Looking for Drugs in All the Wrong Places: Use of GCPII inhibitors outside the brain

Abstract

In tribute to our friend and colleague Michael Robinson, we review his involvement in the identification, characterization and localization of the metallopeptidase glutamate carboxypeptidase II (GCPII), originally called NAALADase. While Mike was characterizing NAALADase in the brain, the protein was independently identified by other laboratories in human prostate where it was termed prostate specific membrane antigen (PSMA) and in the intestines where it was named Folate Hydrolase 1 (FOLH1). It was almost a decade to establish that NAALADase, PSMA, and FOLH1 are encoded by the same gene. The enzyme has emerged as a therapeutic target outside of the brain, with the most notable progress made in the treatment of prostate cancer and inflammatory bowel disease (IBD). PSMA-PET imaging with high affinity ligands is proving useful for the clinical diagnosis and staging of prostate cancer. A molecular radiotherapy based on similar ligands is in trials for metastatic castration-resistant prostate cancer (mCRPC). New PSMA inhibitor prodrugs that preferentially block kidney and salivary gland versus prostate tumor enzyme may improve the clinical safety of this radiotherapy. The wide clinical use of PSMA-PET imaging in prostate cancer has coincidentally led to clinical documentation of GCPII upregulation in a wide variety of tumors and inflammatory diseases, likely associated with angiogenesis. In IBD, expression of the FOLH1 gene that codes for GCPII is strongly upregulated, as is the enzymatic activity in diseased patient biopsies. In animal models of IBD, GCPII inhibitors show substantial efficacy, suggesting potential theranostic use of GCPII ligands for IBD.

Discovery and characterization of GCPII in brain, prostate and gut

Michael Robinson played a pivotal role in identifying, characterizing, and localizing the protein known to biochemists as glutamate carboxypeptidase II (GCPII), known to neuroscientists as NAALADase, known to the cancer world as prostate specific membrane antigen (PSMA), and known to intestinal researchers as folate hydrolase 1 (FOLH1). In this review, we will first detail Mike’s contribution in the early days of GCPII research, then describe the parallel work that identified the same protein in the prostate and gut. Finally, we will provide an overview of where that early work has led: therapeutic development of GCPII inhibitors and ligands outside of the brain for cancer and intestinal inflammation. It’s ironic that decades of work to develop small molecule inhibitors of GCPII to treat CNS disorders of presumed hyper-glutamatergic pathogenesis has actually led to their therapeutic promise outside of the nervous system, hence our title Looking for Drugs in All the Wrong Places.

As postdoctoral fellows in Joseph Coyle’s laboratory at Johns Hopkins, Mike and Randy Blakely characterized the enzymatic activity responsible for the cleavage of N-acetylaspartylglutamate (NAAG) to N-acetylaspartate (NAA) and glutamate [1]. At the time, NAAG had been proposed as an excitatory neurotransmitter as it was produced in the brain and mediated an excitatory effect [2,3], but Mike and Randy’s seminal 1986 publication [1] was the first to provide direct evidence of brain NAAG peptidase activity and liberation of glutamate from NAAG .

Through subsequent publications, the Blakely-Robinson duo pharmacologically characterized this membrane-bound metallopeptidase, which they named NAALADase [4,5]. These first characterizations of brain NAALADase described four-fold higher levels of activity in kidney than in in forebrain, with lower levels in other tissues tested including heart, lung, small intestine and adrenal gland [5]. Because the Coyle lab was focused on understanding the role of NAAG in synaptic transmission, the peripheral distribution of enzymatic activity was noted but not further studied by Mike or others in the lab.

Following his initial work characterizing NAAG hydrolysis, Mike played a key role in the isolation of the NAALADase protein, through his guidance and mentorship of Barbara Slusher, then a pharmacology graduate student in the Coyle lab. Mike and Barbara subsequently generated polyclonal antibodies and confirmed the localization of the protein in the brain and renal proximal tubules [6,7]. Later localization studies were extended to the cellular level and demonstrated expression by non-myelinating Schwann cells in peripheral nerves [8] and astrocytes in the CNS [9]. But again, expression outside of the nervous system was not pursued.

