Tumor-specific CD4⁺ melanoma tumor-infiltrating lymphocytes

Abstract

Adoptive cell therapy using tumor infiltrating lymphocytes (TIL) can mediate objective and durable tumor regressions in patients with metastatic melanoma. CD8⁺ tumor reactive TIL are well studied in humans and animals yet the function of tumor infiltrating CD4⁺ T lymphocytes patient treatments remains controversial. We recently demonstrated that CD4⁺ TIL are not necessary for objective responses in patients. Co-infusion with tumor-specific CD4 TIL may enhance or increase the durability of tumor regressions but the number of patients with tumor-reactive CD4 TIL is unknown. We screened forty-four CD8⁺ depleted TIL for in-vitro reactivity against autologous tumor. Nine (20%) showed specific reactivity by interferon-gamma (IFNγ) release assay of which eight were specifically blocked by an anti-HLA-DR antibody. Flow-cytometric analysis of these reactive TIL confirmed a high CD4⁺ composition (median 89%). Highlighting the contribution of CD4⁺ TIL to tumor regression, a patient with widespread metastatic disease was administered TIL containing HLA class II-restricted tumor activity with high dose interleukin-2 therapy following lymphodepletion that mediated regression of extensive metastatic disease in the liver and spleen. These results demonstrate that at least 20% of metastatic melanomas contain CD4⁺ lymphocytes with specific tumor recognition and suggest a possible role for CD4⁺ cells in the effectiveness of adoptive cell therapy.

Introduction

Malignant melanoma is the sixth-most common form of cancer in the United States ¹. Despite comprising less than 5% of all skin cancers it remains responsible for 79% of the deaths attributed to this malignancy. Furthermore the incidence of melanoma continues to rise 4–6% annually ². Until recently the only Food & Drug Administration (FDA) approved treatments included dacarbazine based chemotherapies and high does interleukin-2 (IL-2). Recently approved FDA therapies, including ipilimumab (Yervoy^®) and vemurafenib (Zelboraf^®), have achieved clinical responses with significant impact on overall survival ^{3, 4}. IL-2 and anti-CTLA-4 can mediate durable objective responses in 5–10 percent of metastatic melanoma patients and encourages further exploration of immune-based therapies^{5, 6}. Adoptive cell transfer (ACT) is an alternative, albeit still experimental, means to mediate tumor regressions ^{7, 8}. Some recent progress has been made using genetically engineered lymphocytes for ACT in the setting of metastatic melanoma and other tumor histologies ^9–13. Much more experience is available from the study of tumor infiltrating lymphocyte (TIL) ACT ^14–17. TIL based immunotherapy in the NCI Surgery Branch can mediate durable complete response (CR) rates as high as 40% in metastatic melanoma, with many patients experiencing ongoing regressions at 5 years ¹⁸.

The cellular composition of the infused T cells may impact tumor regressions. In one study, the number of CD8⁺ T cells in unselected “young” TIL was correlated with patient response ¹⁴ and in another study the fraction of CD4⁺ T cells was inversely correlated with response in patients receiving selected TIL ¹⁹. This led to the optimization of a protocol for the enrichment of CD8⁺ lymphocytes from bulk TIL ¹⁹, and implementation of a clinical protocol to test their efficacy. We recently reported observing a 49% objective response rate according to Response Criteria in Solid Tumor (RECIST) standards, to ACT with CD8⁺ enriched TIL ¹⁵. However, the role of CD4⁺ cells in tumor immunity remains poorly understood. The function and immunologic impact of a CD4⁺ T cell depends on its phenotype and specificity. Recent evidence suggests T-helper (Th) subsets of lymphocytes may prove beneficial in tumor surveillance. The administration of tumor reactive CD4⁺ Th17 cells can cause the complete regression of a transplanted B16 tumor in syngeneic C57Bl/6 mice ²⁰, while the administration of CD4⁺CD25⁺ regulatory T cells have the opposite effect and abrogate the therapeutic efficacy of transferred tumor reactive CD8⁺ cells ²¹. Human CD4⁺ lymphocytes exhibit numerous fates and they have a complexity of functions in tumor immunity similar to mouse CD4⁺ cells. Hunder et al report of a case study of a patient treated with a CD4⁺ T cell clone specific for the cancer testis antigen NY-ESO-1 showing the capability of CD4⁺ T cells to mediate potent long-term anti tumor function in a patient with metastatic melanoma ²². In contrast, other reports suggest a correlation between regulatory T cell infiltration in tumors and poor clinical prognosis ^23–26. These data in mouse and man suggest a complex and patient-specific interaction between CD4⁺ T cells and growing tumors.

