Increased Replication of Human Cytomegalovirus in Retinal Pigment Epithelial Cells by Valproic Acid Depends on Histone Deacetylase Inhibition

Abstract

purpose. Human cytomegalovirus (HCMV) replication depends on different cellular pathways, including histone acetylation and extracellular-signal regulated kinases 1 and 2 (Erk 1/2). In the present study, the influence of therapeutic valproic acid (VPA) concentrations was investigated on HCMV replication in retinal pigment epithelial (RPE) cells.

methods. HCMV antigen expression and replication were detected by immunostaining, real-time RT-PCR, and determination of virus titers. Histone acetylation and Erk 1/2 phosphorylation were detected by Western blot.

results. Pretreatment with VPA ≤1 mM enhanced HCMV antigen expression and replication by up to ninefold. In addition to histone deacetylase (HDAC) inhibition, VPA stimulated Erk 1/2 phosphorylation in RPE cells. Investigation of six VPA derivatives revealed that S-2-pentyl-4-pentynoic acid was the only derivative that induced histone hyperacetylation, indicating HDAC inhibition, in the observed concentrations ≤1 mM and that increased HCMV antigen expression. Other derivatives did not enhance HCMV replication in the tested concentrations, although some were found to induce Erk 1/2 phosphorylation. The mitogen-activated protein kinase kinase (MEK) inhibitor PD98059 inhibited VPA-induced Erk 1/2 phosphorylation but did not affect VPA-induced increased HCMV replication. In addition, the structurally nonrelated HDACI trichostatin A enhanced HCMV replication but did not affect Erk 1/2 phosphorylation in RPE cells.

conclusions. The data demonstrate that VPA stimulates HCMV replication by HDAC inhibition independent of Erk 1/2 phosphorylation in therapeutic concentrations in RPE cells. Therefore, patients at risk of HCMV retinitis who are treated with VPA or other HDAC inhibitors should be carefully monitored.

The histone deacetylase inhibitor (HDACI) valproic acid (VPA) ^{1 2} is one of the most frequently prescribed antiepileptic drugs, ³ and it is becoming the first-choice treatment for bipolar disorder worldwide. ⁴ In addition, VPA is clinically used for several other diseases, including schizophrenia and different forms of headache ^{5 6 7} and is under experimental and clinical investigation as an anticancer drug. ⁸  

Therapeutic VPA concentrations increase human cytomegalovirus (HCMV) replication in fibroblasts. ⁹ Infection with HCMV leads to a life-long persistence in 50% to 90% of the adult population. In contrast to immunocompetent individuals in whom HCMV infection usually does not induce clinical symptoms, ¹⁰ HCMV infection and reactivation leads to serious diseases of different organs, such as the colon, lung, or central nervous system, including the retina of immunocompromised patients. ^{11 12} It is of interest to investigate the influence of VPA on HCMV replication in specific cell types that are involved in HCMV diseases. 

The retinal pigment epithelium (RPE) plays a central role in HCMV retinitis. ^{11 13} Therefore, RPE cells are a common model for the investigation of pathologic conditions during HCMV retinitis and examination of drugs that potentially influence HCMV retinitis. ^{13 14 15 16 17 18 19 20 21 22} Compared with HCMV infection of fibroblasts, HCMV infection of RPE cells results in a smoldering infection characterized by delayed HCMV replication. ^{11 14} In contrast to other permissive cell types, HCMV replication in RPE cells does not depend on nuclear factor (NF)-κB activation and cannot be inhibited by NFκB inhibition. ^{11 16} However, the stimulation of the extracellular signal–regulated kinases 1 and 2 (Erk 1/2), mitogen-activated protein kinase (MAPK) was found to be of crucial importance for HCMV replication in RPE cells. ^{11 16}  

In the current study, we investigated the influence of VPA on histone acetylation and HCMV replication in RPE cells. The acetylation status of cellular histones is controlled by histone acetyltransferases that promote histone acetylation, and HDACs that promote histone deacetylation. ^{23 24 25} The histone acetylation status influences the expression of numerous cellular genes involved in cell proliferation, differentiation, and programmed cell death. HDACs had already been found to be involved in repression of the HCMV major immediate early promoter (MIEP) ²⁶ and the HCMV UL123-encoded immediate early antigen 1 (IEA1) was demonstrated to inhibit HDAC-3. ²⁷ In addition to its HDAC inhibiting activity, VPA was found to stimulate Erk 1/2 phosphorylation in certain cell types. ^{28 29 30} Therefore, the influence of VPA-induced histone hyperacetylation and Erk 1/2 phosphorylation on HCMV replication in RPE cells was examined. 

