HDAC7 is an actionable driver of therapeutic antibody resistance by macrophages from CLL patients
Abstract
Resistance, to therapeutic antibodies used to treat chronic lymphocytic leukemia (CLL) patients is common. Monocyte- derived macrophages (MDMs) are a major effector of antitumour responses to therapeutic antibodies and we have previously reported that resistance to therapeutic antibodies, by MDMs, increases as CLL disease progresses. In this study, we examine the effect of a Class IIa-selective HDAC inhibitor (TMP195) on the phagocytic response to opsonised tumor cells or non- opsonised targets by MDMs derived from CLL patients. We report that TMP195 enhances phagocytic responses to antibody-opsonised CLL cells and E. coli within 30 min of treatment. The enhanced response is phenocopied by knockdown of the Class IIa HDAC, HDAC7, or by low concentrations of the pan-HDAC inhibitor, vorinostat. HDAC7 knockdown and inhibition induces hyperacetylation and hyperphosphorylation of Bruton’s tyrosine kinase (BTK). Moreover, BTK inhibitors abrogated the enhanced response to HDAC7 inhibition. Our data show that HDAC7 is an actionable driver of resistance to therapeutic antibodies by MDMs derived from CLL patients.

Introduction

Chronic lymphocytic leukemia (CLL) is a common adult B-cell leukemia [1]. Despite the introduction of promising new targeted therapies to the clinic, CLL remains incurable [1]. CLL is characterized by complex genomic changes that are accompanied by changes in the transcriptome, epigen- ome and B-cell receptor function. These changes drive aberrant B-cell differentiation, maturation, apoptosis, and immune function (Reviewed in [1]). In addition to the leukemic-cell changes, CLL is an exemplar of the infl uence
of the tissue microenvironment as a driver of tumor beha- vior. For example, myeloid cells infl uence CLL leukemic- cell survival, progression, and treatment sensitivity [2–5]. In particular, depletion of macrophages in murine models of CLL reduced tumor growth [4, 5]. Thus, targeted depletion of tumor associated macrophages may be an effective way to reduce tumor burden and induce chemo- sensitivity. However, macrophage depletion strategies that show promise in mouse models have been disappointing in patients [6].
Paradoxically, preserving macrophages and harnessing their immune functions may be advantageous in CLL dis- ease control. For example, macrophages have been shown

Supplementary information The online version of this article (https://
doi.org/10.1038/s41388-020-01394-w) contains supplementary material, which is available to authorized users.

* N. A. Saunders [email protected]

1University of Queensland, Woolloongabba, QLD, Australia
2Department of Haematology, Cancer Services Unit, Princess Alexandra Hospital, Woolloongabba, QLD, Australia
3School of Medicine, Translational Research Institute, University of Queensland, Woolloongabba, QLD, Australia
4University of Queensland Diamantina Institute, Translational Research Institute, Woolloongabba, QLD, Australia
to (a) modify T cell mediated antitumor responses [7], (b) phagocytose non-opsonised tumor cells via pattern recog- nition receptors [8] and, (c) engage in antibody dependent phagocytosis (ADP) of opsonised targets [2, 9–12]. Thus, a major component of macrophage-mediated antitumor responses involves different types of phagocytosis. In one instance, recognition and binding of antibody-opsonised targets to macrophage Fcγ Receptors (FcγR; [2, 11]) is mediated via activation of a SYK/BTK pathway or a PI3K/p110δ pathway [2, 9]. The SYK/BTK pathway is antagonized by FcγRIIB via activation of the inhibitory phosphatase, Ship1 [2, 9]. In a second instance, FcγR- independent internalization of non-opsonised targets occurs

via pattern recognition molecules [8, 13]. Both FcγR- dependent and independent phagocytosis are antagonized by CD47 [8, 14] and the FcγR-independent pathway is additionally repressed by CD24 [8]. Early-phase trials have demonstrated the clinical potential of targeting these repressors in the context of malignancy (reviewed in [15]). Thus, there is evidence to suggest that strategies to enhance macrophage phagocytosis may promote antitumor immune responses to therapeutic antibodies and non-opsonised targets.
Monocytes from CLL patients can be used to generate monocyte-derived macrophages (MDMs) [2, 16]. MDMs, in the context of CLL and lymphomas, are often referred to as nurse like cells (NLCs) [3]. MDMs are capable of FcγR- dependent ADP in response to therapeutic antibodies and are a robust model of macrophage responses in CLL [2, 9, 11]. Indeed, evidence suggests that macrophages are a major effector cell mediating antibody responses in CLL (discussed in [11]). MDMs derived from normal healthy donors or patients with early/stable CLL recognize and kill CLL leukemic-cell targets in response to therapeutic anti- bodies (sensitive phenotype) [2]. In contrast, patients with advanced/progressing or relapsed CLL produce MDMs with reduced ADP activity (resistant phenotype) [2]. Although, the “resistance” phenotype is stable and main- tained in in vitro cultures for greater than 3 weeks we have shown that resistance can be reversed using inhibitors of Ship1 [2]. Ship1 is an FcγR2B activated phosphatase that dampens responses of activating FcγRs (e.g., FcγR1, 2A, and 3). Thus, Ship1 inhibitors derepress FcγR-dependent ADP leading to heightened ADP responses [2]. Similarly, FcγR2B-specific antibodies or antibodies that target the inhibitory CD47:SIRPα interaction also enhance responses to therapeutic antibodies [17]. Thus, strategies that target the effectors of “resistance” in MDMs can enhance FcγR- dependent antibody responses. Whilst these findings are encouraging, they rely on targeting specific steps in the resistance machinery. In contrast, strategies that globally reverse the “resistance phenotype” in MDMs may be more attractive as they reduce the probability of acquired resistance.
Macrophages are traditionally classifi ed as M1- or M2- like which refl ects the opposing ends of a spectrum of macrophage functionalities [18, 19]. Epigenetic modifiers have been shown to modify the differentiation status of macrophages in vitro and in vivo [7, 20, 21]. Studies have identifi ed that the immunomodulatory effects of HDACIs may be selective for isoforms belonging to one of the four HDAC classes (I, IIa/IIb, III, and IV) [7, 22, 23]. For example, HDAC class IIa-selective inhibitors (TMP269 or TMP195) induced phagocytosis and expression of char- acteristics traditionally attributed to M1-like macrophages [7, 21, 23]. In this study we examine the impact of class IIa

HDAC inhibition on FcγR-dependent and -independent phagocytic activity by phenotypically sensitive or resistant MDMs derived from CLL patients.