CD4 T cells recognizing peptides restricted by class II human leukocyte antigens (HLA) are contained in TIL from melanoma, breast, and ovarian tissues ^27–29. The frequency of patients with HLA class II-restricted tumor-reactive TIL and whether they can restrict tumor growth in vivo is unknown. We developed a unique set of reagents to examine whether a tumor reactive CD4⁺ cell population could be identified and characterized from human melanoma tumor-infiltrating lymphocytes. These bulk TIL were generated and expanded without any prior knowledge of their tumor recognition. They provided an opportunity to examine the unbiased frequency of TIL cultures containing tumor-reactive CD4⁺ T cells. We discovered that a substantial portion of these samples contained in vitro anti-tumor function as assayed by IFNγ release after co-culture with autologous tumor targets. In this report we present the detailed analysis of these CD4⁺ TIL, as well as anecdotal evidence that CD4 TIL can recognize HLA class II restricted tumor antigens in vivo. A predominantly CD4⁺ TIL was administered to a patient and caused a dramatic tumor regression followed by progression of HLA class II antigen-negative disease.

MATERIALS & METHODS

Established melanoma cell lines & fresh tumor digests

Fresh tumor (FrTu) targets for cocultures were generated by washing then immediately cryopreserving single cell suspensions of resected melanoma tumors after overnight enzymatic digestion in enzyme media (RPMI-1640 (Lonza) supplemented with 100 U/mL penicillin G (Lonza), 100 μg/mL streptomycin (Lonza), 10 μg/mL gentamicin (Lonza), Collagenase, type IV, 1g/liter stock enzyme medium (Sigma-Aldrich), and Pulmozyme (Dornase Alfa Pulmozyme Inhalation Solution, Genentech, Inc.). FrTu were thawed and used immediately in TIL stimulation “coculture” assays. Patient tumor cell lines (TC) were generated from fresh tumors by mechanical dissociation after culture in T2 media (RPMI-1640 supplemented with 10% fetal bovine serum (Hyclone), 25mM HEPES (Lonza), 100 U/mL/100 μg/mL penicillin/streptomycin (Lonza), 2 mM L-glutamine (MediaTech), 10 μg/mL gentamicin (Lonza), 1.25 μg/mL amphotericin B (XGen Pharmaceuticals), and 10 μg/mL ciprofloxacin (Bedford laboratories)) in a humidified 37°C incubator with 5% CO2. Melanoma cell line 624 was generated from a human leukocyte antigen (HLA) -A*02,03, HLA-DR*0401,0701 patient. Melanoma cell line 624-CIITA was generated after retroviral transduction of the class II transactivator (CIITA) and florescent-activated cell sorting after staining with an anti-HLA-DR antibody (BD biosciences). HLA expression was also induced by treatment for 48 hours with recombinant human IFNγ (1000u/ml).

Generation and Fractionation of TIL

The schema used to generate bulk and CD8⁺-depleted TIL available for treatment and experimental evaluations in this study is shown in Figure 1. Limited tissue acquisition restricted the samples available for research use. All patients signed an informed consent approved by the Institutional Review Board of the National Cancer Institute. TIL were resected from a metastatic deposit for initiation of minimally cultured TIL as previously described ^{15, 30}. Briefly, TIL cultures were processed to single cell suspensions by enzymatic digestion and plated for growth at 5 × 10⁵ viable cells/ml. TIL cultures were initiated in 2ml wells in complete medium (CM, RPMI-1640 supplemented with 10% Human AB serum (Valley Biotech or Gemini, Inc), 25mM HEPES (Lonza), 100 U/mL/100 μg/mL penicillin/streptomycin (Lonza), 2 mM L-glutamine (MediaTech), 10 μg/mL gentamicin (Lonza), and 5.5 × 10⁻⁵M 2-mercaptoethanol) supplemented with IL2 (6000 IU/ml, Chiron Corp., Emeryville, CA) in a humidified 37°C incubator with 5% CO2. Five days after initiation, one half of the media was aspirated from the wells and replaced with fresh CM and IL-2, and media was replaced every two to three days thereafter as needed. Wells were inspected visually for TIL growth, and when lymphocytes became confluent in the well with elimination of all adherent cells, the cultures were expanded (if needed) by splitting the wells to maintain lymphocyte density between 0.8–1.5 × 10⁶ cells/ml. When TIL culture reached 1–3 × 10⁸ total lymphocytes (typically 16–24 days) the TIL were cryopreserved pending coordination of clinical and laboratory aspects of treatment protocols.