Materials and Methods

All cell culture supplements were purchased from Biochrom (Berlin, Germany). VPA was obtained from Sigma-Aldrich (Taufenkirchen, Germany). The VPA derivatives (Fig. 1)were synthesized according to methods described before. ^{31 32 33} The structurally nonrelated HDACI trichostatin A (TSA) was obtained from Merck Biosciences (Darmstadt, Germany) and served as a control substance throughout the experiments. Erk 1/2 is commonly phosphorylated by the mitogen-activated protein kinase kinases 1/2 (MEK 1/2). To inhibit Erk 1/2 phosphorylation the MEK 1/2 inhibitor PD98059 (Merck Biosciences) was used during experiments. 

Cells

RPE cells were established from donor eyes and cultivated as described previously. ^{15 16} Cells in passages 3 to 6 were used for experiments. The homogeneity of cultured RPE cells was confirmed by positive immunostaining with mAb to cytokeratins (Pan; Sigma-Aldrich) and to cellular retinaldehyde binding protein (mAb was donated by John Saari, Department of Ophthalmology, University of Washington, School of Medicine, Seattle, WA). 

Viruses

Strain Hi91 was isolated from the urine of a patient with AIDS who had HCMV retinitis. ³⁴ The HCMV laboratory strains Towne and AD169 were obtained from American Type Culture Collection (Manassas, VA). Virus stocks were prepared in human foreskin fibroblasts (HFF) grown in MEM with 4% fetal calf serum (FCS). The titers were determined by plaque titration, as described previously. ³⁵  

Virus Infectivity Assay

Confluent cultures of RPE cells were incubated with HCMV at a multiplicity of infection (MOI) of 0.1, as described before, ³⁶ resulting in less than 0.1% infected cells. Cells producing HCMV-specific antigens were detected 24 and 72 hours after infection by immunoperoxidase staining using monoclonal antibodies directed against the UL123-coded 72-kDa immediate early antigen 1 (IEA1; DuPont, Bad Homburg, Germany) and UL55-encoded late antigen gB (LA; kindly provided by Klaus Radsak, Institute for Virology, Marburg, Germany), respectively. 

Virus Yield Assay

The amount of infectious virus was determined by virus yield assay in a single-cycle assay format, as described before. ³⁶ In contrast to the virus yield assay in fibroblasts in which virus titers were determined 3 days after infection, in RPE cells, virus was titrated 5 days after infection. 

Cell Viability Assay

Viability of mock-infected, VPA (derivative)-treated RPE cells was assessed by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) dye-reduction assay, as described previously. ⁹  

Real-Time PCR

Total RNA was isolated from infected cell cultures using extraction reagent (TRI; Sigma-Aldrich). RNA was reverse transcribed by using random hexamer priming. The relative quantification of cDNA for IEA1 and -2 (encoded by UL122) and β-actin was obtained by real-time PCR (Prism 7900HT; Applied Biosystems, Inc., [ABI] Foster City, CA). All reactions were performed with a primer/probe assay according to a standard thermal profile: denaturation at 95°C for 15 seconds, annealing/extension at 60°C for 60 seconds with 40 repeats. The following primer sequences were used: IE1 forward 5′-GCC GCA CCA TGT CCA CT, IE1 reverse 5′-CAA CGA GAA CCC CGA GAA AG, IE1 probe 5′-(VIC) CCT TAA TCT GTT TGA CGA GTT CTG CCA GGA C-(TAMRA); IE2 forward 5′-CTC GCT ATC AGA GAT CAC GAT ACA G, IE2 reverse 5′-CAA CGA GAA CCC CGA GAA AG, IE2 probe 5′-(VIC) CTC ggg CTT gAT gTC TTC CTg TTT gAT g-(TAMRA); and β-actin forward 5′-CGC GAG AAG ATG ACC CAG AT, β-actin reverse 5′-CAG AGG CGT ACA GGG ATA GCA, β-actin probe 5′-(FAM) TGA GAC CTT CAA CAC CCC AGC CAT GT-(TAMRA). Relative quantification was determined by computer (SDS ver. 2.1 software; ABI) provided with the PCR system. Normalization was obtained by using β-actin as the endogenous control and the infected cell culture as a calibration sample in comparison to the treated and infected cell culture. The results are presented as x-fold change. 

