleted by anti-CD8β Ab administration (13, 14). There is no equivalent strategy for specific CD4⁺ T cell depletion as cells express CD4 as monomers or disulfide-linked homodimers (15, 16).

Importantly, the precise effects of anti-CD4 or anti-CD8 Ab injections on non–T cell populations are not well characterized and, in many studies, not assessed or considered, despite their potential effect on experimental outcomes. In this study, we analyzed the effects of anti-CD4 or anti-CD8α Ab injections on several leukocyte populations in the spleen and blood of naive and virus-infected mice.

Scientific literature was reviewed by searching for titles and abstracts using the following search strings: “(CD4 or L3T4) and (deplet* or ablat* or remov*)” or “(CD8 or Lyt-2) and (deplet* or ablat* or remov*)” in PubMed with entries up until June 1, 2018, being considered in this study.

Female C57BL/6 mice aged between 8 and 12 wk were obtained from Australian BioResources and housed in specific pathogen-free conditions in the animal facility of the University of Sydney. Ethical approval for the use of all mice was obtained from the University of Sydney Animal Care and Ethics Committee (AEC 1056/16). The lymphocytic choriomeningitis virus (LCMV) Armstrong (ARM) 53b stock was obtained originally from a triple plaque-purified clone that was subsequently passaged twice in BHK cells in Michael Oldstone’s (17) laboratory and kindly provided by I. L. Campbell (18). For virus inoculation, mice were given i.p. injection of 500 PFU (low-dose) or 200,000 PFU (high-dose) LCMV-ARM 53b in 200 μl of PBS plus 2.5% FBS. Mock-infected mice received the same volume of PBS plus 2.5% FBS without the virus. The anti-CD4 (clone GK1.5; ATCC TIB-207) and anti-CD8α Abs (clone 2.43; ATCC TIB-210) were purchased from Bio X Cell (West Lebanon, NH) or produced in house from hybridoma supernatant as described previously (18) and quantified by ELISA according to the manufacturer’s instructions (Invitrogen). Two different strategies were used to achieve sufficient depletion in mice infected with low-dose or high-dose LCMV: mice infected with low-dose LCMV and corresponding mock-infected control mice were injected i.p. with 10 μg in-house made anti-CD4 or anti-CD8α Abs in 100 μl of PBS on the day of infection and on days 2, 4, and 6 postinfection. By contrast, mice infected with high-dose LCMV and corresponding mock-infected control mice were injected i.p. with 200 μg of anti-CD4 or anti-CD8α Abs on the day of infection and on days 3 and 6 postinfection. Although the first is sufficient to achieve >95% reduction of CD4⁺ or CD8α⁺ leukocytes in peripheral blood from uninfected mice, the latter is more commonly used (e.g., Refs. 19–22). Control mice received an equal volume of PBS without Ab or 200 μg anti–keyhole limpet hemocyanin (clone LTF-2; Bio X Cell; isotype control), respectively.