Results

TMP195 enhances ADP responses in primary cultures of phenotypically “resistant” MDMs from CLL patients

All peripheral blood mononuclear cell (PBMC) cultures (>98% CD68 + myeloid cells) are screened for their ability to mount an ADP response to the therapeutic antibodies, obinutuzumab, or rituximab [16]. Cultures are considered “sensitive” if >40% of MDMs participate in ADP and “resistant” if <20% MDMs participate in ADP (Fig. 1a and Supplementary Fig. 1). The phenotypic sensitivity/resis- tance is an inherent property of the MDMs. For example, co-culture of MDMs derived from a patient with a stable phenotype with leukemic B cells from a patient with a resistance phenotype will not alter the MDM phenotype. Similarly, co-culture of MDMs derived from a patient with a resistance phenotype with leukemic B cells from a patient with a sensitive phenotype will not alter the MDM pheno- type [2]. CLL cells will convert monocytes from normal healthy donors into MDMs that are similar in activity to phenotypically sensitive MDMs. Specifically, therapeutic antibody treatment induced a 64 ± 6.4% kill vs. a 66 ± 11.4% kill in MDMs generated from CLL patients with phenotypically sensitive MDMs vs those from normal healthy donors. In contrast, normal B cells do not have the ability to generate NLCs. For example, MDMs generated using healthy donor B cells are unable to respond to ther- apeutic antibody (100 + 8% viability in untreated controls vs. 90 ± 7 % viability, respectively). This is consistent with their inability to induce an NLC phenotype [24].
We examined phenotypically sensitive and resistant cultures to see if TMP195-enhanced ADP responses to obinutuzumab. The viability of untreated control cultures of CLL cells or TMP195-treated cultures was not altered by TMP195 (2 μM) treatment (100 ± 11% vs. 95 ± 10% viabi- lity, respectively; n = 8). Co-culture of PBMCs for 2 days with 2 μM TMP195 did not alter the ADP response of “phenotypically sensitive” MDMs to obinutuzumab (Fig. 1a, b). Specifi cally, 58.2 ± 7.7% “sensitive” MDMs participated in ADP (Fig. 1a, b). In contrast, similar treat- ment of PBMCs derived from patients with a “resistant” phenotype resulted in an enhanced ADP response to a level approaching that of “phenotypically sensitive” cultures (Fig. 1a, b). Specifi cally, 17.0 ± 1.5% of “resistant” MDMs participated in ADP which was increased to 42.4 ± 4.3% following incubation with TMP195 (Fig. 1a, b). Thus,

Fig. 1 TMP195 and vorinostat enhances FcγR-dependent ADP responses to obinutuzumab and rituximab. MDMs were generated from PBMCs obtained from patients with CLL as described in “Materials and Methods”. ADP response was characterized as “phe- notypically sensitive” or “phenotypically resistant” as described in “Materials and Methods”. a MDMs were generated from CLL PBMCs for over 5 days, followed by 48 h of treatment with 2 μM TMP195 (as shown). Cultures were then treated with obinutuzumab (10 μg/ml) or rituximab (10 μg/ml) for 2 h (as shown). Cells were then fixed and stained with Giemsa stain and phagocytosis was quantitated as the percent MDMs engulfi ng CLL cells as a fraction of the total pool of

MDMs in the culture. b Representative images of ADP activity of phenotypically resistant MDMs as visualized after Giemsa staining. c Phenotypically sensitive or d phenotypically resistant MDMs were generated from PBMCs, over 7 days, in the presence or absence of 2 μM TMP195 or 0.1 μM vorinostat from day 5–7. Cultures were then treated with obinutuzumab or rituximab for 2 h in the presence or absence of Fc-blocking reagent. Cells were fixed and stained with Giemsa stain and phagocytosis was quantitated “as above”. Data presented as mean ± sem from triplicate determinations of at least three biological replicates. Statistical differences assessed by Student’s t test. **P < 0.01, ***P < 0.001, ****P < 0.0001.

TMP195 enhances ADP response to therapeutic antibodies in phenotypically resistant MDMs.

TMP195 enhances FcγR-dependent and FcγR-independent phagocytosis

We examined the effect of the Fc-blocking antibody on ADP in vehicle- or TMP195-treated phenotypically “sen- sitive” and “resistant” MDMs (Fig. 1c, dand Supplementary Fig. 2a, b). In “sensitive” samples, Fc-blocking signifi cantly (P < 0.05) inhibited obinutuzumab-induced ADP (Fig. 1c and Supplementary Fig. 2a). Similarly, Fc-blocking sig- nificantly (P < 0.05) inhibited the TMP195-induced ADP response to obinutuzumab in “phenotypically resistant” MDMs (Fig. 1d and Supplementary Fig. 2b). These obser- vations were not restricted to the HDAC Class IIa-selective inhibitor, TMP195, since the pan-HDAC inhibitor, vorino- stat induced FcγR-dependent ADP responses to obinutu- zumab, albeit to a lesser extent compared to TMP195