Figure 1. CD4⁺ Tumor infiltrating lymphocytes (TIL) originate from a resected metastatic lesion via CD8⁺ depletion and in vitro culture.

Tumor specimens were obtained and processed under sterile conditions to single cell suspensions by enzymatic digestion. The first priority for tissue use was to generate a TIL culture for protocol treat; when available, some of the autologous fresh tumor (FrTu) single cell suspension was cryopreserved immediately for use later as “targets” in coculture assays. The single cell suspension was plated in lymphocyte growth media and maintained in vitro until a “bulk” young tumor infiltrating lymphocyte cultures (bulk TIL) was established. The majority of the bulk TIL underwent CD8⁺ enrichment using a CliniMACS^(R) magnetic cell separator. All three fractions, the bulk TIL, the CD8⁺ enriched fraction, and the CD8⁺-depleted fraction (containing CD4⁺ cells) were rapidly expanded with OKT3 (anti-CD3), IL-2 and feeder cells. The resulting TIL cultures were evaluated for phenotype and tumor recognition.

Some TIL were subjected to CD8⁺ enrichment for rapid expansion and patient therapy. These TIL samples analyzed in our survey represented 44 consecutive TIL cultures initiated for patients reported previously ¹⁵. Bulk TIL and CD8⁺-depleted TIL were obtained as a byproduct of the CD8⁺ enrichment procedure ¹⁹. Briefly, CD8⁺ (CD4⁻) tumor infiltrating lymphocytes where selected by positive selection with immunomagentic labeling accomplished on a CliniMACS® apparatus with a CD8⁺ reagent (Miltenyi). Three TIL populations were available after CliniMACS® selection: CD8⁺-enriched, CD8⁺-depleted (flow through), and bulk (unfractionated) TIL. All TIL populations were resuspended at a concentration of 1x10⁶ cells/ml in CM supplemented with 6000 IU/mL IL-2 in a humidified 37°C incubator with 5% CO₂ and rested overnight or used immediately.

Rapid expansions of TIL cultures was performed using the Rapid Expansion Protocol (REP) as previously described ^{31, 32.} Briefly, TIL cells were cultured in T175 flasks with a 200 fold excess of irradiated (40 Gy) allogeneic peripheral blood mononuclear “feeder” cells in a 1:1 mixture of CM and AIM V (Invitrogen) with 30 ng/ml anti-CD3 antibody and 3000 IU/ml IL-2. Half of the media was exchanged on day 5, and cells were split with AIM V supplemented with 3000 IU/ml IL-2 (without human serum) as needed thereafter. After 14 days, all TIL fractions were evaluated for expansion, phenotype and specificity. Lymphocyte expansion of all cultures was observed to be 200–2500 fold, and was typically around 1400 fold.

TIL line DMF5 has been described previously ^{16, 33}. The DMF5 TIL line is dominated by a single HLA-A2 restricted, MART-1:26–35 reactive T cell clone that is CD8⁺ dependent ³⁴, and this was used as the class I positive control. TIL E11 was used as a class II restricted positive control. E11 recognizes the HLA-DR*07-restricted epitope gp100170-190 (data not shown). Inhibition of interaction with HLA class I or HLA-DR was performed by pre-incubation of the tumor cells with 100 ug/mL of either W6/32 (pan-HLA-class I) or L243 (HLA-DR-specific) blocking antibodies, respectively, for thirty minutes prior to the addition of TIL.