On day 7 postinfection, mice were deeply anesthetized by halothane inhalation and euthanized by exsanguination via cardiac puncture. In the case where liver leukocytes were isolated, the mice were perfused with sterile PBS. A one-tenth volume of 0.5 M EDTA was added to collected blood to prevent coagulation, and RBCs were lysed using buffered ammonium chloride solution (170 mM NH₄Cl, 20 mM HEPES [pH 7.4]). Single-cell suspensions of blood leukocytes, splenocytes, and liver leukocytes were stained for cell surface markers using specific fluorophore-conjugated Abs optimized for flow cytometry. The reagents and Abs used in this study were as follows: LIVE/DEAD Fixable Blue Dead Cell Stain (Thermo Fisher Scientific), anti-CD11b-BUV395 (clone M1/70; BD Biosciences), anti-B220-BUV737 (clone RA3-6B2; BD Biosciences), anti-MHCII(I-A/I-E)-BV510 (clone M5/114.15.2; BioLegend), anti-Ly6C-BV605 (clone HK1.4; BioLegend), anti-Ly6G-BV650 (clone 1A8; BioLegend), anti-CD8α-FITC (clone CT-CD8α; Thermo Fisher Scientific), anti-Siglec-H-PE (clone 551; BioLegend), anti-CD3ε-PE/Cy5.5 (clone 145-2C11; Thermo Fisher Scientific), anti-NK1.1-PE/Cy5 (clone PK136; BioLegend), anti-CD8β-PE/Cy7 (clone H35-17.2; Thermo Fisher Scientific), anti-CD4-allophycocyanin (clone RM4-4; BioLegend), anti-CD8α-allophycocyanin (clone 53-6.7; BioLegend; for in vitro blocking experiments), anti-CD45-Alexa Fluor 700 (clone 30-F11; BioLegend), and anti-CD11c-allophycocyanin/Cy7 (clone N418; BioLegend). For in vitro blocking experiments, two million splenocytes were incubated with serially diluted anti-CD8α Ab (clone 2.43) in 200 μl of PBS for 20 min at 4°C in the dark prior to staining with fluorophore-labeled Abs. Stained and fixed cells were analyzed with a Becton Dickinson custom 10-laser LSR II flow cytometer and FlowJo software (v.10.4.1). The staining index was calculated using median fluorescence intensity (MFI) as follows: staining index = (MFI_{positive population} − MFI_{negative population})/SD_{negative population}.

As a dimensionality reduction technique for multicolor flow cytometry data, Barnes–Hut-implemented t-distributed stochastic neighbor embedding (t-SNE) was used (23, 24). Following color compensation, leukocytes were gated on live CD45⁺ cells, followed by subgating for CD4⁺ and/or CD8α⁺ cells, of which 30,000 representative events were randomly sampled using FlowJo. The t-SNE plugin in FlowJo was used to calculate t-SNE parameters. Subsequently, data were exported as comma-separated values and visualized using a script in RStudio (v1.1.383) (25).

To survey the usage of anti-CD4 or anti-CD8 Ab injection regimens for depleting specific leukocyte subsets, we performed a literature search on scientific publications in PubMed. A total of 13,080 and 9335 entries were returned on searches for “(CD4 or L3T4) and (deplet* or ablat* or remov*)” or “(CD8 or Lyt-2) and (deplet* or ablat* or remov*),” respectively (Fig. 1A). Although PubMed allows for species restriction using medical subject headings keyword searches, species restriction to Mus musculus resulted in removal of true-positive hits, and hence the search was not limited to murine models. Instead, full-text analysis was performed to remove false positives and to identify experimental details. For feasibility, the detailed full-text analysis was limited to articles published between January 1, 2016, and June 1, 2018 (Fig. 1B). Of the 1149 manuscripts that were found on CD4 depletion, 1138 manuscripts had English full texts available. Two hundred and nine publications used an in vivo approach to deplete CD4⁺ cells in a mouse model, of which 145 used anti-CD4 Ab injection regimens. Of these, 137 manuscripts attempted to deplete CD4⁺ T cells only, whereas just seven manuscripts acknowledged that any CD4⁺ cells could have been affected by this methodology (26–32). More than 80% of all published work performing anti-CD4 Ab–mediated depletion method used GK1.5. Similarly, of the 809 manuscripts found on CD8 depletion, 801 manuscripts had English full texts available, and 185 papers performed anti-CD8 Ab–mediated depletion of CD8⁺ cells in vivo. Six articles stated that any CD8-expressing cells could have been affected (27, 29, 30, 33–35), whereas 178 papers did not comment on cells other than T cells. Diverse clones of anti-CD8α and anti-CD8β Abs were used, with anti-CD8α Ab clones 2.43 and 53-6.7 being the most common choices (Fig. 1B). These findings suggested that although there is robust evidence for CD4 and/or CD8 surface expression on non–T cell populations, there is widespread uncertainty as to whether they are also affected by Ab injection regimens.