(Fig. 1c, d and Supplementary Fig. 2a, b). These data indicate that enhanced ADP responses to TMP195 and vorinostat in “resistant” MDM are FcγR-dependent.
Next, we examined whether the effects of TMP195 and vorinostat could be extended to FcγR-independent phago- cytosis of non-opsonised targets. Utilizing pHrodo E. coli BioParticles, we found that 54.7 ± 8.0% of phenotypically “sensitive” MDMs phagocytosed Escherichia coli particles, which remained unchanged following TMP195 or vorino- stat treatment (Fig. 2a–c). In contrast, TMP195 and vor- inostat increased the percentage of phenotypically “resistant” MDMs phagocytosing E. coli from 17.6 ± 1.8% to 50.9 ± 5.8% and 55.0 ± 5.6%, respectively (Fig. 2d–f). Similar to the ADP response, we found that phagocytosis was increased to approximately the same level as observed for the phenotypically “sensitive” MDMs (Fig. 2d vs. 2A). Further quantitation showed that TMP195 and vorinostat also induced a greater than twofold increase in the number of phagocytic events for each MDM (Fig. 2e, f). This

Fig. 2 TMP195 and vorinostat increase FcγR-independent pha- gocytosis. Phenotypically sensitive (a–c) or phenotypically resistant (d, e). MDMs were generated in the presence or absence of 2μM TMP195 or 0.1μM vorinostat for 2 days after which they were incu- bated with 100μg/ml of pHrodo E. coli BioParticles, for 60min. Pha- gocytosis was then estimated by counting fl uorescent MDMs and presenting that as the percent of total MDMs that participate in

phagocytosis (a, c, d, f) or quantitating the internalized particles by fluorimetry (b, e) as described in the “Materials and methods”. c, f Representative images of phagocytosis assay. Data presented as mean ±sem from triplicate determinations of at least four biological repli- cates. Statistical differences assessed by Student’s t test. P values as written.

increase was observed only in phenotypically “resistant” MDMs. These data suggest that TMP195 and vorinostat can induce phagocytic activity in phenotypically “resistant” MDMs that approximates the ability of the phenotypically “sensitive” MDMs. To confirm that the E. coli phagocytosis was bona fi de we treated MDMs with a phagocytosis inhibitor, cytochalasin D [25]. Cytochalasin D was able to inhibit the TMP195- and vorinostat-enhanced E. coli pha- gocytosis (Supplementary Fig. 3a, b). These data demon- strate that TMP195 and vorinostat enhance FcγR-dependent and independent phagocytic responses by phenotypically resistant MDMs derived from CLL patients.

HDAC7 contributes to ADP responses by phenotypically “resistant” MDMs

In order to define which class IIa HDAC isoforms contribute to the ADP resistance phenotype, we examined the expres- sion levels of HDAC4, 5, 7, and 9 in MDMs. MDMs

expressed HDAC4, 7 and 9 at variable levels but expressed no detectable HDAC5 (Fig. 3a). Detection of HDAC5 in the positive control THP-1 cells confirmed the rigor of the negative result for HDAC5 (Supplementary Fig. 4a). Quantitation of HDAC7 expression revealed no significant difference in expression between sensitive or resistant MDMs (Supplementary Fig. 4b; 1.02 ± 0.21 (n = 9) vs. 1.24 ± 0.24 (n = 10) arbitrary expression units, respectively).
We examined which HDAC isoform(s) may contribute to ADP by using HDAC isoform-selective inhibitors (Fig. 3b–d). Preliminary experiments established concentrations of the inhibitors that were not cytotoxic to the MDMs or leukemic cells (Supplementary Fig. 5a, b). TMP195 and TMP269 are Class IIa-selective inhibitors [23]. Both caused similar levels of enhancement of ADP activity from 12.39 ± 1.485% MDMs participating to 36.9 ± 3.2% and 34.6 ± 2.7%, respectively MDMs participating (Fig. 3b). Con- sistent with previous studies [23], TMP195 and TMP269 did not cause hyperacetylation of histone H3 (Fig. 3c, d).

Fig. 3 Global histone H3 hyperacetylation is not required for TMP195/vorinostat-enhanced ADP. a Four donor MDMs were harvested and proteins extracted and analyzed for Class IIa HDAC expression by western blotting. Expression levels for β-actin are shown as a “loading” control. b MDMs were treated with vorinostat (0.1 μM), TMP195 (2 μM), TMP269 (2 μM), or LMK235 (0.1 μM) for

2 h and the ADP response to obinutuzumab estimated. c The effect of the inhibitor treatment on global histone H3 acetylation was quanti- tated (c) from western blots (representative blot shown), d following a 48-h treatment. Data presented as mean ± sem from triplicate deter- minations of at least four biological replicates. *P < 0.05, ***P < 0.001.

The structurally dissimilar pan-HDAC inhibitor [26], vor- inostat, induced a similar increase from 12.4 ± 1.5% to 32.5 ± 4.6% MDMs participating in ADP and induced hyperacetylation of H3 (Fig. 3b–d). Finally, an HDAC4/5- selective inhibitor, LMK235 [27], did not significantly alter the ADP response but did induce hyperacetylation of H3 (Fig. 3c, d). Combined, these data show that the resistance phenotype aligns most closely with HDACs 7 and 9 and that TMP195-enhanced ADP responses do not correlate with changes in global histone H3 acetylation status.
Recent studies in breast cancer models indicated that TMP195-induced antitumor effects of macrophages were mediated via HDAC7 [7]. Thus, we examined whether this may be the case in CLL-derived MDMs using siRNA knockdown of HDAC7. siRNAs targeting HDAC7 pro- foundly reduced HDAC7 expression to <10% of expres- sion detected in cells treated with the scrambled siRNA control (Fig. 4a, b). These same siRNAs did not alter