Immunoblot Analysis

Immunoblot analysis was performed as described before. ⁹ Proteins were detected with specific antibodies against β-actin (Sigma-Aldrich), acetylated histone H4 (Upstate Biotechnology, Lake Placid, NY), Erk 1/2 (Cell Signaling Technology, Frankfurt am Main, Germany), and phosphorylated Erk 1/2 (pErk 1/2; Cell Signaling Technology). 

Statistics

Results are expressed as the mean ± SD of at least three experiments. Comparisons between two groups were performed using Student’s t-test. Three or more groups were compared by ANOVA followed by the Student-Newman-Keuls test. P < 0.05 was considered to be significant. 

Results

Influence of VPA on HCMV Replication in RPE Cells

VPA pretreatment for 24 hours increased HCMV IEA1 and LA expression in a concentration-dependent manner in RPE cells infected with HCMV strain Hi91 at an MOI of 0.1 (Fig. 2A) . VPA (1 mM) caused an approximately fourfold increase of IEA1 and LA expression, as shown by immunostaining. Real-time PCR indicated an approximate eightfold increase of HCMV IEA1 and HCMV IEA2 mRNA levels in RPE cells pretreated with 1 mM VPA for 24 hours (Fig. 2B) . In contrast, addition of VPA during and/or after HCMV infection did not affect HCMV IEA expression. Concentrations higher than 1 mM further increased HCMV strain Hi91 IEA1 and LA expression (data not shown), but were not investigated in detail, since these concentrations are above the range of therapeutically achievable plasma levels in patients. ^{37 38} Higher MOIs brought similar results (data not shown). 

Determination of virus titers revealed that VPA pretreatment resulted in enhanced infectious titers. Incubation of RPE cells with HCMV Hi91 at an MOI of 0.1 resulted in 1.4 × 10²± 0.2 × 10² IEFU/mL. Pretreatment of RPE cells with VPA for 24 hours resulted in 1.1 × 10³± 0.9 × 10² IEFU/mL. 

Prolonged pretreatment periods caused a further increase in HCMV Hi91 replication. The maximum effect was achieved after a 7-day pretreatment. A 14-day pretreatment period did not result in a further enhancement of HCMV IEA1 expression. The number of HCMV IEA1–expressing cells increased by more than twofold after a 7-day incubation compared with the 24-hour pretreatment time (Fig. 2C) . 

To investigate whether VPA also stimulates HCMV antigen expression of other virus strains in RPE cells, the influence of VPA 1 mM on IEA1 expression of the HCMV strains AD169 and Towne was examined. Preincubation for 24 hours induced HCMV strain Towne IEA1 expression of 293% ± 89% and HCMV strain AD169 IEA1 expression of 280% ± 97% compared with the untreated control. A 7-day preincubation resulted in HCMV strain Towne IEA1 expression of 607% ± 78% and HCMV strain AD169 IEA1 expression of 622% ± 105% compared with the untreated control. 

VPA did not affect RPE cell viability, as indicated by the MTT assay, which showed that cell viability was more than 90% in all experiments. 

Influence of VPA on Histone Acetylation in RPE Cells

VPA is an inhibitor of HDAC. ^{1 2} Inhibition of HDAC causes histone hyperacetylation. To investigate the influence of VPA on HDAC, we determined the level of acetylated histone H4 in RPE cells. VPA concentration (Fig. 3A)and time (Fig. 3B)dependently increased the level of acetylated histone H4. The maximum histone acetylation was achieved after a 7-day VPA treatment. 

Influence of TSA on HCMV Replication in RPE Cells

TSA caused histone hyperacetylation (data not shown) and stimulated HCMV strain Hi91 IEA1 expression in a dose-dependent manner (Fig. 4) . The maximum effect of TSA on HCMV replication was achieved after 24 hours of preincubation. Longer incubation times did not further increase HCMV IEA1 expression. A 7-day pretreatment period did not result in increased HCMV IEA1 expression compared with the untreated control. 