It was previously reported that anti-CD4 Ab clone RM4-4 is a suitable detection Ab postdepletion with anti-CD4 Ab clone GK1.5, being noncompetitive in its binding capacity (36). In the case of anti-CD8α Ab–mediated depletion of CD8α⁺CD8β⁺ T cells, Abs targeting CD8β have previously been used for detection (37, 38). However, as CD8α⁺CD8β⁻ T cells and CD8α⁺ DCs do not have surface expression of CD8β, it was necessary to identify an anti-CD8α Ab clone that does not compete with the commonly used anti-CD8α depleting Ab clone 2.43.

We examined the binding capacity of anti-CD8α Ab clones 53-6.7 and CT-CD8α on splenocytes preincubated with anti-CD8α Ab clone 2.43. The anti-CD8β Ab clone H35-17.2 was included as a control, as it binds to a different Ag from the blocking Ab, and hence no competition was expected in CD8α⁺CD8β⁺ T cells. The binding capacity of each clone was assessed by calculating the staining index, an estimate of the separation between the positive and the negative populations, and the proportion of cells expressing the marker. Clone 53-6.7 showed a rapid decline in both the staining index and the percentage of positive cells to zero as the amount of clone 2.43 used in preincubation increased (Fig. 2A, 2B). Although staining with CT-CD8α showed a decline in the staining index as the amount of 2.43 Ab increased, it remained well above the basal level even after preincubation with the maximal amount of 2.43 Ab. There was also a decline in the percentage of positive cells as the amount of blocking Ab increased. However, this decline in the percentage of positive cells with increasing amount of blocking Ab was also observed to a similar degree in anti-CD8β Ab clone H35-17.2–stained cells (Fig. 2A, 2B). These results suggest that anti-CD8α Ab clone CT-CD8α, but not 53-6.7, is a suitable, noncompeting detection Ab postdepletion with clone 2.43.

To determine the effects of repeated anti-CD4 or anti-CD8α Ab injections on multiple leukocyte populations, we analyzed splenocytes and blood leukocytes from naive and low-dose (500 PFU) LCMV–infected mice. LCMV infection was chosen as it is commonly used to study immune responses, and Ab-mediated cell depletion strategies have regularly been used in this model [e.g., (39, 40)]. Our results had shown that although the anti-CD8α detection Ab clone, CT-CD8α, is suitable to detect CD8α⁺ cells after depletion, this clone nevertheless showed a significant decline in the staining index (Fig. 2A, 2B). To account for this, we used the fluorescence minus one controls as the negative cutoffs to include the CD4^lo and CD8α^lo events when gating for CD4⁺ and/or CD8α⁺ cells.

To examine the different leukocyte populations affected by the Ab injection regimens, t-SNE was performed as an unbiased categorization and visualization method (Figs. 3, 4). t-SNE analysis on mock-infected splenocytes pregated for live CD45⁺, CD4⁺, and/or CD8α⁺ revealed that most leukocytes with surface expression of CD4 and/or CD8α were also CD3ε⁺ (CD4^hi or CD8α^hi) or B220⁺ (CD4^lo or CD8α^lo) (Fig. 3B). As depicted by the insets, pDCs (Siglec-H⁺), CD11b⁺ DCs (CD11c⁺ CD11b⁺), and CD8α⁺ DCs (CD11c⁺ CD8α⁺) were also readily identified. When the effects of low-dose Ab injection regimens and low-dose LCMV infection were examined, we found that Ab-mediated depletion was highly effective and specific, with the most recognizable change being the near-complete disappearance of CD3ε⁺ splenocytes with surface expression of CD4 in both mock- and LCMV-infected mice treated with anti-CD4 Ab (Fig. 3C). Similarly, there was a widespread loss of CD3ε⁺ cells with surface expression of CD8α when mice were treated with anti-CD8α Ab (Fig. 3C). By contrast, B220⁺ cells (CD4^lo or CD8α^lo) were unaffected by Ab injection (Fig. 3C). There were variable effects on the DC populations, with CD11b⁺ DCs being largely unaffected following treatment with either anti-CD4 or anti-CD8α Ab, whereas CD8α⁺ DCs were greatly reduced following anti-CD8α Ab injection regimen in both mock- and LCMV-infected mice (Fig. 3C; arrowheads). Although there was a reduction of CD4^hi subset of pDCs following anti-CD4 Ab injection regimen, CD4^lo pDCs were retained (Fig. 3C; arrowheads). However, t-SNE analysis could not determine whether the disappearance of CD4^hi pDCs was because of specific ablation by anti-CD4 Ab injections or the reduction in CD4 fluorescence intensity mediated by interactions between the depletion and detection Abs. Near-identical observations were made with t-SNE analysis of blood samples (data not shown).