HDAC9 expression (Fig. 4a). HDAC7 knockdown sig- nifi cantly (P < 0.05) enhanced ADP response to obinutu- zumab by 3.14-fold in phenotypically resistant MDMs from CLL patients (Fig. 4c, d). Signifi cantly, HDAC7 knockdown enhanced ADP response to a level equivalent, or greater, than that induced by TMP195 treatment (com- pare with Figs. 2 and 3). These data indicate that HDAC7 contributes to ADP resistance and also indicate that HDAC7 knockdown alone can phenocopy the TMP195 response with respect to ADP.
Next, we examined whether HDAC7 also contributed to the muted FcγR-independent phagocytosis of non- opsonised targets in phenotypically resistant MDMs. siRNA knockdown of HDAC7 signifi cantly (P < 0.05) enhanced phagocytosis of E. coli in phenotypically resistant MDMs (Fig. 4e, f). Significantly, HDAC7 knockdown enhanced E. coli phagocytosis to a level equivalent, or greater, than that induced by TMP195 treatment. Thus,

Fig. 4 HDAC7 knockdown enhances FcγR-dependent and FcγR- independent phagocytosis. Phenotypically resistant MDMs were transfected with 40 nM scrambled or HDAC7-selective siRNAs (a–c). Forty-eight hours later protein was harvested and HDAC7, HDAC9, and β-actin expression visualized (a) and quantitated (b). c MDMs were transfected with 40 nM scrambled or HDAC7 C siRNA and 48 h

later ADP response to obinutuzumab (c, d) or phagocytosis of pHrodo E.coli BioParticles (e, f) was visualized (c, f) or quantitated (d, e). Quantitative data presented as mean ± sem from triplicate determina- tions of at least four biological replicates. Statistical differences assessed by Student’s t test. ns not significant, *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.001.

HDAC7 knockdown can phenocopy the TMP195 response with respect to E. coli phagocytosis.
Combined these data demonstrate that “resistant” MDMs derived from CLL patients acquire an immune pathology in which phagocytosis of opsonised and non-opsonised targets are suppressed. This pathology is reversible following HDAC7-specific depletion/inhibition.

HDAC7-dependent regulation of phagocytosis does not correlate with an alteration in the expression of M1/M2 phenotypic markers

We examined the time dependent effect of TMP195 on the change in ADP and phagocytosis (Fig. 5). ADP (Fig. 5a, b) and E. coli. phagocytosis (Fig. 5c–e) were signifi cantly enhanced within 30 and 120 min of TMP195 treatment and remained elevated for 48 h (Fig. 5a–e). The immediacy of TMP195 action suggests that HDAC7- dependent effects on phagocytosis of opsonised and non- opsonised targets may be mediated by a posttranslational mechanism.

An earlier study reported that TMP195 could induce an M2–M1 phenotypic switch in macrophages [7]. Thus, we examined whether the effects of TMP195 may be attribu- table to the induction of a phenotypic switch in the MDMs (Supplementary Fig. 6a–f). Supplementary Fig. 6b, e show that 12.6 ± 2.3% and 14.3 ± 1.0% of untreated MDMs express M2-like markers CD163 or CD206, respectively. Following 48-h treatment with TMP195 the expression of the M2 surface markers, CD163 or CD206 were unaltered or modestly increased (Supplementary Fig. 6a–f). This did not correlate with the signifi cant increases observed in ADP and E. coli phagocytosis observed in Figs. 1, 2, and 5. Paradoxically, after 7 days treatment, TMP195 induced an almost complete loss of detectable CD163 and CD206 expression (Supplementary Fig. 6a–f). This suggests that the increased ADP and phagocytosis are not correlated with the expression of M2-like macrophage markers, CD163 and CD206. This is confi rmed in Supplementary Fig. 6g which shows that obinutuzumab-induced ADP occurs in CD163 positive and negative MDMs. Similarly, the enhanced ADP response to obinutuzumab following TMP195 or vorinostat

Fig. 5 TMP195 and vorinostat induce rapid changes in ADP and phagocytic responses in MDMs. Phenotypically resistant MDMs were incubated with 2 μM TMP195 or 0.1 μM vorinostat for varying times after which ADP response to obinutuzumab was visualized (a) and percent MDMs participating in ADP quantitated (b). Phenotypi- cally resistant MDMs were incubated with 2 μM TMP195 or 0.1 μM vorinostat for varying times after which phagocytosis of pHrodo E. coli BioParticles was visualized (c) and percent MDMs participating in phagocytosis quantitated (d). Data presented as mean ± sem from triplicate determinations of at least four biological replicates. Statistical differences assessed by Student’s t test. *P < 0.05, **P < 0.01.

also occurred in CD163 positive and negative MDMs (Supplementary Fig. 6g). These data demonstrate that ADP is an attribute of CD163 positive and negative MDMs and shows that TMP195-enhanced phagocytic responses are not attributable to a classical M2–M1 phenotypic switch.

HDAC7-dependent regulation of phagocytosis is mediated via direct modulation of BTK activity

We have previously shown that FcγR-dependent ADP responses are controlled posttranslationally via post- receptor signaling through a SYK/BTK pathway [2]. Therefore, we examined the activation status of the SYK/
BTK pathway following a 2-h treatment with 2 μM TMP195 in phenotypically “resistant” MDMs using FcOxyburst as an activator (Fig. 6a). FcOxyburst is an opsonised substrate that activates FcγR-dependent events. The advantage of FcOxyburst over obinutuzumab- opsonised CLL cells is that it activates FcγR-dependent events in the absence of introducing any extraneous cellular protein (i.e., CLL cell protein). Thus, it allows us to accu- rately interrogate MDM-specifi c changes following FcγR activation.