Flow cytometric analysis

Phenotype was determined by six-color flow cytometric analysis using FACS Canto II Flow Cytometer (BD Bioscience, San Jose, CA). Antibodies for analysis included CD3⁺, CD4⁺, CD8⁺, CD56⁺, andW6/32 (Class I), and HLA-DR (BD Biosciences). At least 20,000 events were captured for each analysis.

IFNγ release

Specificity assays were performed by overnight coculture in 96-well flat bottom plates of 1 × 10⁵ responder TIL with 0.5–1 × 10⁵ autologous, HLA-matched, or HLA-mismatched tumor cells. IFNγ was quantified in supernatant by ELISA according to manufacturer’s guidelines (Pierce). For each enrichment, cocultures included bulk (unfractionated) and CD8⁺ -depleted (CD4⁺) populations. Specific reactivity was defined as IFNγ release of at least 200 pg/ml and greater than twice the background (release of HLA-mismatched controls or no tumor controls). Assays were performed as initial screens in one-well format and no statistics are present for the data shown. The experiments were replicated in independent assays unless specifically stated.

Patient Treatment and Response Evaluation

Patient 2950 was treated twice as part of an IRB approved clinical protocol in the Surgery Branch, NCI. Prior to each TIL infusion she received a non-myeloablative lymphodepleting regimen consisting of cyclophosphamide (60 mg/kg/d for 2 days) and fludarabine (25 mg/m2/d for the following 5 days). After the cell infusion, the patient received bolus interleukin-2 (720,000 IU/kg every 8 hr to tolerance). The patient’s response was assessed using standard radiographic studies at 4 weeks after cell administration and at regular intervals thereafter based on response evaluation in solid tumors (RECIST) standards.

RESULTS

CD8⁺-depleted TIL can recognize autologous tumor

Prior efforts to screen active TIL cultures for use in adoptive cell transfer (ACT) protocols often sought to identify lymphocytes that specifically recognized human leukocyte antigen (HLA) class I matched tumor activity. A disadvantage of this method was its inability to detect HLA class II tumor-reactive CD4⁺ T cells. For this reason, the frequency of TIL cultures containing tumor-reactive class II-restricted T cells has not been established. We tested minimally-cultured TIL samples for which an autologous tumor target was available, resulting in 44 unique patient TIL-tumor combinations for testing. Either enzymatically dissociated fresh tumor (FrTu) or tumor cell lines (TC) were used as stimulators. All patient TIL were CD8⁺-depleted, then rapidly expanded (see Materials and Methods) and tested by cytokine release for tumor recognition. Cultures with tumor-reactive CD4 TIL were identified as exhibiting increased IFN-γ secretion concurrent with CD4 T cell enrichment. Nine of the 44 CD8⁺-depleted TIL (20%) released similar or greater amounts of IFNγ compared to the bulk TIL (Table 1) suggesting that these TIL contained tumor-reactive Th1 CD4⁺ T cells. FACS on the nine reactive CD8⁺-depleted TIL fractions confirmed a high CD3⁺CD4⁺ phenotype (median 88.6 %, Table 1).

Table 1.

Recognition of autologous tumor, and CD4⁺ composition of depleted fraction.

Patient1		IFNγ (pg/ml)			CD4 (%)2
		none	938	Auto
2546	Bulk TIL	120	136	974
	CD8 depleted	91	259	13610	89
2950	Bulk TIL	29	35	2810
	CD8 depleted	32	46	3660	72
3108	Bulk TIL	41	78	4460
	CD8 depleted	81	49	22061	25
3146	Bulk TIL	64	184	2700
	CD8 depleted	236	131	12960	95
3269	Bulk TIL	37	15	2060
	CD8 depleted	52	63	6320	97
3277	Bulk TIL	11	60	2700
	CD8 depleted	27	89	3611	90
3334	Bulk TIL	30	49	1520
	CD8 depleted	36	334	2600	69
3107	Bulk TIL	45	24	>33761
	CD8 depleted	34	239	>31130	96
3117	Bulk TIL	39	48	6650
	CD8 depleted	47	93	7090	75

¹IFNγ release TIL is shown after stimulation with no tumor (none), melanoma cell line 938, or autologous tumor targets. Values in bold are > 200 pg/ml and greater than 2x background. Each patient had both a bulk TIL and a CD8⁺ depleted fraction tested against all targets.