At about the same time that NAALADase was being characterized in the Coyle lab, a novel protein was detected in the LNCaP human prostate cancer cell line by Horoszewicz and colleagues [10]. A few years later, Warren “Skip” Heston’s laboratory at the Cleveland Clinic used Horoszewicz’s 7E-C5 monoclonal antibody to purify the protein from LNCaP cells [11], calling it prostate-specific membrane antigen, or PSMA. After sequencing the purified proteins, they reported that Northern blotting showed no expression outside of the normal and cancerous prostate. A subsequent investigation with Western blotting, however, showed that PSMA was predominantly expressed in the prostate but lower levels were detectable in brain, salivary gland and small intestine [12]. Interestingly, they did not detect the protein in the kidney, even though earlier immunohistochemical reports using the same monoclonal antibody suggested its localization there as well [13]. The intestinal localization made by Heston’s group led to the later characterization of PSMA enzymatic activity as a folate hydrolase, cleaving glutamate from dietary folate polyglutamate to its usable form as folate monoglutamate [14]. At the time, however, no connection to NAALADase was made.

Subsequent immunohistochemistry with improved techniques and more selective antibodies showed a distribution [15,16] that essentially matched the previously reported distribution of NAALADase activity [5]. The expression in human prostate turned out to be species specific; it was later established that dogs, the only other species that spontaneously develops prostate cancer, also expresses PSMA in prostate, with a 5-fold increase in prostate cancer [17]. Eventually, the 7E-C5 monoclonal antibody would be approved for use as a diagnostic for prostate cancer as Prostascint. Unfortunately, since the antibody bound to the intracellular domain of the protein, it was a less than optimal imaging agent for living tissue and targeted therapies that used the antibody for delivery radiotherapy and toxins were not successful [18].

The three threads of the story, PSMA, NAALADase, and FOLH1, only came together once the sequence of the gene coding for the brain enzyme was complete [19-22] and it was finally established that PSMA, NAALADase and FOLH1 were all encoded by the same gene. Still, terminology remained confusing with different names used based on the enzyme’s localization. Part of the confusion in these early years was also due, at least in part, to multiple splice variants and homologs [23,24] as well as a profusion of antibodies with varying affinity between species and tissues. In Figure 1, we show recent immunohistochemical studies from our laboratory that illustrate how dependent localization is depending on the antibody used.

Figure 1.

Throughout the literature there are conflicting reports regarding the tissue expression of PSMA/GCPII protein. Recently, we performed immunohistochemistry using three validated monoclonal PSMA/GCPII antibodies recognizing epitopes in the extracellular domain of PSMA/GCPII: 3E6 (Agilent M362029-2), D7I83 (Cell Signaling 12815) and 1A11 (Kind gift Cyril Barinka). As expected, all three antibodies exhibited strong PSMA/GCPII expression in human prostatic epithelium (brown; A-C). In the kidney however, where PSMA/GCPII is known to be expressed selectively in a subset of renal tubules, striking differences were observed. Strong positive signal in renal tubules was present using 1A11 (stars; D, G), while minimal or no staining was present using 3E6 and D7183 (E,F,H,I). These results illustrate the potential confounding impact of antibody selection on immunohistological outcomes. Scale bars 50 μM in all panels.

In 1992, Slusher joined Zeneca where she and colleagues led the discovery of the first potent GCPII inhibitor, 2-PMPA (IC50=300pM) [25]. Then, in 1999, over a decade after the initial characterization of the brain enzyme, Mike Robinson was a key collaborator with Barbara Slusher and James Vornov on a pivotal study that used 2-PMPA to demonstrate a profound neuroprotective effect following acute ischemic brain injury [26]. This was one of the first steps in establishing the important therapeutic potential of GCPII inhibition in the brain. This report was followed by significant efforts by multiple groups to discover additional GCPII inhibitors, including phosphonic acid-, thiol-, hydroxymate-, and urea-based inhibitors (Figure 2). In parallel with these medicinal chemistry efforts, a number of co-crystal structures were determined for GCPII in complex with various inhibitors [27] providing critical insights into their active site binding modes. For instance, the crystal structures of GCPII in complex with (S)-2-PMPA (PDB code: 2JBJ and 2PVW) revealed that the inhibitor forms a network of interactions with the zinc atoms and key residues in the GCPII active site, resulting in the remarkably high inhibitory potency [28,29]. 2-PMPA and other rationally-designed inhibitors were subsequently utilized to demonstrate the robust therapeutic effects of GCPII inhibition in over 40 animal models of diseases in which hyper-glutamatergic neurotransmission was implicated including pain, drug abuse, epilepsy and cognitive deficits (for review see [30,27]).