Influence of VPA Derivatives on HCMV Antigen Expression in RPE Cells

Side-chain elongation and introduction of a triple bond in position 4 has been shown to increase HCMV replication and histone acetylation in fibroblasts. Derivatives with branched side chains or derivatives with a side chain in position 3 influence neither HCMV replication nor histone acetylation. ⁹ S-2-pentyl-4-pentynoic acid (I) stimulated HCMV strain Hi91 antigen expression and histone acetylation in the tested concentrations ≤1 mM. Pretreatment of RPE cells with 1 mM S-pentyl-4-pentynoic acid (I) resulted in an approximate 10-fold enhancement of IEA1 and LA expression compared with VPA 1 mM after infection with an MOI of 0.1 (Fig. 5A) . Investigation of acetylated histone H4 after a 24-hour treatment with 1 mM S-pentyl-4-pentynoic acid (I) revealed substantially increased levels of acetylated histone H4 compared with treatment with 1 mM VPA (Fig. 5B) . None of the other derivatives, including the R-enantiomer R-2-pentyl-4-pentynoic (II) acid, affected HCMV antigen expression or histone acetylation in concentrations ≤1 mM (data not shown). 

Influence of VAP-Induced Erk 1/2 Phosphorylation on HCMV Replication

Confluent RPE cells were kept in protein-free medium overnight and then incubated with 1 mM VPA for different times. Increased Erk 1/2 phosphorylation was detected as early as 5 minutes after addition of VPA (Fig. 6A) . After 240 minutes Erk 1/2 phosphorylation had declined to basal levels. Erk 1/2 is commonly phosphorylated by MEK 1/2. ^{39 40} The MEK 1/2 inhibitor PD98059 (10 μM) inhibited VPA-induced Erk 1/2 phosphorylation (Fig. 6A) . However, 24-hour pretreatment with the combination of PD98059 and VPA did not result in significantly reduced HCMV IEA1 expression compared with VPA pretreatment (Fig. 6B)indicating that VPA-induced Erk 1/2 phosphorylation does not contribute to enhanced HCMV-replication caused by VPA. Similar results were obtained when PD98059 was added 30 minutes before, during, or after infection. Moreover, 24 hours of pretreatment with VPA was necessary to increase HCMV replication, whereas incubation during or after virus adsorption did not affect HCMV replication. Because Erk 1/2 phosphorylation can be detected as early as 5 minutes after addition of VPA, addition of VPA during or after HCMV adsorption would be expected to increase HCMV replication if VPA-induced Erk 1/2 phosphorylation plays a role in VPA-induced increased HCMV replication. In concordance with these findings, the VPA derivative R-pentyl-4-pentynoic acid induced Erk 1/2 phosphorylation but did not induce HCMV replication or histone H4 acetylation in concentrations ≤1 mM (data not shown). This further confirms that VPA-induced Erk 1/2 phosphorylation does not contribute to VPA-induced increased HCMV replication. In addition, this finding suggests that VPA-induced Erk 1/2 phosphorylation is independent of HDAC inhibition. In concordance, TSA did not induce Erk 1/2 phosphorylation in RPE cells in concentrations up to 90 ng/mL, until 240 minutes after TSA addition. A representative Western blot showing Erk 1/2 phosphorylation in untreated and TSA-treated cells is presented in Figure 7 . 

Discussion

In our study, the results show that VPA increased HCMV replication in RPE cells. Increased virus replication was shown for the clinical HCMV isolate Hi91, ³⁴ by staining for IEA1 and LA as well as by virus-yield assay. IEA1 an LA staining resulted in a fourfold increase in the number of RPE cells expressing HCMV antigens after a 24-hour pretreatment with 1 mM VPA, a therapeutically achievable plasma concentration. ^{37 38} Quantitative real-time RT-PCR revealed an approximate eightfold increase in levels of HCMV immediate early gene 1 and 2 mRNA. This shows that VPA not only increased the number of HCMV-positive RPE cells but also stimulated expression of HCMV immediate early mRNA. Investigation of the HCMV strains AD169 and Towne revealed that pretreatment with VPA also enhances HCMV antigen expression, showing that VPA-induced effects are not limited to one specific HCMV strain. These findings are in concordance with previous studies showing that VPA increases HCMV replication in fibroblasts. ^{9 41}  