²The percentage of total viable cells in the CD8⁺ depleted fraction that were CD3⁺CD4⁺ as measured by flow cytometry is shown.

CD8⁺-depleted TIL are predominantly CD4⁺ lymphocytes with HLA class II restricted reactivity

We next investigated whether tumor specific reactivity could be blocked with either the addition of a class II-DR (L243) and/or pan-class I (W6/32) antibody. Autologous tumor targets (FrTu or TC) were not available for one patient, but were available for eight of the nine patients with autologous CD4⁺ TIL reactivity. All eight (100%) of these had reactivity that was blocked by the addition of a class II-DR antibody but not with a pan-class I antibody confirming HLA-class II-restricted anti-tumor function (Figure 2). As controls we examined the specific ability of W6/32 (anti-Class I) block the HLA class I-restricted anti-tumor function of TIL DM5 (HLA-A2/MART-1:27–35) and the specific ability of L243 (anti-HLA-DR) to block the HLA class II-restricted anti-tumor function of E11 (HLA-DRb4*01/gp100:44–59).

Figure 2. Eight patients with autologous tumor activity in the CD4⁺ TIL fraction demonstrated HLA-DR-restricted anti-tumor recogniton.

Eight CD8⁺-depleted TIL with autologous anti-tumor recognition were co-cultured overnight with autologous tumor alone, or with autologous tumor in the presence of a pan-HLA class I (W6/32) blocking antibody, or an HLA-DR (L243) blocking antibody. IFNγ release was assayed by ELISA. Inhibition of IFNγ release is presented relative to the absence of antibodies. The specificity of antibody blocking was confirmed by inhibition of IFNγ release from TIL DMF5 (HLA-A2/MART1:27–35) and TIL E11 (HLA-DRB07/gp100:170–190) after co-culture with the melanoma cell line 624-CIITA. One Autologous reactive TIL could not be tested due to insufficient target sample.

The phenotype of the CD4+ enriched, tumor-reactive TIL was examined by flow analysis, and compared to non-tumor-reactive CD4+ TIL. Six tumor-reactive CD4⁺ TIL were analyzed and four non-reactive TIL were evaluated; all samples were tested after a single rapid expansion. All samples were stained for CD25, CD62L, CD45RA, and PD-1 (CD279) to determine if any obvious differences in activation, differentiation and/or exhaustion could account for the tumor reactive phenotype. As typically observed in highly cultured, rapidly expanded CD4+ lymphocytes, CD45RA and CD62L expression was very low, and CD25 and PD-1 were variable, but there was no correlation between expression and tumor reactivity.

A CD4⁺ tumor-reactive TIL mediated dramatic tumor regression via ACT

An opportunity to better understand the anti-tumor function of CD4⁺ T cells in vivo came when we treated a patient with a TIL containing predominantly tumor-reactive HLA-DR-restricted T cells (TIL 2950). The patient presented at the Surgery Branch with advanced metastatic melanoma which had not responded to several prior treatments including high-dose IL-2. Her disease rapidly progressed during the time necessary to generate a minimally cultured TIL from a metastatic liver deposit, resulting in extensive hepatic and splenic tumor burden at the time of treatment (Figure 3, left vs middle panel). She was treated with bulk TIL 2950 that contained predominantly HLA-DR-restricted anti-tumor function (Figure 2). Her extensive liver and splenic tumor burden decreased substantially after treatment with the HLA-DR-restricted tumor-reactive TIL (Figure 3, right panel). While the patient’s disease regressed, her serum laboratory abnormalities resolved and her performance status improved. However, the patient did not meet criteria for an objective response by RECIST due to the appearance of a new, sub-centimeter lesion in her adrenal gland at two months after TIL infusion. At three months after her TIL infusion, a bulky recurrent pelvic mass was detected.

Figure 3. Clinical response after ACT with predominantly CD4⁺ TIL.

Composite MRI images 6 weeks before treatment (left) at the time of tumor resection for TIL generation, immediately prior to initiation of chemotherapy for ACT (middle) demonstrating pace of tumor progression in the liver, and 6 months after infusion of predominantly CD4⁺ TIL (right) that demonstrated HLA-DR restricted autologous tumor recognition.