Consistent with their “resistant” phenotype, FcOxyburst did not induce significant changes in BTK or SYK phos- phorylation but induced Erk1/2 phosphorylation (Fig. 6a, b). Significantly, only BTK phosphorylation was elevated following TMP195 + FcOxyburst combination treatment (Fig. 6a, b). These data indicate that TMP195- enhanced ADP is accompanied by BTK phosphorylation/
activation. The histone acetyl transferase, p300, acetylates BTK to promote BTK phosphorylation/activation [28]. To determine whether the acetylation status of BTK may be altered by HDAC7 in our system we treated THP-1 cells with FcOxyburst (10 ng/ml) alone or in combination with TMP195 (2 μM) or vorinostat (0.1 μM) for 2 h and then immunoprecipitated BTK and probed for acetyl lysine. The limiting protein yields from MDMs necessitated the use of THP-1 macrophages. THP-1 cells are a macrophage cell line that participates in FcγR-dependent phagocytosis [29]. TMP195 and vorinostat hyperacetylated BTK to levels at least twofold greater than that observed following FcOx- yburst treatment alone (Fig. 6c and Supplementary Fig. 7). These data indicate that hyperacetylation of BTK accom- panies TMP195- or vorinostat-enhanced ADP and BTK phosphorylation/activation. These data also indicate that BTK may be a direct deacetylase substrate of HDAC7.
To determine whether BTK activation was causal in HDAC7-mediated ADP we co-treated phenotypically resistant MDMs with a non-toxic dose (3 μM) [2, 9] of the BTK-selective inhibitor, ibrutinib, combined with either TMP195 or vorinostat for 2 h and then measured ADP (Fig. 6d, e) or E. coli. phagocytosis (Fig. 6f, g). TMP195 and vorinostat-enhanced ADP (Fig. 6d, e) and E. coli. phagocytosis (Fig. 6f, g) whilst ibrutinib had no effect on either ADP (Fig. 6d, e) or E. coli. phagocytosis (Fig. 6f, g). TMP195 or vorinostat + ibrutinib profoundly reduced the ADP response to a level equivalent to that of ibrutinib alone (Fig. 6d, e). Similarly, TMP195 or vorinostat + ibrutinib profoundly reduced phagocytosis of E. coli. (Fig. 6f, g). These data show that the HDAC7 inhibitor-enhanced MDM phagocytosis is mediated via almost immediate BTK activation.

Discussion

In this study we show that HDAC7 regulates FcγR-depen- dent and FcγR-independent phagocytosis in MDMs derived from CLL patients via direct modulation of BTK acetylation and phosphorylation. Moreover, we show that HDAC7 contributes to muted phagocytic responses commonly associated with MDMs derived from patients with pro- gressive, relapsed or refractory disease. Finally, we show that inhibition of HDAC7 with a clinically available agent (e.g., vorinostat) is able to restore phagocytic responses to

Fig. 6 HDAC7-dependent regulation of phagocytosis is mediated via BTK. a Phenotypically resistant MDMs were incubated with 12 μg/ml FcOxyburst with or without 2 μM TMP195 for 2 h. Total cellular protein extracts were generated and the expression of phos- phorylated forms of BTK, SYK, or ERK1/2 quantitated. b A repre- sentative western blot is shown with GAPDH as a reference loading control. c THP-1 cells were incubated with 12 µg/ml FcOxyburst with or without 2 μM TMP195 for 2 h. Total cellular protein was harvested and immunoprecipiated with BTK antibody + protein A dynabeads. Immunoprecipitates were then run on a gel and blotted with an acetylated-lysine antibody. d Phenotypically resistant MDMs/CLL cell cultures were preincubated with 3 μM ibrutinib, 2 μM TMP195, or 0.1 μM vorinostat for 30 min followed by incubation with 10 μg/ml

obinutuzumab for 2 h. ADP is presented as the percentage of MDMs participating in ADP as assessed from the Giemsa-stained cultures. e Representative micrographs of d are shown. f Phenotypically resis- tant MDMs were preincubated with 3 μM ibrutinib, 2 μM TMP195, or 0.1 μM vorinostat for 2 h followed by incubation with 100 μg/ml fluorescent pHrodo E. coli BioParticles for 60 min. Phagocytosis was quantitated in ImageJ is presented as fl uorescent intensity of inter- nalized particles. g Representative micrographs of f are shown. Data presented as mean ± sem of triplicate determinations from at least four biological replicates. Statistical differences determined by an unpaired Student’s t test. ns not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 7 Diagram of the proposed shared pathway controlling FcγR- dependent and FcγR-independent phagocytic responses in MDMs from CLL patients. IC immune complex and PRR pattern recognition receptors.

opsonised and non-opsonised targets. These findings sug- gest that HDAC7-specific inhibitors could be used to improve responses to therapeutic antibodies and to enhance the innate immune response in CLL patients.
Our observation that HDAC7 suppresses phagocytic responses in MDMs is novel and suggests that activation of HDAC7-dependent events occurs during CLL progression as well as contributes to therapeutic antibody resistance and suppressed innate immune responses. The immediacy of the effects of TMP195 or vorinostat, suggest that the pathology is likely due to alterations in the activation state of HDAC7 or downstream effectors of HDAC7-mediated suppression of phagocytosis such as BTK (Fig. 7). This is supported by our observation that TMP195 treatment or HDAC7 inhibi- tion induces almost immediate hyperphosphorylation of BTK. Similarly, BTK phosphorylation was increased fol- lowing HDAC7 knockdown. Moreover, TMP195 did not alter the activation status of SYK which lies upstream of BTK in the FcγR-dependent ADP signaling pathway. Finally, selective inhibition of BTK, with ibrutinib, was able to inhibit TMP195/vorinostat-dependent enhancement of ADP and phagocytosis responses. Combined, these data demonstrate that HDAC7 suppresses BTK activation and BTK-dependent phagocytic events (Fig. 7). This is con- sistent with our earlier finding that BTK activity was sup- pressed in phenotypically resistant MDMs derived from patients with progressive/relapsed disease compared to phenotypically sensitive MDMs derived from healthy donors or patients with early-stage CLL [2].