A recent report using anti-CD3ε Abs had shown off-target effects on cells not expressing the CD3ε Ag (41). To assess any off-target effects on CD4⁻CD8α⁻ cells, we employed a manual gating strategy to determine the changes in the numbers of CD4⁻CD8α⁻ T cells, B cells, NK cells, neutrophils, and Ly6C^hi monocytes (Fig. 3A). We found no significant changes in the numbers of these cells following Ab injection regimens (data not shown). Taken together, these findings suggest that surface expression of CD4 and/or CD8α is a necessary requirement for targeted depletion by the corresponding Ab injection regimens. Furthermore, these findings show that the ablative efficacy of Ab injection regimens differs between specific leukocyte subsets.

To investigate Ab and virus dose-dependent effects on leukocyte depletion efficacy, we performed t-SNE analysis on splenocytes, blood leukocytes, and liver leukocytes from naive and high-dose (200,000 PFU) LCMV–infected mice that were treated with high-dose isotype anti-CD4 or anti-CD8α Abs (200 μg) (Fig. 4). As above (Fig. 3B), t-SNE analysis on leukocytes pregated for live CD45⁺ CD4⁺ and/or CD8α⁺ identified CD3ε⁺ cells (CD4^hi or CD8α^hi), B220⁺ cells (CD4^lo or CD8α^lo), CD11b⁺ DCs (CD11c⁺ CD11b⁺), CD8α⁺ DCs (CD11c⁺ CD8α⁺), and pDCs (Siglec-H⁺) (Fig. 4A). Comparable to low-dose LCMV infection and low-dose Ab injection regimens (Fig. 3C), mice that were given high-dose anti-CD4 or anti-CD8α Ab injections had a near-complete depletion of CD3ε⁺ T cells (CD4^hi or CD8α^hi) but not B220⁺ cells (CD4^lo or CD8α^lo) (Fig. 4B). CD11b⁺ DCs were largely unaffected following high-dose anti-CD4 or anti-CD8α Ab injection regimens, whereas CD4^hi pDCs were greatly reduced following high-dose anti-CD4 Ab injection in both mock- and LCMV-infected mice (Fig. 4B). Moreover, CD8α⁺ DCs were largely ablated following high-dose anti-CD8α Ab injections (Fig. 4B). Near-identical observations were made with t-SNE analysis of blood and liver samples from the same cohort of mice (data not shown). Moreover, CD4⁻CD8α⁻ T cells, B cells, NK cells, neutrophils, and Ly6C^hi monocytes were not ablated following high-dose Ab injections (data not shown).

Our t-SNE analysis had shown that CD4⁺ and/or CD8α⁺ leukocytes were divided into three subsets, T cells (CD3ε⁺), B220⁺ cells, and DC subsets, and that the ablative efficacy of Ab injection regimens varied between these groups. To quantitatively determine the ablative effects of low-dose or high-dose anti-CD4 or anti-CD8α Ab injections on these cell populations separately, we performed manual gating and quantitation. A reduction in number equivalent to more than 95% of CD4⁺ T cells or CD8α⁺CD8β⁺ T cells was observed, independent of low-dose LCMV infection in both the spleen and blood of mice treated with low doses of anti-CD4 or anti-CD8α Abs, respectively (Fig. 5A–E). CD8α⁺CD8β⁻ T cells also showed a significant reduction, equaling ∼75% in both the spleen and blood of mock- and LCMV-infected mice treated with anti-CD8α Ab (Fig. 5A–E). Furthermore, Ab treatments also reduced the number of CD4⁺CD8α⁺ double-positive T cells in the spleen, with anti-CD4 treatment having greater effect (>97% reduction) than anti-CD8α treatment (80–85% reduction) in both mock- and LCMV-infected mice (Fig. 5D). Although double-positive T cells were greatly reduced in the blood of mock-infected, anti-CD4-treated mice when compared with the control (∼90% reduction), anti-CD8α-treated mice did not show a significant reduction in the number of these cells (Fig. 5E). Corroborating the findings in t-SNE analysis (Figs. 3, 4), B220⁺ cells (CD4^lo or CD8α^lo) were not changed in number (data not shown).