Figure 2:

Structures of prototypical GCPII inhibitors and their orally-bioavailable prodrugs.

In the early 2000’s, when both Slusher and Vornov worked at Guilford Pharmaceuticals, they led the early stage clinical testing of a thiol-based GCPII inhibitor termed GPI5693 (2-MPPA) for CNS disorders [31]. Unfortunately, the compound never reached a proof of concept trial in human disease because of immunogenicity findings in GLP animal toxicity studies likely related to the compound’s thiol-containing structure, not its pharmacologic activity. The Slusher laboratory has recently reported efforts to synthesize non-thiol orally bioavailable prodrugs of GCPII inhibitors that could finally allow clinical testing in CNS disorders (Figure 2) [32,33]. They have also reported on intranasal administration of GCPII inhibitors to enhance brain penetration [34,35]. However, as of now, the highly polar nature of the known potent GCPII inhibitors has prevented their clinical development for CNS disorders. In contrast, these inhibitors have emerged with significant diagnostic and therapeutic potential in both cancer and inflammatory diseases.

PSMA Theranostics based on GCPII Binding in Prostate Cancer

Since the original reports of PSMA as a marker for prostate cancer, it has been broadly confirmed that PSMA is highly upregulated (10-100 fold) in the majority of prostate cancer lesions [36,37]. PSMA surface expression correlates with disease progression, androgen independence, and metastasis [36,37]. The association of enhanced PSMA expression to increasing cancer grade defined by Gleason Score has fostered a strong interest in PSMA as a diagnostic and therapeutic target.

In recent years, PSMA imaging with small molecule PET ligands has been shown to be a clinically useful tool for diagnosing and staging prostate cancer [38,39,36,40,41]. As discussed below, these ligands have provided a broader window into changing expression of the enzyme in cancer, inflammation and other pathological processes. Examples of PET imaging using a PSMA-ligand in shown in Figure 3.

Figure 3.

The effect of direct inhibition of PSMA binding by co-administration of 2-PMPA and tracer is shown in mice bearing PSMA -expressing 22Rv1 tumor xenografts. A. In mice given just the tracer, the kidneys, bladder and tumor are clearly shown. B. With the blocking agent, kidney and tumor uptake are reduced. K, kidney; B, bladder, T, tumor. C. Ex vivo quantification of distribution in the model showing direct pharmacologic inhibition results in decrease in normal tissue and tumor (N = 3). Data courtesy of Dr. Daniel Thorek. D. A representative human image of the same tracer in a patient with widely metastatic prostate cancer. The salivary glands and kidney are clearly seen in addition to the tumor uptake. Image courtesy of Dr. Steven Rowe.

In addition to diagnostic utility, PSMA targeted radiotherapy with small molecule ligands has emerged as an extremely promising therapy for metastatic prostate cancer that no longer responds to hormonal therapy, so called metatstatic castration-resistant prostate cancer (mCRPC) [42-44] Because ligand binding to PSMA induces rapid internalization, targeting PSMA actually has the potential to promote accumulation of therapeutic ligands in cancer cells [36,37]. Small molecule PSMA ligands are currently in experimental mCRPC clinical studies including beta therapy (e.g. 177Lu-PSMA-617, 131I-MIP-1095), and alpha therapy (e.g. 225Ac-PSMA-617) [45,46]. Some of these molecules, such as PSMA-617, contain a chelator functional group that enables labeling with isotopes for use as both imaging and therapeutic agents [46,39]. For example, 177Lu-PSMA-617 has been administered to hundreds of late stage mCRPC patients [39,47,48] as a diagnostic. Phase 2 results from a trial of 177Lu-PSMA-617 (ACTRN12615000912583) [49], showed promising results even in a heavily pre-treated patient population who had exhausted benefit from approved therapies [47,48]. Dozens of randomized clinical trials are now registered to confirm the utility and efficacy of PSMA imaging and radiotherapy, with 177Lu-PSMA-617 having recently entered Phase 3 trials in mCRPC ( NCT03511664). These PSMA ligands promise to provide a theranostic approach to mCRPC in which a ligand is used to localize and treat widely metastatic disease in patients with few other options.