Previous results have shown that structurally modified VPA derivatives increase HCMV replication in fibroblasts in strict correlation with their HDAC-inhibiting activity. ⁹ In accordance, VPA caused histone hyperacetylation in RPE cells. Because VPA is known to inhibit HDAC, ^{1 2} the induction of hyperacetylation indicates that VPA also inhibits HDAC in RPE cells in therapeutic concentrations. Investigation of structurally modified VPA derivatives revealed that S-2-pentyl-4-pentynoic acid was the only other VPA derivative tested that induced histone hyperacetylation in the observed concentrations. None of the other derivatives affected histone acetylation or HCMV replication. In addition, the structurally nonrelated HDACI TSA also enhanced HCMV antigen expression. In contrast to VPA pretreatment, which resulted in a maximum effect on histone hyperacetylation and HCMV replication after a 7-day preincubation period, the maximum effect of TSA on HCMV replication in RPE cells was detected after a 24-hour pretreatment period, showing that HDAC inhibition by TSA follows different pharmacokinetic patterns than with inhibition by VPA. 

Although VPA caused increased HCMV replication in fibroblasts as well as in RPE cells, there remain some differences between VPA-induced stimulation of HCMV replication in both cell types. In fibroblasts, the maximum increase of HCMV antigen expression was observed after a 3-day preincubation period with VPA at 1 mM, resulting in a three- to fourfold increase. ⁹ In RPE cells, the maximum effect was seen after a 7-day preincubation with 1 mM VPA, resulting in an approximate ninefold increase. These findings are in concordance with previous investigations showing HCMV replication to be different in RPE cells than in other cell types. ^{11 15 16 20 22 42} In fibroblasts, HCMV infection results in a lytic infection that rapidly destroys all cells. In RPE cells, HCMV causes a smoldering infection characterized by lower HCMV IEA1 expression and subsequent slow viral replication. ^{11 14} Previous examinations had shown that HDAC plays an important role in the regulation of cellular susceptibility to HCMV and that TSA treatment results in the abrogation of HCMV major immediate early promoter repression in cells that are not susceptible to HCMV. ²⁶ Therefore, the higher VPA-induced HCMV stimulation in RPE cells (ninefold) compared with fibroblasts (three- to fourfold) may be explained by the decreased basal susceptibility of RPE cells, possibly due to increased HCMV major immediate early promoter repression compared to fibroblasts. 

VPA induces Erk 1/2 phosphorylation in different cell types, ^{28 29 30 43 44} probably independent of HDAC. ²² This is confirmed by our findings that R-2-pentyl-4-pentynoic acid induced Erk 1/2 phosphorylation, but did not cause histone hyperacetylation in the observed concentrations ≤1 mM, whereas the HDACI TSA did not induce Erk 1/2 phosphorylation. Erk 1/2 phosphorylation plays a crucial role in HCMV replication in RPE cells. ¹⁶ However, VPA-induced Erk 1/2 phosphorylation did not contribute to VPA-induced increased HCMV replication as shown by the use of the MEK inhibitor PD98059. The fact that VPA-induced Erk 1/2 phosphorylation does not contribute to VPA-induced enhanced HCMV replication is further supported by the finding that VPA increased HCMV replication only when added to RPE cells before infection. Addition during or after virus adsorption had no effect, although VPA-induced Erk 1/2 phosphorylation was detected as early as 5 minutes after the addition of VPA. In addition, R-2-pentyl-4-pentynoic acid, which did not cause histone hyperacetylation in the investigated concentrations ≤1 mM but induced Erk 1/2 phosphorylation did not influence HCMV replication. 

The effects of VPA on HCMV replication have not yet been studied in patients. However, a stimulation of HCMV replication by therapeutic VPA concentrations suggests a possible clinical relevance of the present findings. Sporadic HCMV reactivations are controlled by the immune system and do not lead to disease in immunocompetent individuals. ¹⁰ Thus, it is conceivable that VPA-induced increased HCMV replication remains unrecognized in immunocompetent persons. In contrast, VPA-induced HCMV replication could contribute to the development of HCMV disease in immunocompromised patients. A connection between clinical use of VPA and virus replication has already been described for human immunodeficiency virus 1 and human herpes virus 6. ^{45 46} Therefore, the effects of VPA on HCMV disease should be carefully monitored, especially in immunocompromised patients. 