A major conclusion from the present study is that HDAC7 is a major, if not exclusive, HDAC isoform to regulate responses, by MDMs, to both opsonised and non- opsonised targets. Equally important, is our fi nding that HDAC7 is causally involved in resistance to therapeutic antibodies in MDMs from CLL patients. This has clinical relevance since therapeutic antibodies are important agents in CLL treatment and macrophages are the major effector cell type mediating responses to therapeutic antibodies [10]. There are several independent lines of evidence to indicate that HDAC7 is the major HDAC isoform involved in phagocytic responses by MDMs. Previous studies have shown that TMP195 is a class IIa-selective HDAC inhibitor with particular potency for HDACs 7 and 9 that does not affect histone H3 acetylation levels [23]. Consistent with this, we find that TMP195 and its close analog, TMP269, do not alter histone H3 acetylation status but do induce almost immediate increases in phagocytic responses. Although TMP195 can inhibit HDACs 4 and 5 [23] we found that an HDAC4/5-selective inhibitor induced no significant change in ADP activity suggesting they were not major contributors to this response. Moreover, expression analysis did not detect HDAC5 expression supporting the conclusion it is not meaningfully involved. Most significantly, we noted that the effect of HDAC7-selective knockdown recapitu- lated entirely the magnitude of the effect observed with TMP195 on phagocytic responses. Combined, these data indicate that HDAC7 is the major, if not exclusive, HDAC driving resistance to therapeutic antibodies and non- opsonised targets by MDMs derived from CLL patients.
Whilst it remains unresolved whether BTK is a deace- tylase substrate of HDAC7 in MDMs our data clearly show that BTK acetylation is linked to FcγR-dependent and independent phagocytosis. An earlier study of CD19 + leukemic cells, showed that patients with a poor prognosis/
survival have high global HDAC activity in their CD19 + leukemic cells which is predictive of low global acetylation levels [30]. Moreover, Liu et al. [28] have shown that the histone acetyl transferase, p300, directly acetylates BTK lysine residues which promotes hyperphosphorylation (activation) of BTK [28]. Thus, the acetylase/deacetylase system directly modifies the activity of BTK in CD19 + leukemic B cells. Extending this, we now show that the class IIa specifi c inhibitor, TMP195, induces hyper- acetylation and hyperphosphorylation of BTK within 30–120 min in MDMs. This is the first time that BTK has been shown to be a deacetylase target and is the first report of a putative endogenous HDAC7-substrate. It was not determined, in this study, if HDAC7 was the catalyst for BTK deacetylation or whether it acted as a reader, facil- itating deacetylation by an alternate HDAC [23]. However, we do note that functional activation of ADP and BTK is restricted to class IIa-selective inhibitors and is

accompanied by hyperacetylation of BTK. Finally, we show that HDAC7 suppresses FcγR-dependent and independent phagocytosis by MDMs derived from CLL patients. Thus, our findings highlight a novel and actionable pathology that may contribute to therapeutic antibody resistance and innate immune suppression.
CLL is often accompanied by multiple innate immune defects. Some of these defects could be attributable, in part, to HDAC7-dependent BTK suppression. For example, CLL progression and relapse is associated with the acquisition of ADP resistance [2, 9]. In addition, patients with advanced CLL display generalized immune suppression characterized by a high incidence of immune-sensitive squamous cell carcinomas [31–33] and a susceptibility to infections due to their B-cell pathology as well as the immune suppression caused by their treatments [34, 35]. Similarly, patients with X-linked agammaglobulinemia, who lack BTK, are also susceptible to bacterial and fungal infections normally cleared by macrophages (reviewed in [34, 35]). Finally, there is accumulating evidence that therapeutic BTK inhi- bitor use is accompanied by an increased incidence of invasive fungal infections that are normally cleared by macrophages (reviewed in [34, 35]). These observations highlight the potential importance of BTK to innate immunity and ADP responses to therapeutics. Our study now shows this immune pathology may be attributed, in part, to dysregulation of an HDAC7/BTK axis.
Overall, there is clear evidence that impaired FcγR- dependent and -independent phagocytosis are a pressing clinical issue in CLL patients. In this regard, our findings show that targeted inhibition of HDAC7 with TMP195 is able to modify both these pathological phenotypes. Thus, HDAC7-selective inhibitors, could serve multiple clinical uses to (i) reinstate innate antitumor responses, (ii) enhance FcγR-dependent ADP responses in CLL patients, and (iii) restore innate immune activity against infections in CLL patients.

Materials and methods

CLL patient samples and PBMC cultures

Peripheral blood from CLL patients was collected with informed consent according to protocols approved by the Princess Alexandra Hospital (PAH) Human Research Ethics Committee in accordance with the Declaration of Helsinki. Diagnosis of CLL was made according to iwCLL criteria [36]. Clinical features of the patients used is provided in Supplementary Table 1. PBMCs were isolated and cultured as described previously [16, 37]. Purified CLL PBMCs were cultured for 7 days to allow maturation of monocytes into MDMs as described previously [2]. Cultures were

incubated with rituximab (10 µg/ml; Mabthera; Roche) or obinutuzumab (10 µg/ml; GA101; Roche) as indicated.

Reagents

TMP195, TMP269, Vorinostat, LMK235, and Ibrutinib were purchased from Selleckchem. Cytochalasin D was from Sigma-Aldrich, pHrodo E. coli BioParticles and FcOxyburst Green Assay reagent were from Thermo Fisher Scientifi c, the CellTiter 96® AQueous One Solution Cell Proliferation Assay was from Promega and Fc-block reagent was purchased from Miltenyi Biotech.

ADP Giemsa assay

Following 7 days in culture, CLL PBMCs were treated with indicated inhibitors before addition of therapeutic anti- bodies for an additional 2 h where indicated. Non-adherent cells were removed by gentle agitation and slides stained with May–Grunwald–Giemsa [2]. Samples were analyzed as previously described [3]. A minimum of nine fields were examined per well. All quantitation was performed using ImageJ software.