Despite significant progress, clinical development of PSMA-targeted agents has been complicated by the physiologic expression of PSMA in normal tissues as would be expected from the history outlined above. In dosimetry studies, exposure of salivary glands and kidney are dose limiting [37,39]. In these clinical studies, PSMA radiotherapeutics have resulted in transient or permanent dry mouth (xerostomia) from salivary gland toxicity, particularly in a recent trial of alpha radionuclides [50,43]. For beta emitters such as 177Lu-PSMA-617, kidney exposure is dose-limiting, restricting treatment of the prostate cancer [51,52,48,53-56,36,37]. While some renal injury might be acceptable in heavily pretreated, late stage patients with widely metastatic disease, the risk of chronic injury prevents use of radiation treatment at early stages of mCRPC [44,49,57].

GCPII inhibitor prodrugs that preferentially shield kidneys and salivary glands

Our recent work has provided evidence that it may be possible to selectively block binding of PSMA-targeted radiotherapy to normal tissue without affecting tumor exposure. An early attempt to mitigate healthy tissue exposure using the highly selective PSMA inhibitor 2-PMPA (IC50=0.3nM) showed that while the dose to the kidneys could be reduced by 83% and nephrotoxicity reduced by 3 months, the exposure also inhibited tumor uptake of the radiotherapeutic by more than 50%. This accelerated tumor growth and significantly reduced overall survival relative to mice that received the radiotherapeutic alone [55]. An example of PSMA block both normal tissue and tumor tracer uptake is shown in Figure 3. Similar results were obtained when 2-PMPA was paired with 131I-MIP-1095 [58]. These results did provide some evidence that tumor and normal tissue could show differential binding, making a shielding approach possible if selectivity could be improved by modification of the shielding ligand to block normal tissue more effectively than the tumor.

We have recently synthesized a tri-alkoxycarbonyloxy alkyl (TrisPOC) prodrug of 2-PMPA (JHU-2545) that appears to preferentially deliver 2-PMPA to non-malignant tissues such as salivary glands and kidneys versus tumor in rodent models. JHU-2545 achieved a 3- and 53-fold enhanced delivery of 2-PMPA to salivary glands and kidneys, respectively, compared to prostate cancer xenografts. We have shown that JHU-2545 can block uptake in rodent salivary glands and kidneys of PSMA targeted radioligands by up to 85% without effect on prostate cancer xenograft uptake and a patent has been published (Slusher et al. PCT/US2018/027106). If similar preferential delivery can be achieved in patients, JHU-2545 or a similar compound could provide a way to bring PSMA-targeted radiotherapy to a wider range of patients with a broader range of beta and alpha emitting radionuclides.

GCPII Inhibitors to Treat Cancer

Since highly specific and potent GCPII inhibitors have been synthesized, an obvious question is whether enzymatic inhibition might also have therapeutic benefit. One idea is that GCPII provides metabolic substrates for cancer growth. It seems possible that PSMA might not only allow increased supply of folate from polyglutamated folate, but also provide a supply of nitrogen, energy and carbon precursors via glutamate and conversion to glutamine [59]. Recently, a correlation between NAAG concentration and cancer grade has been reported [60], with the observation that inhibition of GCPII and glutaminase had synergistic effects on inhibiting tumor growth, supporting the idea that GCPII plays a metabolic role in cancer.

From 1996 to 1999, Slusher and colleagues published several US patents claiming both diagnostic potential and therapeutic utility of GCPII inhibition for cancer (e.g. Slusher et al., US5804602A US5981209A, US6011021A, US6372726B1). Recently, this therapeutic finding was replicated in a publication from Kaittanis and colleagues that provided additional evidence that the somewhat modest effects of GCPII inhibition on tumor growth were due to blockade of glutamate release from a substrate, presumably folate polyglutamate, and subsequent binding to metabotropic glutamate receptors, activating the IP3K/AKT/mTOR pathway [61]. Interestingly, while there are some effects on cell growth from enzyme inhibition, these are much stronger when combined with androgen receptor blockade.

GCPII over-expression in cancer neovasculature

Soon after the original development of the PSMA antibodies, evidence began to accumulate that GCPII is broadly but not universally expressed in tumor vasculature [62,63,16]. For example, in a survey of 275 samples of non-small cell lung cancer, tumor expression was as low as 6% while neovascular expression was found in 49% of cases [64]. Expression was found in 19.4% of a sample of 779 sarcoma cases, with expression increased in more malignant tumors [65].