 Supported by the society Hilfe für krebskranke Kinder Frankfurt e.V. and their foundation Frankfurter Stiftung für krebskranke Kinder.

 Submitted for publication March 23, 2005; revised May 20, 2005; accepted July 18, 2005.

 Disclosure: M. Michaelis, None; T. Suhan, None; A. Reinisch, None; A. Reisenauer, None; C. Fleckenstein, None; D. Eikel, None; H. Gümbel, None; H.W. Doerr, None; H. Nau, None; J. Cinatl, Jr, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Chemical structures of the investigated VPA acid derivatives.

Figure 2.

 

Dose- and time-dependent influence of VPA on HCMV strain Hi91 immediate early antigen (IEA1) and late antigen (LA) expression. Cells were infected with an MOI of 0.1. (A) RPE cells were pretreated with and without VPA 24 hours before infection. RPE cells were stained for 72-kDa IEA1 24 hours after infection or for a 67-kDa lLA 72 hours after infection. Histograms show number of cells expressing IEA1 or LA relative to control as detected by immunostaining (mean ± SD of three independent experiments). (B) RPE cells pre-treated with VPA 1 mM for 24 hours and HCMV IEA1 and HCMV IEA2 mRNA expression were determined by real-time PCR relative to HCMV-infected control cells (CTL) C) RPE were pretreated with and without VPA 24 hours, 7 days, or 14 days before infection. Histogram shows number of cell expressing IEA1 (mean ± SD of three independent experiments) as detected by immunostaining. *P ≤ 0.05.

Figure 3.

 

Influence of VPA on histone acetylation in RPE cells. Representative Western blot analysis showing (A) levels of acetylated histone H4 (ac. H4) in RPE cells treated with different VPA concentrations for 24 hours. (B) Levels of ac. H4 in RPE cells treated with VPA 1 mM for 24 hours, 7 days, or 14 days.

Figure 4.

 

Influence of the HDAC inhibitor TSA (TSA) on HCMV immediate early antigen expression. RPE cells were treated with different TSA concentrations for 24 hours and stained for 72-kDa HCMV immediate early protein. *P ≤ 0.05.

Figure 5.

 

Influence of S-2-pentyl-4-pentynoic acid (I) on HCMV antigen expression and histone acetylation. (A) RPE cells were pretreated with and without VPA or S-2-pentyl-4-pentynoic acid (I) 24 hours before infection. RPE cells were stained for 72 kDa IEA1 24 hours after infection or for 67-kDa late antigen (LA) 72 hours after infection. Histograms show number of cells expressing IEA1 or LA relative to control (mean ± SD of three independent experiments). (B) Representative Western blot analysis showing levels of acetylated histone H4 (ac. H4) in RPE cells treated with VPA 1 mM or S-2-pentyl-4-pentynoic acid 1 mM for 24 hours. *P ≤ 0.05, CTL = control.

Figure 6.

 

Influence of VPA with or without the MEK inhibitor PD98059 (PD) on Erk 1/2 phosphorylation and HCMV immediate early antigen expression in RPE cells. (A) RPE cells were kept in protein-free medium overnight. The representative Western blot shows cellular levels of Erk 1/2 and phosphorylated Erk 1/2 (pErk) after 5 minutes treatment without (CTL) or with VPA 1 mM, PD98059 10 μM, or VPA plus PD98059. (B) Number of 72-kDa HCMV immediate early antigen (IEA1)–positive RPE cells after pretreatment without (CTL) or with 10 μM PD98059, 1 mM VPA, or PD98059 +VPA 24 hours before infection.

Figure 7.

 

Influence of the HDACI TSA on Erk 1/2 phosphorylation in RPE cells. RPE cells were kept in protein-free medium overnight. The representative Western blot shows cellular levels of Erk 1/2 and phosphorylated Erk 1/2 (pErk) after 5 minutes of treatment without or with TSA 30, TSA 60, or TSA 90 ng/mL.

The authors thank Gesa Meincke, Cornelia Sippel, Janette Spitznagel, and Lena Stegmann for technical assistance.