E. coli phagocytosis assay

CLL PBMCs were cultured as indicated above. CLL cells were removed by gentle agitation and MDMs were treated with indicated inhibitors for 2 h before the addition of 100 µg/ml of pHrodo E. coli BioParticles (Thermo Fisher Scientifi c) for 60 min. Wells were washed with PBS, before resuspending in Live Cell Imaging Solution containing ProLong Live Antifade Reagent (both from Thermo Fisher Scientifi c). Fluorescent signals and differential interference contrast were viewed through an Olympus FV1200 Laser Scanning Confocal Microscope system (Olympus) and analyzed using Olympus FV-10-ASW imaging software.

Immunoblotting

Immunoblotting has been described previously [2]. Anti- bodies used for immunoblot include: total BTK (#8547), phospho-BTK (Tyr223; #5082), total SYK (#13198), phospho-SYK (Tyr525/526; #2710), p44/42 MAPK (Erk1/
2) (#4695), phospho-p44/42 MAPK (Erk1/2) (Thr202/
Tyr204; #9101); GAPDH (#5174); HDAC4 (#15164), HDAC5 (#20458), HDAC7 (#33418), acetylated histone H3 (#9649) all from Cell Signaling, Histone H3 (06-755; Millipore), HDAC9 (ab109446; Abcam), and β-actin (sc47778, Santa Cruz Biotechnology). Horseradish perox- idase conjugated goat anti-rabbit (Sigma-Aldrich) and goat anti-mouse (Sigma-Aldrich) secondary antibodies were used for detection. Where protein expression has been

quantitated, results represent relative protein levels nor- malized to corresponding total protein expression levels using ImageJ software.

HDAC7 knockdown

CLL PBMCs were cultured as described above. Non- adherent cells were removed by gentle agitation before MDMs were transfected with 40 nM predesigned HDAC7 siRNA (HDAC7 Trilencer-27 Human siRNA; SR309850; OriGene) using Lipofectamine RNAiMAX (Thermo Fisher Scientifi c) according to the manufacturers’ instructions. Non-targeting scrambled siRNA (OriGene) was used as a control. Cells were harvested 48-h post transfection for western blotting or functional assays.

Co-immunoprecipitation

Following treatment, MDM cell lysates were immunopre- cipitated with a rabbit anti-BTK antibody (#8547; Cell Signaling) or rabbit IgG XP isotype control (#3900; Cell Signaling) followed by capture with Protein A DynaBeads (Thermo Fisher Scientific). BTK acetylation was visualized with an acetylated-lysine-HRP antibody (#6952; Cell Signaling).

Data analysis and statistics

Results are presented as mean ± sem of at least three bio- logical replicates each performed in triplicate. All statistical analyses used an unpaired Student’s t test estimated using GraphPad Prism and findings were considered signifi cant if P ≤ 0.05.

Acknowledgements The authors acknowledge the generosity of the patients who donated their time and blood for this study. This work was supported by a generous donation from Jeff and Fran Maclean and a research support package to DG from Queensland Health, Princess Alexandra Hospital.

Compliance with ethical standards

Confl ict of interest The authors declare that they have no conflict of interest.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1.Bosch F, Dalla-Favera R. Chronic lymphocytic leukaemia: from genetics to treatment. Nat Rev Clin Oncol. 2019;16:684–701.
2.Burgess M, Mapp S, Mazzieri R, Cheung C, Chambers L, Mat- tarollo SR, et al. Increased FcgammaRIIB dominance contributes to the emergence of resistance to therapeutic antibodies in chronic lymphocytic leukaemia patients. Oncogene. 2017;36:2366–76.

3.Chen YCE, Mapp S, Blumenthal A, Burgess ML, Mazzieri R, Mattarollo SR, et al. The duality of macrophage function in chronic lymphocytic leukaemia. Biochim Biophys Acta Rev Cancer. 2017;1868:176–82.
4.Hanna BS, McClanahan F, Yazdanparast H, Zaborsky N, Kalter V, Rossner PM, et al. Depletion of CLL-associated patrolling monocytes and macrophages controls disease development and repairs immune dysfunction in vivo. Leukemia. 2016;30:570–9.
5.Galletti G, Scielzo C, Barbaglio F, Rodriguez TV, Riba M, Lazarevic D, et al. Targeting macrophages sensitizes chronic lymphocytic leukemia to apoptosis and inhibits disease progres- sion. Cell Rep. 2016;14:1748–60.
6.Salvagno C, Ciampricotti M, Tuit S, Hau CS, van Weverwijk A, Coffelt SB, et al. Therapeutic targeting of macrophages enhances chemotherapy effi cacy by unleashing type I interferon response. Nat Cell Biol. 2019;21:511–21.
7.Guerriero JL, Sotayo A, Ponichtera HE, Castrillon JA, Pourzia AL, Schad S, et al. Class IIa HDAC inhibition reduces breast tumours and metastases through anti-tumour macrophages. Nat- ure. 2017;543:428–32.
8.Barkal AA, Brewer RE, Markovic M, Kowarsky M, Barkal SA, Zaro BW, et al. CD24 signalling through macrophage Siglec-10 is a target for cancer immunotherapy. Nature. 2019;572:392–6.
9.Enya Chen YC, Burgess M, Mapp S, Mollee P, Gill D, Blumenthal A, et al. PI3K-p110delta contributes to antibody responses by macrophages in chronic lymphocytic leukemia. Leukemia. 2020;35:451–61.
10.Church AK, VanDerMeid KR, Baig NA, Baran AM, Witzig TE, Nowakowski GS, et al. Anti-CD20 monoclonal antibody- dependent phagocytosis of chronic lymphocytic leukaemia cells by autologous macrophages. Clin Exp Immunol. 2016;183:90–101.
11.VanDerMeid KR, Elliott MR, Baran AM, Barr PM, Chu CC, Zent CS. Cellular cytotoxicity of next-generation CD20 monoclonal antibodies. Cancer Immunol Res. 2018;6:1150–60.
12.Gul N, Babes L, Siegmund K, Korthouwer R, Bogels M, Braster R, et al. Macrophages eliminate circulating tumor cells after monoclonal antibody therapy. J Clin Investig. 2014;124:812–23.
13.Uribe-Querol E, Rosales C. Control of phagocytosis by microbial pathogens. Front Immunol. 2017;8:1368.
14.Veillette A, Chen J. SIRPalpha-CD47 immune checkpoint block- ade in anticancer therapy. Trends Immunol. 2018;39:173–84.
15.Weiskopf K. Cancer immunotherapy targeting the CD47/SIR- Palpha axis. Eur J Cancer. 2017;76:100–9.
16.Burgess M, Gill D, Singhania R, Cheung C, Chambers L, Renyolds BA, et al. CD62L as a therapeutic target in chronic lymphocytic leukemia. Clin Cancer Res. 2013;19:5675–85.
17.Roghanian A, Teige I, Martensson L, Cox KL, Kovacek M, Ljungars A, et al. Antagonistic human FcgammaRIIB (CD32B) antibodies have anti-tumor activity and overcome resistance to antibody therapy in vivo. Cancer Cell. 2015;27:473–88.
18.Murray PJ. Macrophage polarization. Annu Rev Physiol. 2017;79:541–66.
19.Ginhoux F, Schultze JL, Murray PJ, Ochando J, Biswas SK. New insights into the multidimensional concept of macrophage onto- geny, activation and function. Nat Immunol. 2016;17:34–40.
20.Kimbrough D, Wang SH, Wright LH, Mani SK, Kasiganesan H, LaRue AC, et al. HDAC inhibition helps post-MI healing by modulating macrophage polarization. J Mol Cell Cardiol. 2018;119:51–63.
21.Qi X, Wang P. Class IIa HDACs inhibitor TMP269 promotes M1 polarization of macrophages after spinal cord injury. J Cell Bio- chem. 2018;119:3081–90.
22.Bobrowicz M, Dwojak M, Pyrzynska B, Stachura J, Muchowicz A, Berthel E, et al. HDAC6 inhibition upregulates CD20 levels and increases the efficacy of anti-CD20 monoclonal antibodies. Blood. 2017;130:1628–38.