PSMA enzymatic activity appears to play a role in tumor angiogenesis, as impairments in tumor angiogenesis are observed in PSMA knock out animals and following PSMA inhibitor administration [66]. Recently, a relationship has been established between matrix metalloproteinase 2 (MMP2) and PSMA/GCPII [67]. Sequential proteolysis of laminin (a component of the extracellular matrix) by MMP2 and PSMA/GCPII produced small peptide fragments that increased the rate of endothelial cell migration [68]. This two-step degradation pathway highlights the possible role that PSMA/GCPII activation has on the induction of angiogenesis and metastasis and a potential explanation for its upregulation in cancer neovasculature. However, the impairments in knockout animals and effects of inhibitors do not appear to be powerful enough to make enzyme inhibition a clinically useful chemotherapy, suggesting that the enzyme may be most important as a means of broadly targeting malignancies with toxins or radionuclides as in prostate cancer.

As 68Ga-PSMA-PET imaging has gained favor as a method for staging and restaging of patients with confirmed prostate cancer, binding has been found, as might be expected, in malignant lesions other than prostate such as renal cell carcinoma, breast cancer, bronchial carcinoma and thyroid carcinoma. The literature has recently been reviewed and in general has confirmed that the binding is often due to expression in tumor neovasculature [69]. In Table 1, some examples of tumor uptake are listed to illustrate the potential broad utility of PSMA-targeted PET ligands for imaging tumors over-expressing the enzyme.

Table 1.

Examples of PSMA detection in neoplasms and benign conditions

	Examples	Method of detection
Neoplasm
Breast	Breast cancer[104]	Immunohistochemistry
	Ductal carcinoma[105]	68Ga PSMA-PET/CT
	Lobular carcinoma[106]	Immunohistochemistry
	Triple negative breast cancer[107]	68Ga PSMA-PET/CT
Thyroid	Papillary thyroid carcinoma[108]	68Ga-PSMA-HBED-CC PET/CT
	Follicular thyroid carcinoma[109]	68Ga-PSMA PET/CT
	Differentiate thyroid cancer[110]	68Ga-PSMA-HBED-CC PET/CT
Pancreas	Pancreatic cancer[111]	Immunohistochemistry
Lung	Lung cancer[112]	Immunohistochemistry
Neurogenic tumors	Peripheral nerve sheath tumor[113]	68Ga-PSMA PET/CT
	Paravertebral schwannoma[114]	68Ga-PSMA PET/CT
	MPNST, undifferentiated sarcoma, rhabdomyosarcoma, neovasculature of synovial sarcoma[115]	Immunohistochemistry
Benign
Bone remodeling	Paget Disease[116]	68Ga PSMA-PET/CT
	Sacral Fracture[117]	68Ga PSMA-PET/CT
	Distal radius fracture[118]	68Ga PSMA-PET/MRI
	Fibrous dysplasia[119]	68Ga PSMA-PET/MRI
Bone	Tibia plateau, advance osteoarthritis[120]	68Ga-PSMA PET/CT
Spleen	Splenic Sarcoidosis[121]	68Ga PSMA-PET/CT
Stroke	Subacute stroke[122]	68Ga-PSMA-HBED-CC PET/CT
	Subacute cerebellar infarction[123]	68Ga PSMA-PET/CT
Misc.	Acrochordon (skin tag)[124]	68Ga PSMA-PET/MRI
	Neovasculature of nonneoplastic regenerative and reparative tissues[125]	Immunohistochemistry
	Desmoid tumor[126]	68Ga-PSMA-HBED-CC PET/CT
	Cervical, Celiac, and Sacral Ganglia[127]	68Ga-PSMA-HBED-CC PET/CT
	Intramuscular myxoma[128]	68Ga PSMA-PET/CT

These findings have recently led to the first examples of PSMA being targeted for the treatment of malignancies beyond prostate. In one report, 177Lu-PSMA-617 was administered to a patient with PSMA-positive metastasized leiomyosarcoma [70]. In another Phase II study, the PSMA-targeted cytotoxin conjugate mipsagargin (G-202) resulted in disease stabilization in late stage hepatocellular carcinoma, mediated in part by reducing blood flow to lesions [71].