CD11b⁺ DCs were not ablated by anti-CD4 Ab treatment in either the spleen or blood of mock- and low-dose LCMV–infected mice (Fig. 5A–E). By contrast, anti-CD8α Ab treatment significantly reduced (75–90%) numbers of CD8α⁺ DCs in the spleen, independent of LCMV infection. Although a similar reduction of CD8α⁺ DCs was observed in the blood of uninfected mice, there was no statistically significant reduction in the blood of LCMV-infected mice (Fig. 5E). There was also a significant reduction of CD11b⁺ DCs in the spleen of anti-CD8α-treated, LCMV-infected mice (40–50%) but not mock-infected mice, despite the absence of the targeted Ag, CD8α (Fig. 5D). Contrary to the observations made with t-SNE analysis (Fig. 3C), pDCs (CD4⁺ CD8α^var) showed no ablation in either the spleen or blood of mock- and LCMV-infected mice after being treated with low-dose anti-CD4 or anti-CD8α Abs, suggesting that the apparent disappearance of CD4^hi pDCs in t-SNE analysis was due to decreased CD4 fluorescence intensity from CD4^hi to CD4^lo, mediated by the interactions between the depletion and detection Abs (Fig. 5C–E).

Furthermore, there was a reduced efficacy of Ab-mediated depletion of cells in the blood of low-dose LCMV–infected mice (Fig. 5E) compared with the spleen of infected mice (Fig. 5D). Most notably, 98% of CD4⁺CD8α⁺ double-positive T cells were depleted from the blood of mock-infected mice after anti-CD4 Ab injections, but only 50% were ablated in mice infected with LCMV (Fig. 5E). CD8α⁺ DCs were depleted by 77% in mock-infected mice but only by 50% in LCMV-infected mice after anti-CD8α Ab treatment (Fig. 5E).

We confirmed these findings in mice that were either mock- or high-dose LCMV–infected and treated with high-dose Ab injection regimens. High-dose LCMV infection mediated more widespread changes in the number of leukocyte subsets, including a significant expansion of CD8α⁺CD8β⁺ T cells when compared with low-dose infection (Fig. 5D, 5E, 5I–K), indicating more robust immunological responses. Near-identical observations were made regarding the specific ablative effects of anti-CD4 or anti-CD8α Ab injection regimens on leukocytes from the spleen and blood, with a few exceptions (Fig. 5F–J). Notably, high-dose anti-CD8α Ab injections resulted in a near-complete depletion of CD8α⁺ DCs (>96%) in the spleen of mock- and high-dose LCMV–infected mice. Furthermore, contrary to low-dose Ab injection regimens, injections of high-dose anti-CD4 or anti-CD8α Abs resulted in a significant reduction in the number of pDCs in the spleen (∼50%) but not the blood. In the liver, CD4⁺ T cells were significantly ablated following an anti-CD4 Ab injection regimen in both mock-infected (∼71%) and LCMV-infected mice (∼98%) (Fig. 5K). CD8α⁺CD8β⁺ T cells had a near-complete depletion (∼99% in both mock- and LCMV-infected mice) following an anti-CD8α Ab injection regimen, and CD8α⁺CD8β⁻ T cells and CD8α⁺ DCs were also significantly depleted in both mock- and high-dose LCMV–infected mice (Fig. 5K). The numbers of CD11b⁺ DCs and pDCs were changed following Ab injections only in LCMV-infected mice (Fig. 5K).