23.Lobera M, Madauss KP, Pohlhaus DT, Wright QG, Trocha M, Schmidt DR, et al. Selective class IIa histone deacetylase inhibi- tion via a nonchelating zinc-binding group. Nat Chem Biol. 2013;9:319–25.
24.Boissard F, Fournie JJ, Laurent C, Poupot M, Ysebaert L. Nurse like cells: chronic lymphocytic leukemia associated macrophages. Leuk Lymphoma. 2015;56:1570–2.
25.Ribes S, Ebert S, Regen T, Agarwal A, Tauber SC, Czesnik D, et al. Toll-like receptor stimulation enhances phagocytosis and intracellular killing of nonencapsulated and encapsulated Strep- tococcus pneumoniae by murine microglia. Infect Immun. 2010;78:865–71.
26.Khan N, Jeffers M, Kumar S, Hackett C, Boldog F, Khramtsov N, et al. Determination of the class and isoform selectivity of small- molecule histone deacetylase inhibitors. Biochemical J. 2007;409:581–9.
27.Marek L, Hamacher A, Hansen FK, Kuna K, Gohlke H, Kassack MU, et al. Histone deacetylase (HDAC) inhibitors with a novel connecting unit linker region reveal a selectivity profile for HDAC4 and HDAC5 with improved activity against chemore- sistant cancer cells. J Medicinal Chem. 2013;56:427–36.
28.Liu Z, Mai A, Sun J. Lysine acetylation regulates Bruton’s tyr- osine kinase in B cell activation. J Immunol. 2010;184:244–54.
29.Ackerman ME, Moldt B, Wyatt RT, Dugast A-S, McAndrew E, Tsoukas S, et al. A robust, high-throughput assay to determine the phagocytic activity of clinical antibody samples. J Immunol Methods. 2011;366:8–19.
30.Van Damme M, Crompot E, Meuleman N, Mineur P, Dessars B, El Housni H, et al. Global histone deacetylase enzymatic activity is an independent prognostic marker associated with a shorter

overall survival in chronic lymphocytic leukemia patients. Epi- genetics. 2014;9:1374–81.
31.Jung JW, Veitch M, Bridge JA, Overgaard NH, Cruz JL, Linedale R, et al. Clinically-relevant rapamycin treatment regimens enhance CD8(+) effector memory T cell function in the skin and allow their infi ltration into cutaneous squamous cell carcinoma. Oncoimmunology. 2018;7:e1479627.
32.Ramsay HM, Fryer AA, Hawley CM, Smith AG, Harden PN. Non-melanoma skin cancer risk in the Queensland renal transplant population. Br J Dermatol. 2002;147:950–6.
33.Lindelof B, Sigurgeirsson B, Gabel H, Stern RS. Incidence of skin cancer in 5356 patients following organ transplantation. Br J Dermatol. 2000;143:513–9.
34.Varughese T, Taur Y, Cohen N, Palomba ML, Seo SK, Hohl TM, et al. Serious infections in patients receiving ibrutinib for treat- ment of lymphoid cancer. Clin Infect Dis Off Publ Infect Dis Soc Am. 2018;67:687–92.
35.Rogers KA, Mousa L, Zhao Q, Bhat SA, Byrd JC, E Boghdadly Z, et al. Incidence of opportunistic infections during ibrutinib treatment for B-cell malignancies. Leukemia. 2019;33:2527–30.
36.Hallek M, Cheson BD, Catovsky D, Caligaris-Cappio F, Dighiero G, Dohner H, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111:5446–56.
37.Burgess M, Cheung C, Chambers L, Ravindranath K, Minhas G, Knop L, et al. CCL2 and CXCL2 enhance survival of primary chronic lymphocytic leukemia cells in vitro. Leuk Lymphoma. 2012;53:1988–98.