It should be noted that expression is not limited to tumor neovasculature. In some cases, expression is found in tumors derived from tissues that normally express the enzyme. For example, GCPII is normally expressed in astrocytes [9] and is upregulated in astrocytic tumors. As in prostate cancer, the increased expression is correlated with astrocytic tumor grade [72], In the case of glioblastoma, this has clinically been confirmed with PET imaging [73,74]. Peripheral nervous system structures such as the ganglia of the sympathetic trunk can also be seen on PET images [75], which again is to be expected based on histological findings that GCPII is highly expressed on non-myelinating Schwann cells [76,77] and is increased in tumors derived from Schwann cells such as malignant peripheral nerve sheath tumors [65]. Similarly, cancer cells arising from non-nervous system organs that express GCPII will often also express the enzyme, as reported for renal cell carcinoma [78]. PSMA theranostics may be particularly useful for these tumor types as the enzyme is selectively expressed by malignant cells.

GCPII inhibitors To Treat Inflammatory Bowel Disease

The second human disease outside of the nervous system established to show over-expression of FOLH1 was inflammatory bowel disease. As was the case in prostate cancer, where the link was established through a human cell line, the role of GCPII in IBD was first suggested by study of the human disease, this time by a genome-wide expression investigation [79].

As noted, after PSMA was identified as the gastrointestinal tract folate hydrolase, GCPII’s role in folate absorption was described. GCPII is expressed by epithelial cells of the proximal small intestine [78,16], where it functions to sequentially cleave γ-linked glutamate residues from dietary polyglutamyl folates [14]. This deconjugation to a monoglutamic form is required for folate absorption. Reduced processing of polyglutamyl folates is known to reduce folate bioavailability [80-82]. A naturally occurring mutation in GCPII has been reported, H457Y [83], which correlates with circulating concentrations of folate and its downstream metabolite homocysteine in population studies [83-90]. A detailed mechanistic study highlights the complexity of this relationship, demonstrating that at a structural level the H457Y mutation has no effect on polyglutamyl folate binding or enzyme kinetics, suggesting that an in vivo effect of H475Y polymorphism on folate status is likely to be indirect. [91].

With regard to human disease, multiple independent genome-wide screens have now identified that FOLH1 expression is significantly upregulated in the affected intestinal mucosa of IBD patients [92,93,16,79], where FOLH1 has been described to function as a “hub” gene with significant correlations to over a dozen known IBD gene biomarkers [79]. Correspondingly, it has been validated that GCPII enzymatic activity is consistently and robustly increased in both Crohn’s disease and ulcerative colitis diseased patient biopsies by 300-3000% [94] and that pharmacological inhibition of GCPII ameliorates clinical signs in mouse models [95,94].

GCPII Expression in Other Inflammatory Diseases.

One of the novel observations from the wide use of PSMA PET ligand use in prostate cancer is the expression in benign inflammatory states, including anal fistula [96], sarcoidosis [97], fasciitis [98], and cerebral infarction [99, 100]. PSMA ligand uptake has also been incidentally observed in areas of bone and joint remodeling, such as in Paget disease [101], fractures [102], and synovitis [103], in patients being imaged for their prostate cancer. Table 1 includes some examples of these intriguing incidental findings of increased enzyme expression.

Interestingly, there are no reports of imaging in patients with inflammatory bowel disease to date. There are intriguing suggestions that mGluRs and PI3K/Akt may serve as common active pathogenic signaling systems activated in cancer, inflammatory states and angiogenesis, all influenced by expression of PSMA/GCPII. As the roles of these signaling systems are understood in the pathological states, new approaches to targeted therapy may be possible that address multiple processes across a wide range of diseases, with PSMA PET imaging serving as a means to identify appropriate patients and perhaps useful as a way to assess response to therapy.

Conclusion

Since the first characterization of GCPII enzymatic activity by Mike Robinson and collaborators in the Coyle lab, much has been learned about the enzyme and its function in pathologic conditions in the brain, cancer and inflammatory diseases. The potential therapeutic utility of GCPII is currently most promising using radiochemical ligands as theranostics in cancer and enzyme inhibitors to treat inflammatory bowel disease. While work continues to develop potent brain penetrable GCPII inhibitors, the work started with Mike Robinson many years ago unexpectedly shows the greatest promise to provide treatments for important unmet medical needs outside the nervous system.