To further investigate the tumor recognition of TIL 2950, multiple HLA mis-matched targets were evaluated for their ability to stimulate IFNγ production and three HLA mis-matched TIL were tested with 2950 FrTu target (Figure 4a). These controls demonstrated the variability of IFNγ release when using FrTu targets. The assay variability is presented graphically by plotting the average control value (black bar) and the standard deviation of the controls. The TIL 2950 was specifically blocked by the L243 anti-Class II antibody, and not blocked (within the assay variability) by the anti-Class I W6/32 anti-body (Figure 4a). Fractionation of TIL 2950 into CD8⁺-enriched and CD8⁺-depleted components demonstrated that all of the tumor recognition was contained in the CD8⁺-depleted (CD4⁺) fraction (Figure 4b). These data further support the conclusion that recognition of tumor by the HLA-DR-restricted CD4⁺ TIL mediated the transient reduction of most of the patient’s tumor burden.

Figure 4. The treatment TIL exhibited autologous tumor recognition restricted by HLA class II that was enriched in CD4⁺ lymphocytes.

IFNγ release was assayed by coculture and ELISA as described in Figure 2. Six allogeneic, HLA d mismatched FrTu were used as targets (left of figure) but failed to stimulate substantial IFNγ release. Three HLA-mismatched allogeneic TIL were cocultured with the 2950 FrTu but did not cause IFNγ release, demonstrating that the 2950 FrTu was not non-specifically stimulatory. The horizontal line at 200 pg/ml indicates background. The grey shaded area indicates one standard deviation of the non-specific values around the background. Autologous tumor was used with no antibody blocking, and with W6/32 (anti-Class I), L243 (anti-HLA-DR), or both W6/32 and L243. L243 significantly blocked tumor recognition by the autologous the autologous TIL. The specificity of antibody blocking was confirmed by inhibition of IFNγ release from TIL DMF5 (HLA- A2/MART1:27–35) and TIL E11 (HLA-DRB07/gp100:170–190) after co-culture with the melanoma cell line 624-CIITA.

A recurrent lesion failed to express the antigen recognized by the CD4⁺ TIL

The single progressive pelvic site was surgically removed and a tumor cell line was established from it. FACS analysis revealed that the pelvic tumor maintained HLA class I expression but lacked expression of HLA-DR (Figure 5a). After retroviral transduction of the pelvic tumor line with CIITA (class II transactivator) expression of HLA-DR was restored. Cytokine release assays were used to investigate TIL recognition of these TC. CD4⁺ and bulk TIL recognized CIITA transduced HLA-DR⁺ TC, but not the HLA-DR- TC (Table 2). These results suggest that without HLA-DR expression, the first treatment was not capable of recognizing the pelvic tumor. As an important control, the CD8⁺ enriched fraction from the original TIL never recognized any tumor, including the CIITA transduced pelvic TC, strongly suggesting that all the initial anti-tumor and therapeutic activity was in the CD4⁺ TIL subset. We investigated the induction of HLA-DR on the pelvic tumor cell line by IFNγ (Figure 5b). Treatment of melanoma tumor cell lines with IFNγ often increases HLA class II expression as was seen melanoma cell line 624. However, IFNγ treatment did not appreciably increase HLA-DR expression on the pelvic tumor line. This observation is consistent with the hypothesis that the patient’s pelvic tumor escaped recognition by the tumor reactive TIL cells by lack of HLA-DR expression in vivo.

Figure 5. HLA class II was absent and was not inducible by IFNγ on the melanoma cell line derived from the recurrent pelvic tumor mass.

A) HLA-DR was not expressed by the melanoma cell line derived from the recurrent pelvic mass while HLA class I (W6/32) was expressed (left). After forced expression by retroviral transduction with CIITA, HLA class II was observed while HLA class I was unchanged (right). B) The cell line 624 (upper) and the pelvic tumor line (lower) responded differently to IFNγ induction. HLA class I was expressed at all concentrations of IFNγ for both cell lines (left). HLA class II was induced by IFNγ in a dose dependent manner in the 624 cell line while no class II expression was induced at any IFNγ dose in the pelvic tumor line (right)

Table 2.

Autologous tumor reactive TIL was identified only in the CD8⁺-depleted (CD4⁺) fraction of TIL 2950, but both CD8⁺-enriched and CD8⁺-depleted fractions of TIL from the pelvic recurrence demonstrated autologous tumor activity.