Next, we assessed the relative surface expression levels of CD4 and/or CD8α on T cells and DCs. In both the spleen and blood, CD4⁺CD8α⁺ T cells had a relatively similar level of CD4 as CD4⁺ T cells, whereas their CD8α level was significantly lower than those of CD8α⁺CD8β⁺ T cells. Despite having detectable CD4 expression, CD11b⁺ DCs and pDCs had significantly lower CD4 surface expression when compared with CD4⁺ T cells in the spleen, blood, and liver (Fig. 6A, 6B). By contrast, CD8α⁺ DCs in the spleen but not blood or liver had a similar level of CD8α when compared with CD8α⁺CD8β⁺ T cells, whereas the MFI of CD8α in pDCs was less than one-tenth that of CD8α⁺CD8β⁺ T cells in the spleen, blood, and liver (Fig. 6A, 6B). No significant differences were observed in the levels of CD4 or CD8α when comparing leukocytes from the spleen and blood of mock-versus low-dose LCMV–infected mice, with the exception of a significant decline in the surface level of CD8α on CD8α⁺CD8β⁺ T cells in the blood (Fig. 6A). By contrast, the surface expression level of CD4 was significantly decreased in CD4⁺ splenocytes (CD4⁺ T cells, CD4⁺CD8α⁺ T cells, and pDCs), blood leukocytes (CD4⁺CD8α⁺ T cells and pDCs), and liver leukocytes (pDCs) in mice infected with a high dose of LCMV (Fig. 6B). The same was true for the surface expression of CD8α, which was significantly lower on CD8α⁺CD8β⁺ T cells in the spleen, blood, and liver of high-dose LCMV–infected mice compared with mock mice. CD8α was also significantly lower on CD8α⁺ DCs in the spleen and liver, but not blood, of high-dose LCMV–infected mice compared with mock-infected mice (Fig. 6B). In summary, these findings suggest that CD4 and/or CD8α surface expression is not sufficient for targeted depletion by corresponding Ab injection regimens.

In this study, we demonstrated that the administration of anti-CD4 or anti-CD8α Abs depletes cognate T cell and DC subsets. However, not all cell populations expressing the target Ags were affected to the same degree. Importantly, only leukocytes with surface expression of CD4 and/or CD8α were depleted, indicating minimal off-target effects.

After anti-CD4 Ab injections, CD4⁺ T cells and CD4⁺CD8α⁺ double-positive T cells were significantly ablated. However, CD11b⁺ DCs were not affected by both low- and high-dose Ab injection regimens, despite expressing CD4. This finding differs from a previously published report that showed significant reduction of CD4⁺CD8α⁻ DCs following anti-CD4 Ab injection (8). This discrepancy may be the result of differences in the Ab injection regimens. Martín et al. (8) injected 300 μg of anti-CD4 clone GK1.5 daily for three consecutive days, whereas we used an injection regimen of 10 μg every other day or 200 μg every third day. Although both doses of Ab injections were sufficient to deplete >95% of all CD4⁺ leukocytes in the spleen, peripheral blood, and liver, our results indicate that depletion rates differ between specific subsets. The reduced susceptibility of DCs compared with T cells may be due to the significantly lower level of CD4 on these cells. Notably, using t-SNE analysis, we observed that the CD4^hi, but not all of the CD4^lo, subset of pDCs appeared to be reduced following low-dose anti-CD4 Ab treatment. By contrast, manual gating showed no effect on the overall number of pDCs. The apparent depletion of pDCs seen in t-SNE analysis is thus likely a reduction in CD4 fluorescence intensity from CD4^hi to CD4^lo, mediated by interactions between the depletion and detection Abs. However, in contrast to low-dose Ab injections, high-dose anti-CD4 and anti-CD8α Ab injections resulted in a significant reduction in the number of pDCs in the spleen but not blood or liver, suggesting a dose-dependent increase in collateral ablative effects on non–T cell populations.