	Controls1		1st TIL (2950)			2nd TIL (Pelvic recurrence)
	None	DMF5	Bulk (Rx1)	CD8+ depleted	CD8+ enriched	Bulk	CD8+ depleted	CD8+ enriched (Rx2)
None	42	135	115	134	56	153	112	58
624mel	29	13590	87	65	48	49	65	36
Pelvic TC	34	133	113	76	154	118	111	6160
Pelvic TC	51	100	623	857	154	1050	709	6280
- CIITA

¹IFNγ release (pg/ml) was assayed by ELISA after overnight coculture TIL and melanoma cell lines. Examined are TIL from the patient’s original liver tumor (TIL 2950) or the pelvic tumor (2^nd TIL) before (bulk) or after CD8⁺ magnetic separation. The DMF5 TIL (HLA-A2/MART-17–35) is included as a control. Positive recognition (bold and underlined) is defined as values greater than 200 pg/ml and more than twice background.

Following tumor-reactive Th1 CD4⁺ TIL infusion, new CD8⁺ tumor-reactive TIL were isolated and used for ACT

TIL were generated from the pelvic tumor and were administered to the patient as a second treatment for progressive disease, including several new liver lesions. The second treatment consisted of CD8⁺ enriched TIL that were reactive with the pelvic TC (Table 2). The recovery of tumor-reactive CD8⁺ lymphocytes is an interesting observation and may suggest that the first tumor reactive CD4⁺ TIL therapy acted to induced de novo immune reactivities.

Discussion

CD4⁺ T lymphocytes express diverse fates and perform an array of functions in the immune system, and this variety is reflected in a large diversity of tumor-lymphocyte interactions. Preclinical models have described regulatory CD4⁺ T cells associated with tumor progression as well as inflammatory CD4⁺ T cells that mediate tumor regression. Similarly in humans, T regulatory cell frequencies in tumors have been correlated with poor prognosis, while anecdotal cases of administration of inflammatory CD4⁺ T cells have resulted in anti-tumor responses. These diverse and conflicting reports of CD4⁺ T cell roles in tumor immunology complicate immunotherapy efforts targeting CD4⁺ T cells. To directly assess the potential for CD4⁺ T cell therapy in ACT of melanoma, we undertook a survey of the frequency of endogenous tumor-reactive CD4⁺ T cells in melanoma TIL. All CD4⁺ TIL evaluated in this study were derived from consecutive patients with appropriate materials, and the lymphocytes were grown using standardized procedures to improve the reliability and consistency of the analysis. A bulk TIL and a CD8⁺-depleted TIL (containing the CD4⁺ lymphocytes) were expanded in parallel and only patients with both of these TIL populations as well as autologous tumor targets (cryopreserved single cell enzymatic tumor digests or a tumor cell line) were evaluated. All TIL were tested for tumor recognition by evaluating the secretion of IFNγ a Th1 cytokine. Using these criteria, we demonstrated that 9 of 44 samples, or 20.5%, exhibited specific tumor recognition that was HLA-DR restricted and present in the CD4⁺ enriched fraction. Further phenotypic analysis of these TIL did not elucidate factors that identify tumor-reactive CD4 cultures and is consistent with prior TIL studies that found that extensive culture minimizes phenotypic differences. Cultured TIL consistently present phenotypes associated with T cell activation, differentiation, and contain a paucity of CD4 T regulatory cells.

Other examples of tumor reactive CD4⁺ TIL have been reported. CD4⁺ T cells, like CD8⁺ T cells, may recognize “shared” antigens expressed by multiple HLA-matched tumors ²⁹. Alternately, TIL recognition of “unique” antigens encoded by mutations specific to individual tumors have been described ^35–37. Autologous tumor represents a clinically relevant target, expressing both shared antigens and mutated antigens uniquely expressed by the patient’s own tumor. However, autologous tumor targets in coculture assays can have variable background, and may under-represent the true frequency of tumor-reactive CD4⁺ TIL due to poor HLA class II expression or tumor antigen presentation in the experimental assays. Moreover, we chose to survey tumor reactive CD4 by tallying IFNγ which has been identified in mouse models to have both direct and indirect effects on tumor regressions ³⁸. Yet, using IFNγ release to identify tumor-reactive TIL omits CD4 TIL which exclusively secrete IL-17 or IL-2. Thus, 20% represents a minimum estimate of patients who have tumor reactive CD4⁺ lymphocytes in their TIL. These results suggest that a relatively high frequency of patients have TIL with tumor reactive CD4⁺ lymphocytes, and emphasize the possibility for exploiting these tumor reactive CD4⁺ T cells for therapy.

There are many examples of patients who received CD4 T cells as a part of a TIL therapy ¹⁸, but to our knowledge, this is the first described case of a patient who experienced tumor regression after receiving a predominantly CD4⁺ TIL containing HLA-DR restricted tumor reactivity with no evidence of CD8⁺ tumor recognition. After receiving her CD4⁺ TIL, the patient experienced regression of bulky liver and spleen disease. The mechanism(s) of tumor destruction in vivo by CD4⁺ TIL are not known. Some CD4⁺ T cells are capable of exerting an anti-tumor effect without an appreciable contribution from CD8⁺ T cells by mobilizing inflammatory cells, cytokine secretion and direct tumor cytotoxicity ^{20, 39–43}. For the patient treated with CD4⁺ TIL in this study, a recurrent tumor mass appeared within a few months of CD4⁺ TIL infusion, and this recurrent tumor failed to express HLA-DR. The infused CD4⁺ TIL was unable to recognize the recurrent tumor, even after induction with IFNγ. Supply of the original liver tumor deposits was extremely limited and prevented multiple experiments evaluating HLA class II expression. The antigen responsible for TIL recognition was still expressed but tumor recognition required transduction with the CIITA gene (that directly induces transcription from the HLA Class II locus), rescuing the surface expression of HLA-DR. This is the first example to our knowledge of a recurrent tumor with HLA class II “antigen loss” phenotype further supporting in vivo selection by CD4 T cells.

Unlike CD8⁺ lymphocytes, tumor-specific CD4⁺ Th1 cells may contribute to anti-tumor immunity through mechanisms that do not require direct tumor recognition. For instance, Th1 CD4⁺ cells might contribute to tumor destruction by providing an “inflammatory” context for newly released tumor antigens through the activation of antigen presenting cells. These events might lead to the proliferation and survival of new anti-tumor lymphocytes, including CD8⁺ T cells, reactive with additional new tumor antigens. In a case study of one patient who received CD4⁺ NY-ESO-1 reactive T cells clones to treat metastatic melanoma, Hunder et al describe a lack of persistence of the transferred CD4⁺ cells, but induction of new tumor reactive CD8⁺ specificities subsequent to the CD4⁺ cell infusion ²². It is tempting to speculate that the potent tumor-reactive CD4⁺ Th1 TIL described in this report may also have resulted in a similar “epitope spreading” scenario. The TIL from the initially resected tumor had no evidence of CD8⁺ tumor-reactive or class I restricted lymphocytes. Resection of a recurrent lesion a few months after administration of the initial CD4⁺ TIL resulted in the generation of a new TIL that contained both CD4⁺ tumor-specific lymphocytes similar to the original CD4⁺ TIL as well as CD8⁺, HLA class I restricted lymphocytes with potent anti-tumor activity. Obviously, no causative role can be proven for the initial CD4⁺ TIL in the generation of the subsequent CD8⁺ TIL; and the tumor-reactive CD8⁺ lymphocytes in the second TIL may have been present in the recurrent lesion even in the absence of CD4⁺ T cell therapy. More research into the potential role of CD4⁺ Th1 cells in tumor inflammation and patient therapy is needed.

In summary, this study examined 44 TIL samples for the presence of HLA-DR restricted, CD4⁺ tumor reactive lymphocytes. Nine TIL (20%) were found to contain lymphocytes capable of secreting IFNγ after autologous tumor stimulation. One patient was treated with her CD4⁺ tumor-reactive TIL, and experienced a dramatic but short-lived objective tumor regression that was terminated by the outgrowth lacking HLA class II-restricted antigen-expression variant metastasis. The high frequency of tumor-reactive Th1 CD4⁺ cells in TIL and their demonstrated therapeutic potential suggests that these cells could be exploited clinically for patient benefit in ACT.