In contrast to anti-CD4 Ab injections, anti-CD8α Ab injection regimens ablated CD8α⁺ T cells and DCs in the spleen, blood, and liver but not CD4⁺CD8α⁺ double-positive T cells in the blood of either LCMV-infected or mock-infected mice. The collateral effects of anti-CD8α Ab injections on splenic CD8α⁺ DCs have been reported previously by Zhan et al. (9) and are possibly due to the equally high surface expression level of CD8α on DC and CD8α⁺ T cell subsets in the spleen. The comparatively higher level of CD4 versus CD8α on CD4⁺CD8α⁺ double-positive T cells may also explain the greater efficacy of the anti-CD4 Ab than the anti-CD8α Ab in depleting these cells. However, despite having significantly lower surface expression of CD8α than CD8α⁺CD8β⁺ T cells, CD8α⁺CD8β⁻ T cells in the spleen, blood, and liver and CD8α⁺ DCs in the blood and liver were successfully depleted. Furthermore, the similar level of CD8α on CD8α⁺ DCs from the blood of mock- versus low-dose and high-dose LCMV–infected mice indicates that the reduced depletion efficacy is not due to intrinsic reduction of the surface Ag level or increased numbers of immature CD8α^lo DCs emigrating from the bone marrow in response to infection. Interestingly, we observed an overall reduction in the efficacy of low-dose anti-CD4 and anti-CD8α Ab injection regimens in the blood of LCMV-infected mice, which was not seen in mice treated with high-dose Abs. The finding in the low-dose depleted animals is consistent with a previous report that increased levels of immune complexes of viral Ags and host Abs in circulation impair FcγR-dependent Ab-mediated leukocyte depletion in the peripheral blood of mice persistently infected with LCMV (42, 43) and suggests a general inhibitory effect of Ab-mediated depletion during infection. Importantly, the absence of this effect in high-dose treated mice emphasized the necessity to determine the effective dosage for each experimental condition. Further, it has been proposed that collateral effects on DC subsets could be avoided by using Abs targeting anti-CD8β instead of anti-CD8α to specifically ablate CD8α⁺CD8β⁺ T cells (10, 11). However, this strategy would not ablate the CD8α⁺CD8β⁻ T cells, which are mostly present in intestinal epithelium (44) and hence may not be suitable for studying intraepithelial immune interactions.

We observed no depletion of leukocytes that did not express the target Ags following low- or high-dose Ab injection regimens. By contrast, a recent study looking at the effects of anti-CD3ε Ab injections reported off-target effects on non–CD3ε-expressing cells (41). These effects correlated with the dose of depleting Ab used. It is unclear if a similar effect would occur at an even higher dose; nevertheless, off-target effects should be assessed for each Ab injection regimen anew.

Furthermore, our findings demonstrate the need for using detection Abs that do not compete with depletion Abs to assess treatment efficacy accurately. This is also evident from Zhan et al. (9), who used two different gating strategies in parallel to assess depletion of CD8α⁺ DCs; the first was based on the presence of CD8α, whereas the second approach gated for CD205⁺CD11b⁻ DCs based on the observation that CD8α⁺ DCs are positive for CD205 and negative for CD11b (45). The first gating strategy indicated a near-complete depletion, whereas the second strategy showed a reduction of ∼70% in the number of CD205⁺CD11b⁻ DCs (9). The difference between the two gating strategies is likely due to competitive binding between the depletion and detection Abs in the first approach and further demonstrates the importance of using appropriate detection strategies when assessing Ab-mediated depletion.

In summary, our results demonstrate that both anti-CD8α and anti-CD4 Ab injection regimens have no off-target effects on leukocytes that lack the cognate surface Ags. However, anti-CD4 and anti-CD8α Ab injections have dose-dependent collateral effects on Ag-expressing, non–T cell populations. Furthermore, efficacy and collateral effects of Ab depletion varied between uninfected and infected mice, emphasizing the need for multivariate assessment of treatment effects in all experimental groups.

We thank Iain L. Campbell for LCMV ARM 53b, Brendon Davis for technical assistance in generating neutralizing antibodies, and Claire L. Thompson for editorial help with the manuscript. We thank the Sydney Cytometry facility for assistance with flow cytometry and data analysis. We thank Monica Cooper at the University of Sydney Library for assistance with literature review.

The authors have no financial conflicts of interest.